Sponsored by the Society of Tribologists and Lubrication Engineers
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
Volume III Monitoring, Materials, Synthetic Lubricants, and Applications Editor
E. Richard Booser, Ph.D. Consulting Engineer Scotia, New York
CRC Press, Inc. Boca Raton Ann Arbor London Tokyo
Copyright © 1994 CRC Press, LLC
Copyright © 1994 CRC Press, LLC
PREFACE—VOLUME III
Volume III extends this Handbook series to cover new developments and topics in tribology during the past decade. Together with the practical application practices covered in Volume I, in 1983, and theory and design in Volume II, in 1984, the three volumes provide a comprehensive reference tool for those involved with lubrication, friction, and wear. Among the diverse new developments included in Volume III are the revolutionary magnetic bearings finding use in demanding applications in compressors, high speed spindles, and aerospace equipment. Extensive coverage is given to tribology developments in office machines and in magnetic storage systems for computers. Authors have also developed new unified coverage for fatigue life of ball and roller bearings, for design and application practices with porous metal bearings, for self-contained lubrication involving wicks and oil rings, and for plastic bearings. Each of these classes of bearings is used by millions daily throughout industry. Synthetic lubricants are covered in a section of nine chapters by outstanding specialists in this rapidly developing field. Synthetics are coming to the forefront in such widely diverse areas as automobiles, aerospace, compressors using the new ozone-layer-friendly refrigerants, and a variety of extreme-temperature and environmentally sensitive applications. Water- and gas-lubricated bearings are given similar coverage. Preventive maintenance and monitoring were emphasized by the Handbook Committee for a variety of tribology elements: from ideas on monitoring sensors in the opening chapter; to specific monitoring techniques for automobiles, diesels, and rotating machines; and to procedures for tracking the remaining life of lubricants. These three volumes reflect the efforts of the Society of Tribologists and Lubrication Engineers plus hundreds of authors, reviewers, and associate editors. We trust that our series will continue to serve as a tool in achieving success for those involved in tribology. This should be reflected by improved performance of materials and lubricants in a broadening range of applications, along with tribological innovations, as we search for new paths in our future. E. R. BOOSER EDITOR
Copyright © 1994 CRC Press, LLC
THE EDITOR
Dr. E. Richard Booser has been active in tribology and lubrication for over 40 years. In academic training at The Pennsylvania State University, his research studies focused on refining procedures and performance testing for petroleum lubricants. He was then employed by the General Electric Co. for 39 years in development work on bearings and lubricants for steam and gas turbines, motors, generators, aerospace and nuclear plant equipment, and a variety of related electrical products. He currently works as a consulting engineer on bearings and lubrication. Assignments have covered lubrication of nuclear power plant equipment; bearing performance and problem analyses for steam turbines, gas turbines, generators, and accessory power plant equipment; friction and wear testing; locomotive, aerospace, and appliance bearings; electric motors; and failure analyses of ball and roller bearings. His 80 publications cover turbulence and parasitic power loss in high-speed oil-film bearings, fire-resistant lubricants, oil oxidation, grease life in ball bearings, design of circulating oil systems, lubrication of electric motors, and selection of bearing materials. He also co-authored the McGraw-Hill book Bearing Design and Application. While President of the Society of Tribologists and Lubrication Engineers (STLE) in 1956, he participated in initiating the annual joint Tribology Conferences with ASME. He has served as chairman of the STLE Lubrication Fundamentals Committee, and, in 1992, received the STLE National Award. He has also organized and taught bearing and lubrication courses for over 750 designers and engineers. Dr. Booser and the Handbook Committee have drawn on their broad associations and, especially, on the resources of STLE to organize this book. Together with the earlier Volumes I and II, this Handbook series represents a compilation by 120 authors of practices and developments in the still emerging field of tribology: the science of friction, wear, and lubrication.
Copyright © 1994 CRC Press, LLC
Donald G. Flom, Ph.D. Chairman Flom Consulting Scotia, New York
ADVISORY BOARD
Norman S. Eiss, Jr., Ph.D. Professor Department of Mechanical Engineering Virginia Polytechnic Institute and State University Blacksburg, Virginia Traugott E. Fischer, Ph.D. Professor Department of Materials Science and Engineering Stevens Institute of Technology Hoboken, New Jersey
Copyright © 1994 CRC Press, LLC
Robert M. Gresham, Ph.D. Vice President-Technology E/M Corporation West Lafayette, Indiana Michael Khonsari, Ph.D. Professor Department of Mechanical Engineering University of Pittsburgh Pittsburgh, Pennsylvania
George H. Kitchen President International Lubrication and Fuel Consultants Rio Rancho, New Mexico Charles A.. Moyer (Retired) The Timken Company Canton, Ohio
EDITORIAL REVIEW BOARD
K. Bajaj Stewart Warner Corporation of Canada, Ltd. Belleville, Ontario, Canada
A. Jackson Mobil Research and Development Corporation Paulsboro New Jersey
G. C. Barber Detroit Diesel Corporation Detroit, Michigan
K. R. Januszkiewicz Alcan International Ltd. Kingston, Ontario, Canada
F. J. Blatz Auto Research Laboratories, Inc. Chicago, Illinois
William D. Marscher Concepts ETI Parsippany, New Jersey
K. J. Brown Ontario Hydro Toronto, Ontario, Canada
S. H. Roby Lubrizol Corporation Wickcliffe, Ohio
P. W. Centers U.S. Air Force Wright-Patterson AFB, Ohio
F. Sadeghi Perdue University West Lafayette Indiana
W. J. Crecelius GE Aircraft Engines Cincinnati, Ohio
H. J. Sneck Rensselaer Polytechnic Institute Troy, New York
C. M. Ettles Rensselaer Polytechnic Institute Troy, New York
R. Timsit Alcan International, Ltd. Kingston, Ontario, Canada
Lois Gschwender U.S. Air Force Wright-Patterson AFB, Ohio
C. S. Yust Oak Ridge National Laboratory Oak Ridge Tennessee
Selda Gunsel Pennzoil Dormagen, Germany
Copyright © 1994 CRC Press, LLC
CONTRIBUTORS
Paul Allaire, Ph.D. School of Engineering and Applied University of Virginia Charlottesville, Virginia
Bharat Bhushan, Ph.D. Department of Mechanical Engineering Ohio State University Columbus, Ohio
Donald M. Bornarth Evanston, Illinois
Raymond G. Bayer Consultant Vestal, New York
E. D. Brown Schenectady, New York
William L. Brown Union Carbide Corporation Tarrytown, New York
Cris Cusano, Ph.D. Department of Mechanical Engineering University of Illinois Urbana, Illinois
Thomas W. Del Pesco, Ph.D. Specialty Chemicals E.I. DuPont de Nemours & Co., Inc. Deepwater, New Jersey William J. Derner Indianapolis, Indiana
Richard C. Elwell Schenectady, New York
Bruce J. Beimesch Henkel Corporation - Emery Group Cincinnati, Ohio
Copyright © 1994 CRC Press, LLC
Trangott E. Fischer, Ph. D Department of Materials Science and Engineering Stevens Institute of Technology Hoboken, New Jersey
Donald G. Flom, Ph. D Flom, consulting Scotia, New York
Robert M. Gresham, Ph.D. E/M corporation West Lafayette, Indiana
Lois J. Gschwender U. S. Air Force WL/MLBT Wright-Patterson Air Force Base, Ohio
Robert R. Humphris, D.Sc. Department of Mechanical, Aerospace, and Nuclear Engineering University of Virginia Charlottesville, Virginia Said Jahanmir, Ph.D. National Institute of Standards and Technology Gaithersburg, Maryland Warren E. Jamison, Ph.D. E/M Corporation Everett, Washington
Robert E. Kauffman University of Dayton Research Institute Dayton, Ohio
Michael Khonsari, Ph.D. Department of Mechanical Engineering The University of Pittsburgh Pittsburgh, Pennsylvania
Elmer E. Klaus, Ph.D. Professor Emeritus Department of Chemical Engineering Pennsylvania State University University Park, Pennsylvania
Carl R. Knospe, Ph.D. Department of Mechanical, Aerospace, and Nuclear Engineering University of Virginia Charlottesville, Virginia
Ranga Komanduri, Ph.D. Mechanical and Aerospace Engineering Oklahoma State University Stillwater, Oklahoma Dennis A. Lauer, P.E. Kluber Lubrication North America Londonderry, New Hampshire
David W. Lewis, Ph.D. Department of Mechanical, Aerospace, and Nuclear Engineering University of Virginia Charlottesville, Virginia
William D. Marscher Concepts ETC Parsippany, New Jersey
Charles A. Moyer North Canton, Ohio
William N. Needelman Pall Corporation Glen Cove, New York
R. L. Orndorff, Jr. B F Goodrich Engineered Polymer Products Wilmington, North Carolina
Joseph M. Perez, Ph.D. Office of Transportation Materials U.S. Department of Energy Washington, D.C. Douglas G. Placek FMC Corporation Princeton, New Jersey
Jack Poley Lubricon Beech Grove Indiana
Yongbing Liu Mechanical and Production Engineering Department National University of Singapore Subrat Ray Singapore Department of Metallurgical Engineering
Eric H. Maslen, Ph.D. Department of Mechanical, Aerospace, and Nuclear Engineering University of Virginia Charlottesville, Virginia.
Lee A. Matsch, Ph.D. Garrett Engine Division Phoenix, Arizona
Michael P. Marino FMC Corporation Philadelphia, Pennsylvania
Copyright © 1994 CRC Press, LLC
University of Roorkee Roorkee, India
Pradeep K. Rohatgi, Ph.D. Materials Department University of Wisconsin Milwaukee, Wisconsin
Shirley E. Schwartz, Ph.D. Fuels and Lubricants Department General Motors North American Operations Research and Development Center Warren, Michigan
Wilbur Shapiro. Mechanical Technology Incorporated Latham, New York Glenn D. Short CPI Engineering Services Midland, Michigan
Ronald L. Shubkin Ethyl Corporation Baton Rouge, Louisiana
Copyright © 1994 CRC Press, LLC
Don J. Smolenski, Ph.D. Fuels and Lubricants Department General Motors North American Operations Research and Development Center Warren, Michigan
Carl E. Snyder, Jr. U.S. Air Force WL/MLBT Wright Patterson Air Force Base, Ohio Esko Venalainen Esko Industries Ltd. North Vancouver British Columbia, Canada
TABLE OF CONTENTS MONITORING AND MAINTENANCE Condition Monitoring Sensors and Systems Automotive Engine-Oil Condition Monitoring Diesel Engine Lube Analysis Rotating Machinery Vibration Testing, Condition Monitoring, and Predictive Maintenance Filtration Rapid Determination of Remaining Useful Lubricant Life
MATERIALS Friction and Wear of Ceramics Plastics and Plastic Matrix Composites Metal Matrix - Solid Lubricant Composites Bonded Solid Film Lubricants
SYNTHETIC LUBRICANTS Aerospace Applications of Synthetic Fluids and Lubricants Industrial Applications Automotive Applications Polyalphaolefins Dibasic Acid and Polyol Esters Polyalkylene Glycols Phosphate Esters Perfluoroalkylpolyethers Silicones
APPLICATIONS Tribology of Magnetic Storage Systems Computers and Office Machines Refrigeration and Air Conditioning Oil-Mist Lubrication Tribology in High Speed Machining
BEARINGS AND SEALS Hydrodynamic and Hydrostatic Seals Rolling Bearing Fatigue Life Porous Metal Bearings Self-Contained Bearing Lubrication: Rings, Disks, and Wicks Water and Process Fluid Bearings Gas Bearings Magnetic Bearings INDEX
Copyright © 1994 CRC Press, LLC
603
CRC HANDBOOK OF LUBRICATION AND TRIBOLOGY E. Richard Booser, Editor
Volume I Application and Maintenance
Applications Industrial Lubrication Practices Maintenance Appendixes Volume II Theory and Design
Friction, Wear, and Lubrication Theory Lubricants and Their Application Design Principles Volume III
Monitoring, Materials, Synthetic Lubricants, and Applications
Copyright © 1994 CRC Press, LLC
PERMISSIONS
Chapter 1, Figure 9: From Conlley, R. E., ASLE Trans., 20, 244, 1977.
Chapter 7, Figure 1: From Dong, X. and Jahanmir, S., Tribological characteristics of alumina at elevated temperatures, J. Am. Ceram. Soc, 74, 1036, 1991. Reprinted by permission of the American Ceramic Society.
Chapter 15, Figure 5: From Lahajani, J., Lockwood, F. E., and Klaus, E. E., ASLE Trans., 25, 25, 1982.
Chapter 19, Figures 3 and 6: From Demby, D. et al., in Synthetic Lubricants and High Performance Fluids, 1992, p. 183, by courtesy of Marcel Dekker, Inc. Chapter 20 Figures 28 and 30: From Bhushan, B., Tribology and Mechanics of Magnetic Storage Devices, Springer-Verlag, 1990, p. 23. Figure 16: From Bhushan, B., Bradshaw, R. L., and Sharma, B. S., ASLE Trans., 27, 89, 1984.
Chapter 24, Figures 9 and 10: From Kottenstette, J. B., in High-Speed Machining, American Society of Mechanical Engineers, pp. 91 and 372, 1984.
Chapter 26 Figure 4: From Tallian, T., ASLE Trans., 5, 183, 1962. Figure 10: From Cantley, R. E., STLE Trans., 20, 244, 1977.
Chapter 28, Figures 9 and 10: From Kaufman, H. N., Szeri, A. Z., and Raimondi, A. A., Trans. ASLE, 21(4), 315, 1978. Figures 11, 12, and 13: Courtesy of Kingsbury, Inc., Philadelphia, PA.
Chapter 30, Figure 12: From Raimondi, A., ASLE Trans., 4, 131, 1961.
Copyright © 1994 CRC Press, LLC
CONDITION MONITORING SENSORS AND SYSTEMS Esko Venalainen
Due to increased speeds of machinery and substantial costs of unexpected shutdowns, general interest in parameters influencing bearing performance has greatly increased. This issue was extensively discussed at the Tappi Engineering conference in Boston, 1984. Further development of data communication and computer hardware since 1984 has offered costeffective means to continuously monitor certain conditions of the machinery and its oil to avoid unexpected bearing failures and to obtain more efficient and organized maintenance work through scheduled shutdowns. Sensor technology to measure certain parameters has been available for quite some time, but these applications have lacked the opportunity to communicate cost effectively into centralized control rooms. Without present data communication, every sensor had to be hardwired from its location all the way to the control room or annunciator panel.
OIL FLOW SENSORS
While common in the industry, conventional “sight glasses” cannot be considered as oil flow sensors for monitoring purposes because they cannot provide any electronic output. Quite a large variety of oil flowmetering or sensing devices have, however, been offered over recent years. Measurement of oil flow is rather specific and requires special attention to the requirements of each application. This means that compromises may have to be made. An ideal oil flowmetering device or sensor should have the following features:
• • • • • • •
No leaks. Easy to maintain, i.e., take apart. Relatively accurate and repeatable. Easy to read, even with dark, dirty oil. Built-in valve for flow adjustment. Influence of oil viscosity should be minimal. Reliable in continuous operation to provide flow indication with no “sticking” behavior.
Several types of oil flowmeter for continuous measurement of oil flows to the bearing are available. These can be grouped basically as the two following types: (1) variable area rotameter, and (2) meters with a continuous electronic output signal (on/off or continuous). Variable Area Flow Monitors Different types of variable area meters, known as rotameters or simply as “sight glasses” are shown in Figure 1. Traditionally, they are made of different materials, including steel, aluminum, brass, glass, and plastic. The indicating device is a float inside the flow tube which can be glass, plastic, or metal. Connecting parts are always plastic or metal. The whole meter can also be made of metal, in which case it is called an armored flowmeter, which is very rarely used because of its high price. The major benefit is fire safety and freedom from leaks. Use of plastic components has been increasing. One reason is their capability to endure higher temperatures; some are now available for routine use up to 150°C. Another reason is their elasticity: with proper construction plastic components can take some stress and are relatively easy to make leak-proof. Copyright © 1994 CRC Press, LLC
3
CRC Handbook of Lubricants and Tribology
4
FIGURE 1. Variable area oil flow sensors.
Sensors with Continuous Electronic Output The following types of oil flow sensors generate an electronic signal, analog or digital, for continuous monitoring of oil flows.
Positive Displacement Meter. The most common of this type is the so-called oval gear meter, developed during World War I to accurately detect oil quantity in submarine periscope control. As shown in Figure 2, the measuring elements are oval shaped and the flow is rotating them. The piston type positive displacement meter has also occasionally applied in lubrication. A major drawback is the back and forth moving piston, which easily becomes a high maintenance item. Turbine Meter. Turbine meters are also applied for continuous oil flow measurement. They are not as accurate as positive displacement meters. While they are viscosity sensitive, at higher viscosities the sensitivity error is not very high, and this type of meter has been found to be generally acceptable. Gear Meter. While gear meters are also available, they have not often been used for lubrication purposes. They are very accurate and can be made to take high pressures. Pressure Differential Oil Flow Meter. In these meters, as illustrated in Figure 2, the sharpedged orifice plate generates pressure differential which is proportional to the flow. It is nonlinear, but viscosity immune. If supply pressure is constant, then downstream pressure can be measured as an indication of oil flow.
OIL FLOW SENSORS PROVIDING ON/OFF AND ALARM SIGNALS
These are typically variable area flowmeters with a signal from a float used to indicate the lack of adequate oil flow. Of on-off sensors in this category, Figure 3 illustrates the freefloating type and shows three types of detectors used, and Figure 4 illustrates the technical characteristics of inductive proximity sensors. The main difference between free-floating and spring-loaded units (see Figure 5) is the mounting. The free-floating one has to be mounted vertically, but the spring-loaded flowmeter can be mounted in other positions as well. Another basic difference between these two types of sensors is their viscosity sensitivity. Free-floating oil flowmeters are more viscosity sensitive than spring-loaded ones. The float on a free-floating unit is often made with sharp edges to reduce the viscosity sensitivity. If we Copyright © 1994 CRC Press, LLC
Volume III
5
FIGURE 2. Oval and pressure differential types of flowmetering devices.
FIGURE 3. Different types of detectors to provide low flow alarm with free-floating devices.
try to maximize the flow, however, then a sharp edge may have to be removed from the float and the meter is then more viscosity sensitive. A free-floating device also cannot become stuck, while a spring-loaded float needs support from tight side clearances which introduces the possibility of getting stuck. This sticking Copyright © 1994 CRC Press, LLC
6
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
FIGURE 4. Electrical function and characteristics of an inductive proximity sensor.
Volume III
FIGURE 5. Spring-loaded variable area flowmeter at top, swinging vane type at bottom.
7
behavior may cause a false indication of adequate oil flow, even when the bearing may be starving. An alarm arrangement with a spring-loaded float is often controlled by micro-switch. Tight clearance construction of the spring-loaded switch calls for clean oil to make this meter reliable. Figure 3 illustrates the following three different means to detect the position of the float:
• • •
Magnetic reed relay Infrared Inductive proximity sensors
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
8
FIGURE 6. Standard oval gear meters cover flow ranges up to 100 l/min (25 gpm).
The magnetic reed relay is low in cost, but its reliability is less than that of inductive proximity sensors. Inductive proximity sensors do not attract steel particles like magnetic ones. Infrared is rarely used because it is sensitive to oil color and contamination. The inductive proximity sensor shown in Figure 4 has proven to be the most reliable and accurate device to detect the position of the indicating float in a variable area flowmeter. At the tip of the sensor, there is an electromagnetic field. Presence of the metallic float will “overload” this electromagnetic field and in that way provide on/off indication. Another basic proximity sensor is the capacitive type. Its functions are reversed, but the basic operation principle is the same as that of the inductive sensor. The basic function in all these types of flow alarms is to detect the position of the float or pointer in the indicating mechanism. Whenever the float enters the proximity of a sensor, then an alarm will be provided, as illustrated in Figure 3. Also, an inductive proximity sensor is often applied as a low flow alarm in the bottom of the meter. Basic features of inductive proximity sensors are given in Figure 4.
OIL FLOW SENSORS PROVIDING CONTINUOUS DIGITAL OR ANALOG SIGNALS
These types of sensors will provide remote indication. Alarming with these sensors is normally arranged in receiving electronics, a basic difference from the sensors mentioned above. For continuous oil flow measurement, four basic types are available: • • • •
Oval gear meters or piston meters Gear meters Turbine meters Orifice plate/pressure differential meter
Oval Gear Meters A typical example of an oval gear meter is illustrated in Figure 6. This technology was developed several decades ago for submarine periscope control. It was applied for the first time Copyright © 1994 CRC Press, LLC
Volume III
9
in lubrication applications almost 10 years ago in Finland, and today there are tens of thousands of these meters employed successfully in oil flow measurement. The meter generates a pulse increment proportional to flow and in practical terms provides reliable viscosity-immune measurements. Lubrication oval gear meters typically use inductive proximity sensors to detect rotation: in every rotation, the meter gives two pulses. Table 1 gives a tabulation of different sizes of oval gear meters with their pulse rate per U.S. gallon and per liters per minute. The oval gear meter normally has a capability of 1:10 flow ratio; for instance, a size 2 meter is capable of from 200 cm3/min up to 2 l/min at its normal speed limit of 900 rpm.
Gear Meter The gear meter gives a more accurate measurement than the oval gear meter, because gear meters normally give a pulse on each tooth of the gear. Gear meters are also available for high pressure hydraulic applications ranging up to 9000 psi. In principle, there is no difference between these two, except that today the oval gear meter dominates the market because of price. The electronic pick up is a special crystal which will very reliably pick up magnetization in each tooth of the gear. This provides outstanding accuracy as compared to any other meter.
Turbine Meter The turbine meter is not a positive displacement meter, although it does provide continuous indication of the oil flow. The basic limit for a turbine meter in lubrication applications is its sensitivity to viscosity variations above 30 cSt. There have been developments and trials, but this meter has not achieved popularity in lubrication monitoring. While piston-type metering devices have found some use, they are declining in popularity because they are difficult to maintain.
Orifice-Type Oil Flowmeter Another type of continuous flow measurement is the well-known and reliable orifice plate. As indicated in Figure 2, the sharp-edged orifice plate will provide low pressure differential proportional to the flow. Because of the sharp-edged orifice plate, it is considered practically viscosity immune. Since constant supply pressure can be arranged, it is possible in oil flow applications to use a pressure gauge on the downstream side of the orifice plate as a secondary measurement. Obviously, there is a potential error if supply pressure varies. Often this approach is used only for alarming by mounting the pressure gauge with an alarm contact on the downstream side. Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
10
FIGURE 7. Thermowell and sensors used for monitoring bearing temperatures.
OIL AND BEARING TEMPERATURE SENSORS As mentioned before, excessive bearing temperatures have been acknowledged as a main cause or indication of bearing damage. At certain temperatures, thermal stresses occur and bearings will be damaged and their life shortened. Also, at high temperatures, the viscosity and lubricating effectiveness of the oil are significantly reduced. Extra heat can come from several sources. If steam is present, then obviously insulation of the bearing and the oil is very important. Often, inadequate desuperheating can introduce overheated steam adjacent to the bearing. Bearings can also be overheated because of excessive friction as from overloading. Obviously, lack of sufficient oil flow also will lead to increased friction and reduced heat transfer from the bearing, resulting in overheating. To sense temperatures, several technologies are available. Heat-sensitive paints which change colors at certain temperature have been used. For continuous monitoring, the following are possibilities: • • • •
Resistance temperature detectors (RTDs) Thermocouples Semiconductors Infrared temperature sensors
The electrical resistance of all metals will vary as the temperature varies. For this kind of temperature sensor, typically platinum and nickel are used. These sensors are then inserted into the thermowell shown in Figure 7. When applying an RTD for temperature measurement, a varying resistance signal has to be converted into an electronic signal acceptable for the monitoring systems. The required converter often increases the price of the system and therefore is not used. Copyright © 1994 CRC Press, LLC
Volume III
11
FIGURE 8. An example of infrared temperature detection system with a focused lens used to view heated surface from predetermined distance away from R.F. field.
Thermocouples In a thermocouple, the junction of two different metals will provide a low millivolt signal which is a function of its temperature. This signal again can be converted into a current signal and then applied to the monitoring systems. This sensor has the same requirement as RTD, i.e., it requires a converter to generate a milliampere signal for monitoring purposes.
Semiconductor Sensor The semiconductor sensor utilizes a tiny transistor having an accurately known temperature coefficient. This type of sensor provides a linear signal from 0 to 150°C (32 to 302°F). A good feature of semiconductor sensors is that receiving electronics, normally data acquisition, will accept this low milliamp signal as an input without a converter. In practice, this means that elimination of converters will make this type most cost effective.
Infrared Sensors Infrared temperature measurement as illustrated in Figure 8 detects radiation from the object for which temperature is being measured. The infrared thermometer can measure temperature from a distance without being in contact with the object. The signal provided can be transmitted electronically or through fiber optics. The main advantages of an IR system are fast detection of temperature variation, the fact that the sensor does not need to be physically in contact, and high reliability in a hostile environment. It is also reliable in eliminating socalled noise at the pick-up point. Good examples are bearing temperature measurements on induction motors or on heat-generating devices that can interfere with other types of sensing devices.
OIL CONTAMINATION ANALYZERS
Oil contamination has a proven effect on bearing life. Oil can be contaminated by particles or by other liquids, typically water. Particles can be of different types, and if the particle size is larger than the thickness of the oil film, its effect on bearing life through wear and extra stress can be significant. Harder metal particles are known to be more damaging than nonmetallic particles. Sensors have been developed for both particles and water contamination. Water Contamination Water contamination can be very damaging to ball and roller bearings, gears, and other machine components, although there is a fair amount of debate as to what is actually affected
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
12
FIGURE 9. Effect of water contamination on ball bearing life. (Data from R. E. Conlley, ASLE Trans., 20, 244– 248, 1977.)
by water in the oil. Figure 9 gives an indication of how water affects bearing life in certain conditions. Very low concentrations, down to 0.01% (100 ppm) can reduce bearing life to one half. On the other hand, water contamination has very little effect in reasonable amounts on tin babbitt bearings in power plant equipment (turbines, etc.). While many traditional measurements do not have sufficient sensitivity, water absorption of infrared light through the oil sample will provide measuring sensitivity below 100 ppm. These units have an oil sample flow through a measuring cell between two glass surfaces, after which the sample flow is directed back to the system. Two different wavelengths of infrared are commonly applied, one for measuring and the other as a reference. While the measuring wavelength is being absorbed by water molecules, the reference one is unaffected by water. These two signals are then amplified and processed by the monitoring system to indicate water contamination level. These units normally offer the following features:
• • • •
Measuring range from 100 to 5000 ppm Accuracy ± 50 ppm Outputs: analog 4 to 20 or digital RS422 Alarm output: dry relay contact normally open or normally closed
A water evaporation technique is also used. Its drawback is insensitivity for low concentrations. It is reliable for high concentrations only.
Metallic Particle Sensors Quite recently, sensors to detect metallic particle distribution in lubrication oil have become commercially available. They are based on inductive behavior to detect the quantity of contamination. They are relatively low cost and easily applicable for lubricating oil applications.
VIBRATION SENSORS
Vibration measurement and vibration analyses can become very complex and are discussed in a later chapter of this handbook. While very easy to include them in a continuous monitoring system, detailed continuous analysis on-line involves a costly system and is often difficult to justify. Continuous measurement from several hundred bearings up to a frequency of 30 to 60 kHz requires very sophisticated electronics. Copyright © 1994 CRC Press, LLC
Volume III
13
FIGURE 10. Piezoelectric accelerometer with preamplifier.
FIGURE 11. Typical vibration sensors.
Simple vibration amplitude measurement, however, is possible at a substantially lower cost on an on-line basis. When an alarm of certain amplitude is given, then the rotor-bearing system should be checked with more sophisticated equipment. Typically, for vibration measurement, the following sensor technology is available: • • •
Motion sensor Velocity sensor Acceleration sensor
The first two have commonly been used over the years to monitor large rotor systems in oilfilm bearings. The third one has come to the forefront in today’s technology. These sensors are often piezoelectric and built with a preamplifier to lower impedance. This gives more reliability for conditions where dirt and moisture may enter the system. Figure 10 gives the basics of a piezoelectronic sensor. These piezoelectric sensors are capable of very high frequencies, up to 30 to 60 kHz. Their basic element is a crystal which will be affected by vibration in its electrical phenomena, and will then provide accurate information on acceleration. Major benefits are their durable construction and ability to take mechanical punishment. When mounting vibration sensors such as are shown in Figure 11 in the bearing housing, one should consult a specialist because there are several critical details. They normally require Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
14
mounting so that the sensor tip will not be in touch with the bearing house in a certain manner. Also, positioning to sense maximum amplitude within the bearing is rather important.
CONTINUOUS MONITORING SYSTEMS
With sensors monitoring variables influencing bearing performance, the signals can easily be communicated to different types of computing and monitoring systems. Receiving electronics can be fairly sophisticated if so desired, but also simple to use. Two different types of components have to be considered: data acquisition from sensors communicated to host computer, and host computer systems.
Data Acquisition from Sensors Communicated to Host Computer Data communication from sensors to the host computer is possible with commercially available programmable logic controllers, known as PLCs. Since PLCs were not designed initially for these types of applications, the equipment selected for this can be an overkill and may be costly. Some recent units, however, are designed for data acquisition and some specifically for bearing performance monitoring. Figure 12 gives two typical examples of how to apply this data acquisition for monitoring purposes. Several interfaces are available. The most common are RS232, RS422, and RS485. It is recommended that when considering the purchase of such a system, one should use an expert to consider the following basic features of these systems.
Communicating Electronics Programmable logic controller (PLC)—A programmable logic controller has all the features needed for communication of signals from an inductive proximity sensor or analog signal from a temperature sensor to the host computer. Several proximity sensors are also available that are specifically designed for PLC systems. Among available PLCs, several input/output circuitries are specifically designed for inductive proximity sensors. The only significant drawback with commercial PLCs is that low-cost units have not been designed for this kind of data communication and are not capable of continuous electronic monitoring. While higher price units are more than capable of this type of application, the cost may not be justifiable. ASDA Systems (Application Specific Data Acquisition)—This unit is more cost effective and is specifically designed for continuous monitoring of electronic signals. Like PLCs, this unit is furnished with RS232 and/or RS422 ports, is adaptable to most host computers, is less expensive than most PLCs, and is specifically designed for data acquisition. ASDA systems are furnished with several significant features:
• •
• •
LED light panel indicates alarm condition individually for each bearing being monitored. Spring return switch allows display of oval gear rotation. This makes it easy to check low flow in each bearing. Spring return switch allows temperatures to be examined. Diagnostic programs indicate malfunction of the system.
ALARM SYSTEMS
Multilevel Alarm The first alarm level in a multilevel system will indicate conditions which are undesirable, but do not necessarily require machine shutdown. The second level of the alarm indicates a hazardous condition which may require machinery shutdown. Copyright © 1994 CRC Press, LLC
Volume III
15
FIGURE 12. Bearing performance monitoring system.
Obviously all events are recorded by the alarm printer. The report indicates the time of the occurrence of the condition and also the time when the situation was corrected, even if it corrected itself.
Fluctuation Alarm This gives information when there is continuing variation in oil flow or other operating parameter which does not necessarily go below or above alarm level. This information can also be stored on the disk and the behavior may be reported or stored for failure examination purposes. If the monitoring is expanded to its full extent, it is possible to calculate heat transfer from the bearing. Oil temperature increase through a bearing in many applications should not exceed 30 to 40°F. In a very simple manner, it is possible to calculate and alarm all these types of situations. Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
16
SOFTWARE FEATURES
In case of an alarm, the software package allows operating maintenance personnel to select the following from a menu:
• •
CRT display of the flows and alarms of any selected section of a machine or the mill. History file check as to what specifically has happened in a certain selectable time interval, such as 3 days, 6 days, 90 days, etc.
EXPERIENCE AND CONCLUSIONS
Experiences to date have proven that continuous predictive monitoring systems are easy, user-friendly, and a justifiable solution to collect information and inform maintenance personnel of troublesome conditions occurring in multi-bearing machinery such as paper machines. Not only do they provide oil flow alarms, but they also supply valuable information on the effects of machine speed and lack of oil flow. Desuperheating discrepancies in bearing temperatures has also been gathered in practice. These kinds of systems are cost-effective tools that offer the following benefits:
• • • • •
• • •
Continuously monitors and reports operating conditions Updates, reports, and records failures and corrections of the events Compatible with higher level computers Individual substation continues to function independently if the host computer fails Operating or maintenance personnel can easily check on the condition of the machinery and program engineering units and alarm levels through the keyboard Program package is easy to use and customize to a particular mill Cuts wiring costs as a result of data highway technology System has built-in diagnostic programs
Copyright © 1994 CRC Press, LLC
AUTOMOTIVE ENGINE-OIL CONDITION MONITORING Donald J. Smolenski and Shirley E. Schwartz INTRODUCTION
Over many years, investigators noted that certain operating cycles caused specific kinds of lubricant degradation1,2 and that certain kinds of lubricant degradation caused specific kinds of damage to the engine or loss of performance.3 Investigators also learned that analysis of used engine oil could yield useful information relative to:
• • •
Condition of the oil Engine condition and early detection of problems in the absence of an inspection Probable causes of problems or failures observed when actual engine inspections are performed
In this chapter, examples of lubricant degradation (primarily in gasoline-fueled engines) are presented, along with a description of oil analysis methods to document and interpret them.3–9 Techniques to monitor or model oil degradation during vehicle operation are also presented.
TAKING AN OIL SAMPLE AND CHECKING FOR ABNORMAL OIL
To understand changes occurring in used engine oil, a sample of the same brand and, if possible, the same batch of new oil is needed for comparison. Routinely setting aside a fresh oil sample at the start of an engine or road test is a good policy. Knowing that an oil was misblended (as shown in Figure 1 from the last few oil changes of a high-mileage durability test) saves the investigator from blaming surprising test results on the engine. To illustrate what may happen if abnormal oil is used in a test, three oil-change intervals from the end of a 200,000-km test are shown in Figure 1. The first oil-change interval shown (12,704 km) exhibited typical oil degradation for warm-weather highway service: •
TAN (acid number; terms defined in Table 1) low initially and rising to the dashed line which represents a condemning limit (described later) TBN (base number) high initially and falling to a condemning limit DSC (oxidation induction time by differential scanning calorimetry) high initially and falling nearly to a condemning limit
The second change interval, though short (1,198 km), also exhibited normal oil characteristics. The third oil-change interval (11,367 km) was abnormal; TAN of the fresh oil was above the warning limit and remained above it, TBN started out higher than for normal oil and remained high, and DSC started out near the warning limit and remained near it throughout the oil-change interval. Because fresh oil samples had been retained, the authors were able to determine that there was an oil problem rather than an engine malfunction. A clean, dry bottle is needed for an oil sample. The oil should be warm at the time the oil is collected (a 4-km trip provides sufficient warming). A good time to obtain a sample is during an oil change, since the oil has to be removed at this time anyhow. Collection may be either via a vacuum tube inserted into the dipstick opening (sample a centimeter or two above the bottom of the oil pan) or via the drain plug (discard the first and last portion of oil, sample from 0-8493-3903-0/94/$0.00 + $0.50 © 1994 by CRC Press, Inc.
Copyright © 1994 CRC Press, LLC
17
CRC Handbook of Lubrication and Tribology
18
FIGURE 1. Normal oil aging (circles) compared to abnormal aging (squares).
the middle portion). Sampling from the mid portion of the oil is preferable since the top and bottom portions are more likely to be contaminated, and the mid portion is more likely to represent what is flowing through the lubrication system.10
OIL ANALYSIS TECHNIQUES AND INTERPRETATIONS
A number of factors can impair engine oil performance. Oil thickening, loss of wear protection, and deposit control are of concern primarily in high-temperature, high-load service.3 Oil thinning, loss of corrosion protection, and low-temperature sludge formation are of concern primarily in short-trip, winter service.3 Other types of service may also cause particular types of degradation. Various analytical techniques to document these effects are listed and briefly explained in the following paragraphs. Details of these methods can be found in the references cited in each section (primarily ASTM methods described in Reference 11). SAE J357 in the SAE Handbook (Reference 12) is also a useful guide. Table 1 describes the analytical techniques and gives the abbreviations used in this Chapter, and Table 2 lists recommended analyses to quantify the nature and extent of oil problems. Copyright © 1994 CRC Press, LLC
Volume III
19
Viscosity (ASTM D 445, D 4683, D 4684, D 4741, and D 5293) The viscosity of monograde oils such as SAE 20, 30, 40, or 50 is defined at 100°C by the SAE Engine Oil Viscosity Classification, J300.13 The viscosity requirements of multigrade oils such as SAE 5W-30, 10W-30, or 15W-40 are defined by this same classification at both low-temperatures (designated by the letter “W”) and high-temperatures. One technique for measuring high temperature viscosity is to determine the time required for a specified volume of liquid to flow through a calibrated capillary viscometer (ASTM D 445). Viscosity obtained in this manner is called the “kinematic viscosity”, and is expressed in units of centistokes (cSt or mm2/s). High-temperature, high-shear viscosity, measured at shear conditions typical of operating engines, is determined using ASTM D 4683 or ASTM D 4741. Low-temperature (-5 to -30°C) viscosities are primarily determined by two separate tests, each defining a specific viscometric property. The viscosity that reflects the viscous drag of an oil at low temperature is determined with a cold-cranking simulator (CCS) using ASTM D 5293. As defined in SAE J300, the viscosities of the different grade oils must be lower than the stated maximum values at the temperatures indicated. Another requirement of the “W” Copyright © 1994 CRC Press, LLC
20
CRC Handbook of Lubrication and Tribology
low-temperature grade is the pumping viscosity, as measured by ASTM D 4684. This test method determines an apparent viscosity of the oil at low temperature and thus indicates whether an engine oil will flow to the oil pump inlet and provide adequate oil pressure during the initial stages of engine operation. Changes in engine oil viscosity indicate some form of oil degradation during vehicle operation. Increases in high-temperature viscosity can be due to oil oxidation, soot contamination, or volatilization of light base stocks in the oil.3 Decreases in viscosity can be due to fuel or low-viscosity soluble contaminants in the oil, or to shearing of the viscosity index improver.3 The consequences of an excessive viscosity increase include difficulty in starting a vehicle at low temperatures3 and the possibility of inadequate lubrication if the oil is so viscous that it can’t reach all the critical components that it must lubricate.3 The consequences of excessively low viscosity include insufficient oil film thickness and wear of parts such as bearings and crankshaft journals, which depend on an adequate oil film for lubrication.3 The requirements for engine oil viscosities are continually being upgraded. Thus one must consult a current version of SAE J300 for the latest test methods.
Acid Number and Base Number (ASTM D 664 and D 2896) The acid number and the base number (TAN and TBN, formerly designated total acid number and total base number) are determined by titrating an engine oil either to a fixed end point or to an inflection point using a standard base or a standard acid solution. TAN is a measure of the concentration of acidic species in the oil, which may include weakly acidic components of the fresh oil, acids formed during oil oxidation, and weak to moderately strong acids generated during the combustion process. TBN provides a measure of the remaining amount of protective alkaline agents (“reserve alkalinity”) placed in the oil by the manufacturers to neutralize acids (particularly weaker ones). High values of TAN and low values of TBN, compared to values for the fresh oil, indicate that the oil has lost some of its ability to neutralize acids and that corrosion of engine components is more likely to occur.12
Pentane and Toluene Insolubles (ASTM D 893) The amount of insoluble material in the oil is determined by mixing a sample with the appropriate solvent (pentane or toluene) and weighing the dried solid removed during centrifuging. Pentane insolubles indicate the total amount of insolubles in the oil; toluene insolubles indicate the inorganic portion. If the oil sample contains primarily oxidized engine oil, fuel, or other hydrocarbon components, toluene insolubles typically will be negligible. However, if large amounts of foreign material are ingested through the intake system or enter the crankcase via some other means, toluene insolubles will be appreciable. The value
Copyright © 1994 CRC Press, LLC
Volume III
21
determined by subtracting toluene insolubles from pentane insolubles is sometimes designated as “resins” and represents the organic portion. There are two versions of the test, “coagulated” (procedure A) and “uncoagulated” (procedure B) insolubles. Under long-trip service conditions, a high content of insolubles generally indicates that the dispersant in the oil is no longer performing its intended function. As a result, small oil passages in the engine can become plugged, resulting in oil starvation of critical engine parts. Under extreme short-trip winter driving cycles in which the water content of the oil is high, the amount of insoluble material determined by ASTM D 893 can be very high (greater than 5% per Reference 10) and can even exceed the percentage of water in the oil. At the same time that significant amounts of low-temperature pentane insolubles are found in the oil, visual observation of the oil indicates the presence of a considerable amount of white sludge.10 Once longer trips are taken, the amount of insoluble material can diminish rapidly. That is, the formation of low-temperature pentane insolubles appears to be highly reversible.10 Fuel in Oil (ASTM D 322 and D 3525) Fuel contamination of the engine oil (“fuel dilution”) can be determined by either of two ASTM methods. In ASTM method D 322, water is added to the oil sample and the mixture is distilled (any fuel in the sample is vaporized with the water). In ASTM method D 3525, the fuel concentration in the oil is determined by means of gas chromatography. Since ASTM method D 322 may erroneously indicate fuel contamination in fresh oil samples, method D 3525 is generally preferred. Under certain conditions, excessive unburned fuel in the oil from cold weather operation can reduce oil viscosity so that it no longer provides adequate oil-film thickness in critical areas within the engine. However, much of the fuel may be eliminated when the engine oil reaches stabilized operating temperatures. For example, near 100°C (212°F) approximately 50% of the gasoline may evaporate, and near 150°C (300°F) 90% of the fuel may evaporate.3 Furthermore, the heavier ends of the fuel, which remain after lighter ends are boiled off, have less tendency to reduce the viscosity of the engine oil. Fuel in the oil contributes significantly to the loss of oxidative stability of the oil.10 Partially oxidized fuel components, arising during short-trip service, can also be condensed in the engine oil3 and can cause an increase in TAN and a decrease in TBN.10 This effect can sometimes be reversed at higher oil temperatures.10
Water in Oil (ASTM D 1744 and D 4928) Water in engine oil can be determined by a Karl Fischer titration, a coulometric titration, or other methods. Excessive water contamination in engine oil can cause increased wear in an engine.14 Water can also cause excessive corrosion and can affect the solubility of the oil’s additive package, sometimes causing “additive drop out”, that is, precipitation of additives from the oil.10
Ethylene Glycol in Oil (ASTM D 2982 and D 4291) If there is an engine condition such mat coolant (antifreeze: ethylene glycol plus water plus additives) leaks into the oil, testing for glycol becomes important. The presence of glycol can be detected by several methods. For example, ASTM D 2982 involves the use of reagents to determine semiquantitatively the glycol concentration by means of an indicator color change. Test kits, based on a color change to indicate glycol, are commercially available. ASTM D 4291 determines glycol concentration by means of a gas chromatographic technique. Glycol concentrations in engine oil as low as 0.5% can cause significant sludging15,16 and may cause a more rapid than normal aging of the oil.17 Some oil additives and some oil Copyright © 1994 CRC Press, LLC
22
CRC Handbook of Lubrication and Tribology
contaminants such as aldehydes produce a positive test for glycol when, in fact, no glycol is present.
Infrared Spectrum Infrared spectroscopy involves passing a beam of infrared light through a sample and measuring the fraction of radiation absorbed at various wavelengths, to provide a spectrum which yields qualitative information on oil condition.18–21 When a sample of fresh oil is available for comparison, differential infrared spectroscopy can be used to obtain semiquantitative information on species concentrations.21.22 Various chemical species have “fingerprint” infrared absorbances at different wavelengths, as shown in Table 3. Thus, an infrared spectrum can indicate semiquantitatively the extent of oil contamination (fuel, water, glycol, soot), oxidation, nitration, zinc dithiophosphate (ZDP) depletion, or, using differential infrared spectroscopy, the concentration of various oil components.
Differential Scanning Calorimetry Differential scanning calorimetry (DSC) is used to measure the oxidation induction time of an engine oil. This method involves heating a drop of oil in a high-pressure oxygen atmosphere and measuring the time required (under isothermal conditions) for the onset of oxidation (termed oxidation induction time).2.22–24 Generally, the shorter an oxidation induction time, the less thermally stable an oil. For a typical DSC test in which a fresh oil might have an induction time near 100 min when tested at 165°C under an oxygen atmosphere of 3.8 MPa (550 psi), a very short induction time (less than 3 min, for example) for used oil of the same brand indicates that little or no active antioxidant remains. Values of 5 min or below, under the conditions previously mentioned, have been used to indicate that the oil is at or beyond the recommended change point.2 DSC values are reduced by the presence of fuel or contaminants in the oil2 and are strongly dependent on the temperature at which DSC is conducted.
Metal Analyses Various analytical techniques can be used to determine the concentration of metals in engine oil, for example: wet chemical methods, atomic absorption, X-ray fluorescence, or inductively-coupled plasma spectroscopy. The various metals of interest and their significance are included in Table 4. In determining the significance of the metals content of used engine oil, it is necessary to know the metals content of the fresh oil as well as the type of materials used in engine components, the fuel properties, type of service, and oil-change interval. It may also be helpful to know the type of metals which are present in the coolant if there is concern about a possible coolant leak.
Copyright © 1994 CRC Press, LLC
Volume III 23
Copyright © 1994 CRC Press, LLC
24
CRC Handbook of Lubrication and Tribology
Copyright © 1994 CRC Press, LLC
Volume III 25
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
26
Other Analyses Several other analyses are often used to characterize engine oils, primarily in quality control checks or special applications. These analyses include specific gravity, density, Brookfield viscosity, pour point, flash point, distillation characteristics, sulfated ash, chlorine, sulfur, nitrogen, nickel, silver, other metals, and ferrographic analysis. References 11, 12, 25, and 26 provide details.
INTERPRETATION OF ANALYTICAL RESULTS
Once a sample of oil has been obtained, delivered to an oil-analysis laboratory, and oil analyses selected, the remaining hurdle is interpretation of oil analysis results. Table 4 summarizes “warning limits” at which engine oil should be changed. These warning limits do not indicate that, if the limits are exceeded, the engine faces immediate and catastrophic failure. Rather, the limits indicate that someone at some time has found a potential for reduced engine performance or reduced durability under those conditions. In some cases, published limits incorporate a safety factor. To indicate the range of various analytical results, Table 4 includes examples of some extreme values the authors have observed. In Table 4 the column entitled “Significance of out-of-limits results” explains the underlying concern when a parameter is out of limits. The column “Related analyses” lists alternative analyses which might help explain, refute, or strengthen inferences based on results from the original analysis. For example, determination of fuel in the oil may confirm fuel contamination as the cause of reduced viscosity. “Comments” relates to further actions suggested, based on the results. In many cases the need for an oil change in the near future is implied. There are no universally accepted warning limits. Different investigators and different operating conditions may suggest different limits for a given oil-analysis measurement. Despite this variability, the limits still provide a valuable indication of the point at which oil should be changed, since once a particular oil property begins to degrade, the rate of further degradation may be extremely rapid.1–3 For example, even though a condemning limit for pentane insolubles (PIN) may be 1.5 in one reference and 5.0 in another, the time elapsed between the lower and higher values may represent only a vanishingly small fraction of the total operation time.
RELATING OIL ANALYSIS TO ENGINE DAMAGE OR IMPAIRED PERFORMANCE
When relating oil-analysis results to engine performance, used oil results are most meaningful when compared to new oil results. In the case of viscosity, the SAE viscosity grade implies a limited range of values for viscosity at high and low temperatures.12 If the low-temperature (-20°C) viscosity of an oil nominally listed as SAE 10W-30 is found to be appreciably greater than 3500 cP, poor flow, possible oil starvation, and poor cranking at low temperatures could occur.3 Fuel typically condenses in the oil during driving conditions in which the oil never warms (wintertime short trips in northern climates). Under these conditions the detrimental effects of fuel in the oil can be minimized by reducing the oil-change interval, as suggested in typical vehicle owner’s manuals. If ambient temperatures are moderate to high and trips are sufficiently long that the oil reaches equilibrium operating temperatures, substantial amounts of fuel in the oil (greater than around 3%, for example) indicate a possible vehicle malfunction with the fuel supply system (a plugged fuel injector or a stuck choke, for example). Copyright © 1994 CRC Press, LLC
Volume III
• • •
If oil analysis indicates glycol in the oil, it may mean:
27
The original oil contained a component behaves like glycol in a glycol test. Combustion products have entered the oil and given a positive test for glycol. There is a coolant leak.
Even if there is a significant coolant leak, glycol may not be detected since glycol can be removed or changed chemically when heated in oil.17 The cause of a coolant leak should be identified and corrected. Infrared spectra yield much information on oil condition when correctly interpreted.21 Warning limits for spectra, as presented in Table 4, are somewhat vague because other chemical species may interfere with spectral peaks of interest. In addition, species of interest may change chemically, with a consequent shift in their absorption band, even though these species may retain many of their original functions. An example is the disappearance of the ZDP antiwear additive absorption band without a corresponding increase in wear.8 Differential scanning calorimetry, useful in determining the oxidation resistance of an oil, may also provide insight into its wear protection as well,2,26,27 since ZDP provides both antioxidant and antiwear protection.
MATHEMATICAL MODELS FOR OIL DEGRADATION AND ENGINE DAMAGE
Various mathematical models have been created to relate chemical or physical changes in the oil to mechanical or physical measurements in an engine. The ultimate purpose of some of these models is to predict the appropriate point at which to change the oil in the vehicle (that is, to develop oil-change indicator systems based on engine measurements). These models differ substantially, as can be seen in the following descriptions from Reference 28.
Chemical Reactor Model One of the earliest and most thorough studies of aging of engine oil was done by Dyson, Richards, and Williams, who developed equations to predict the loss of oil alkalinity in Diesel engines.29 These authors assumed that an internal combustion engine behaves as if it were a chemical reactor and found that the rate of throughput of fuel and the fuel sulfur content govern the loss of oil alkalinity. Volume Effects Sobanska and Wachal30 assumed that, as oil ages, its properties change; oil consumption causes a reduction in the volume of the oil that has aged; and each addition of fresh oil creates an additional volume of new oil which then begins to age. Thus, the authors calculated the age of the oil in each separate volume segment as a function of time and noted the age of each segment in a table.
Average Oil Life When makeup oil is added to partially aged oil in an operating engine, the makeup oil will be exposed to harsh engine conditions for a shorter period of time than the original oil. Bardy and Asseff31 developed the concept of “average oil life” to describe the effects of oil of different age in the system. To determine average oil life, these authors multiplied the weight of each portion of oil by the number of hours on test for that portion. The various
Copyright © 1994 CRC Press, LLC
28
CRC Handbook of Lubrication and Tribology
relative ages were summed, and the sum was divided by total oil weight to yield the average oil life.
Chemical Kinetics Yasutomi et al. considered that chemical kinetics govern the rate of loss of beneficial chemical properties of the oil.32,33 For a variety of engines, these authors determined the reaction rate constants for oil parameters such as acid number or antioxidant stability.33
Oil Film Surface Area and Number of Combustion Events Mahoney et al.34,35 related the rate of loss of antioxidant for several engines to the number of combustion events multiplied by the surface area swept out by the pistons. They determined the minimum volume of fluid which must have been exposed if one assumed that all the antioxidant in that volume were destroyed. Their value for this “antioxidant loss” volume was 0.23 mm3 per combustion event for fresh oil.35
Severity of Service Schwartz and Smolenski2,36 considered that the rate of loss of oxidative stability of the engine oil is the best determination of oil age. Other indications of oil aging such as acid number, accumulation of insoluble compounds, or loss of antiwear protection were found to occur only after the oxidative stability of the oil had been reduced in their tests. Their studies indicated that the most important analytical method for predicting the loss of oxidative stability of engine oil was oxidation induction time by DSC, which remained high so long as the oil operated at temperatures near 110°C, but which diminished more rapidly when the average oil temperature during operation was above or below 110°C.2 Their parabola-shaped oil aging curve (minimum rate of oil aging near 110°C, faster oil aging at other temperatures) was in substantial agreement with the rate of oil aging in operating vehicles.2 The model of Schwartz and Smolenski is the basis for oil-change indicator systems in a variety of production vehicles.36
Volume and Combustion Events Model Schwartz28,37 extended the previous model to include volume changes as well as chemical effects such as coolant leaks and fuel in the oil. Loss of oxidative stability of oil in an operating engine was modeled mathematically as if it were a much simpler system consisting of a reservoir from which oil could be removed, added, or reacted. Simple equations of chemical mixing were used to predict the effects of oil volume changes. To determine the extent of loss of oxidative stability, it was only necessary to count the number of combustion events, correct for volume effects, and correct for the presence of reactive chemicals such as fuel. The rate of oil aging in road tests (engine size from 2.51 to 5.71, a broad range of service conditions including pulling a trailer) as determined by this model correlated well with the rate of aging determined by DSC.2,37 Engine Servicing Model A different approach toward modeling oil-change intervals, by Wisehaupt, Muehlberg, and co-workers,38.39 related engine parameters to frequency of servicing. Their model is also the basis for production oil-change indicator systems.39
Application of Models of Oil Aging and Engine Durability Each of the models listed previously is or could be the basis for an oil-change indicator system using the assumption that oil aging or engine degradation is predictable from engine measurements. Basing oil-change indicator systems on a model assumes that different oil
Copyright © 1994 CRC Press, LLC
29
Volume III
brands exhibit enough similarity that a model will be suitable. When an oil quality (API SF, for example) has been marketed for a long time, the authors have indeed observed some similarity among many oils, since all had to meet the same rigorous tests, while competition typically forced prices, and therefore the additive treatments, to remain within a limited range. If the average oil quality in the marketplace improves, for example by progressing from one quality grade such as SF to the next (SG), models in oil-change indicator systems must be updated. The models must also be updated if changes in engine design, fuels, or materials affect engine or oil durability.
DIRECT SENSING OF OIL QUALITY
Models of oil aging, as described previously, provide valuable information to a driver so long as engine conditions are within the predictive capability of the model. A model assumes a particular oil quality and does not adjust for good or poor oils. Models of oil aging as used in oil-change indicator systems also cannot identify: • • • •
Defective or misblended oils Wrong oils put into the crankcase Oils contaminated with inappropriate fluids Oils exposed to dusty environments
Because of these shortcomings, it is desirable to measure directly the status of the oil. Such methods fall into two categories: oil-sampling techniques and on-board measurement of oil degradation or engine damage.
Oil Sampling Techniques To monitor the status of oil in their vehicles, fleet operators have routinely used oil sampling techniques which may include any of the analyses described earlier.6 Additional procedures such as ferrography have also been found useful.40 If time is at a premium, analysis of one or a few selected oil properties can be completed in the field in a matter of minutes.41.42 In some cases an oil-analysis laboratory may create a set of condemning limits, computerize them, and provide an automatic, computerized interpretation of the status of the oil. While such interpretations are often helpful, occasionally they are misleading. For example, in one case with an elevated level of silicon in the oil, the computerized interpretation suggested a coolant leak. There were no other indications of coolant leaks such as glycol in the oil in conjunction with elevated levels of sodium, or increased amounts of insoluble materials. Since the vehicle was new, the authors suggested to the owner that perhaps he was seeing a “green engine” effect in which normal residues of fluids from the manufacture of the engine showed up in the engine oil. During the first few oil-change intervals, engine oil typically has these small amounts of contamination from various industrial or processing fluids. On-Board Measurement of Oil Properties To obtain an instantaneous on-board measurement of oil quality, it is desirable to place an oil-monitoring device directly into the oil of an operating vehicle. Technical difficulties in the successful development of such a device include operation: • •
Over a temperature range from -40°C to approximately + 175°C With extreme oil contamination from water, fuel, acids, sludge, and varnish
Copyright © 1994 CRC Press, LLC
30
CRC Handbook of Lubrication and Tribology
• •
Under severe vibration For the life of the vehicle (preferably)
•
Low-cost and durable infrared sources and filters for the discrete wavelengths where chemical species of interest exhibit absorption, A durable cell material that resists fouling and is transparent to infrared, and Acceptable detectors and associated electronics for signal processing.
While the authors know of no devices for direct determination of oil quality as original equipment in production passenger cars, the literature describes sensors which measure one or more aspects of oil quality.18,43–48 Any of the analytical techniques discussed earlier theoretically have potential for application in an on-board oil-condition sensor. Devices exist for on-line determination of viscosity.48 However, an oil may be well beyond its recommended change point as determined by acidity or oxidative stability without exhibiting an excessive viscosity change. Infrared spectroscopy can yield a wealth of information about oil condition. It should be possible to develop an infrared spectroscopic sensor for measuring oil quality18.43 if all the following features are provided: • •
Other types of sensors may indicate various oil properties in an indirect way. For instance, it may be possible to correlate the output of sensors which measure oil conductivity, impedance, or dielectric properties to such oil properties as acidity, alkalinity, or contaminant accumulation.44–47 Optical techniques such as refractive index, visible light absorption, or fiber optics, have also shown promise for indicating fluid properties.49 Another approach involves a chemical “fuse” that progressively dissolves or degrades as the oil ages.50 It remains to be determined whether these various devices can provide reliable results in all types of service, in the presence of water (as found in low-temperature, short-trip driving), and in the presence of varnish and sludge. The most likely chance of success for a broadly applicable oil-quality sensor probably lies with a sensor array: a complementary combination of two or more of the most promising sensors. Much progress has been made, but the challenge still remains to produce a low-cost, durable, fouling-resistant sensor that accurately determines the end of useful oil life over all operating conditions normally encountered and with a great variety of oil types.
CLOSURE
The fleet owner or individual driver who wants to determine the correct time to Change engine oil has the option of using sampling techniques, one of the modeled systems for oil age or engine degradation, or a sensor. The sampling or modeling methods have enough documented reliability mat the user can feel confident about the results. With continued testing, it is anticipated that several of the sensors for direct measurement of oil quality will also be proven in the field.
ACKNOWLEDGMENT
The authors thank R. H. Kabel (retired), for extensive input to some sections of this chapter; N. M. Potter and T. J. Chapaton of the General Motors North American Operations Research and Development Center for their comments regarding oil analysis, and the General Motors North American Operations Research and Development Center for supporting the completion of this chapter. Copyright © 1994 CRC Press, LLC
Volume III
REFERENCES
31
1. Shilling, A., Motor Oils and Engine Lubrication, 2nd ed., Scientific Publications (G.B.) Ltd., Broseley, Shropshire, England, 1968. 2. Schwartz, S. E. and Smolenski, D. J., Development of an automatic oil-change indicator system, (Society of Automotive Engineers) No. 870403, 1987. 3. Schilling, A., Automobile Engine Lubrication, Scientific Publications (G.B.) Ltd., Broseley, Shropshire, England, 1972. 4. Asseff, P. A., Used engine oil analyses-review, SAE Pap. No. 770642, 1977. 5. O’Hara, J. P., Sarkis, A. B., and Kennedy, W. A., Equipment protection through customized oil analysis, SAE Pap. No. 730745, 1973. 6. Analysis programs for used engine oils, International Trucks Technical Service Information, TSI-85–43, October, 1985. 7. Lube Oil Analysis Primer for Diesel Engines, Detroit Diesel Allison, December, 1984. 8. Used Oil Analysis, Chevron Chemical Company Report 853. 9. Testing Used Engine Oils, Chevron Research Bulletin, 1983. 10. Schwartz, S. E., Observations through a transparent oil pan during cold-start, short-trip service, SAE Pap. No. 912387, 1991. 11. 1990 Annual Book of ASTM Standards, American Society for Testing and Materials, Philadelphia, Section 5, Volumes 5.01–5.03, 1990. 12. SAE recommended practice, physical and chemical properties of engine oils—SAE J357 JUN86, SAE Handbook, Society of Automotive Engineers, Warrendale, PA, 1987. 13. SAE recommended practice, engine oil viscosity classification—SAE J300 OCT91, SAE Handbook, Society of Automotive Engineers, Warrendale, PA, 1992. 14. Firey, J. C., Newcomb, J. C., Niemann, J. F., and Sugges, P. R., Studies of the effects of water on gasoline engine wear at low temperature. Wear, 10, 33, 1967. 15. Graf, R. T., Copan, W. G., Kornbrekke, R. E., and Murphy, J. P., Sludge formation in engine testing and field service, SAE Pap. No. 881580, 1988. 16. Hudgens, R. D. and Stehouwer, D. M., Coolant contamination of diesel engine oils, SAE Pap. No. 840343, 1984. 17. Artenem’ev, V. A., Boikov, D. V., Koltin, I. P., and Timashev, V. P., Changes in the properties of crankcase oil when antifreeze enters the lubricating system, Chem. Tech. Fuels Oils, 17, 513, 1981. 18. Wooton, D. L., Lawrence, B. J., and Damrath, J. G., Infrared analysis of heavy-duty engine oils, SAE Pap. No. 841372, 1984. 19. Coates, J. P. and Setti, L. C., Infrared spectroscopy as a tool for monitoring oil degradation, in Aspects of Lubricant Oxidation: a Symposium, ASTM STP 916, American Society for Testing and Materials, Philadelphia, 57, 1983. 20. McGeehan, J. A. and Fontana, B. J., Effect of soot on piston deposits and crankcase oils—infrared spectrometric technique for analyzing soot, SAE Pap. No. 801368, 1980. 21. Dotterer, G. O., Jr. and Helmuth, W. W., Differential infrared analysis of engine oil chemistry in sequence V tests, road tests, and other laboratory engine tests, Lubr. Eng., 41(2), 89, 1983. 22. Smolenski, D. J. and Kabel, R. H., Effect of engine oil zinc dithiophosphate (ZDP) additive type on cam and lifter wear in taxi service, SAE Pap. No. 831760, 1983. 23. Blane, R. L., Oxidative Stability of Oils and Greases, DuPont Application Brief, TA 41, 1974. 24. Walker, J. A. and Tsang, W., Characteristics of lubricating oils by differential scanning calorimetry, SAE Pap. No. 801383, 1980. 25. Levinson, H., Limitations of atomic absorption spectrophotometry applied to spectrometric oil analysis, ASLE Trans., 27, 24, 1984. 26. Kalnicky, D. J., Barbi, N. E., Schnerr, G., and Hirsch, P., A new quality control instrument to determine concentration of inorganic additive elements in petroleum products, SAE Pap. No. 840265, 1984. 27. Smolenski, D. J. and Kabel, R. H., Evaluation of cam and lifter wear and oil thickening with low-phosphorus engine oils in taxicab service, SAE Pap. No. 861516, 1986. 28. Schwartz, S. E., A model for the loss of oxidative stability of engine oil during long-nip service. I. Theoretical considerations, SIZE, Tribology Trans., 35(2), 235, 1992. 29. Dyson, A., Richards, L. J., and Williams, K. R., Diesel engine lubricants: their selection and utilization with particular reference to oil alkalinity, in Proc. Inst. Mech. Eng., 171, 717, 1957. 30. Sobanska, K. and Wachal, A., A mathematical model of the ageing process of oil lubricating a mechanical device, including the renewal, Eurotrib ‘81—Proc. 3rd Int. Tribology Congr. Warsaw, Poland, Elsevier Scientific, Amsterdam, 223, 1982. 31. Bardy, D. C. and Asseff, P. A., Motor Oil Thickening—A CLR engine test procedure which correlates with field service, SAE Pap. No. 700508, 1970. 32. Yasutomi, S., Maeda, Y., and Maeda, T., Kinetic approach to engine oil. 1. Analysis of lubricant transport and degradation in engine systems, Ind. Eng. Chem. Prod. Res. Dev., 20(3), 530, 1981.
Copyright © 1994 CRC Press, LLC
32
CRC Handbook of Lubrication and Tribology
33. Yasutomi, S., Maeda, Y., and Maeda, T., Kinetic approach to engine oil. 2. Antioxidant decay of lubricant in engine system, Ind. Eng. Chem. Prod. Res. Dev., 20(3), 536, 1981. 34. Mahoney, L. R., Korcek, S., Hoffman, S., and Willermet, P. A., Determination of the antioxidant capacity of new and used lubricants; method and applications, Ind. Eng. Chem. Prod. Res. Dev., 17(3), 250, 1978. 35. Mahoney, L. R., Otto, K., Korcek, S., and Johnson, M. D., The effect of fuel combustion products on antioxidant consumption in a synthetic engine oil, Ind. Eng. Chem. Prod. Res. Dev., 19(1), 11, 1981. 36. Schwartz, S. E., Smolenski, D. J., Wisehart, A. J., and Nguyen, T. N., Development of an Automatic Engine Oil Change Indicator System, U.S. Patent No. 4,762,476, May 3, 1988. 37. Schwartz, S. E., A model for the loss of oxidative stability of engine oil during long-trip service. II. Vehicle measurements, STLE Tribology Trans., 35(2), 307, 1992. 38. Wisehaupt, W., Service-Intervall-Anzeige, Eigendiagnose. Verschleissgerechte Fahrzeugwartung, Motor, 16, 1, 1984. 39. Muehlberge, H., Starmuehle, E., Weishaupt, W., Flohr, P., and Bourauel, F., Service Interval Display for Motor Vehicle, European Patent No. 57820, August 18, 1982. 40. Anderson, D. N., Hubert, C. J., and Johnson, J. H., Advances in quantitative analytical ferrography and the evaluation of a high gradient magnetic separator for the study of diesel engine wear, SAE Pap. No. 821194, 1982. 41. Kauffman, R. E. and Rhine, W. E., Development of a RULLET (remaining useful life of lubricant evaluation technique). II. Colorimetric method, Lubrication Engineering, 44(2), 162, 1988. 42. Geary, P. A., Jr., Evaluation of in-service industrial lubricants through oil analysis kit methods, Lubr. Eng., 40, 352, 1983. 43. Stuart, A. D., Trotman, S. M., Doolan, K. J., and Fredricks, P. M., Spectroscopic measurement of used lubricating oil quality, Appl. Spectrosc., 43, 55, 1989. 44. Hellwig, G., Normann, N., and Uhl, G., Ein Sensor auf dielektrischer Basis zur On-Line-Characterisierung von Motorenölen (Alkalinität, Viskosität), Mineralöl. Techn., 10, 1, 1988. 45. Kauffman, R. E., Development of a remaining useful life of lubricant evaluation technique. III. Cyclic voltammetric methods, Lubr. Eng., 45, 709, 1989. 46. Kato, T. and Kawamura, M., Oil maintenance tester: a new device to detect the degradation level of oils, Lubr. Eng., 42, 694, 1986. 47. Meitzler, A. H. and Saloka, G. S., Method and Apparatus for Sensing the Condition of Lubricating Oil in an Internal Combustion Engine, U.S. Patent No. 4,733,556, 1988. 48. Martin, S. J., Granstaff, V. E., and Frye, G. C., Characterization of a quartz microbalance with simultaneous mass and liquid loading. Anal. Chem., 63, 2272, 1991. 49. Seitz, W. R., Chemical sensors based on fiber optics, Anal. Chem., 56, 16, 1984. 50. Cipris, D., Walsh, A., and Palinasamy, T., Sensor for motor oil quality, Proc. Symp. Chemical Sensors, Electrochemical Society, Vol. 87–9, 401, 1987. 51. Rodgers, J. J. and Kabel, R. H., A Revised Sequence IIIC Engine Oil Test, General Motors Research Laboratories Pub., Warren, MI, 2611, 1978. 52. The ILSAC Minimum Performance Standard for Passenger Car Engine Oils, Appendix L, pp L–2 to L–10, American Petroleum Institute Engine Oil Licensing & Certification System, Draft Document, American Petroleum Institute, Washington, November 15, 1991.
Copyright © 1994 CRC Press, LLC
DIESEL ENGINE LUBE ANALYSIS Jack Poley
BACKGROUND AND DEVELOPMENT OF USED LUBE TESTING
Used lube analysis has existed as long as lubricants have been used. The very first means of inspecting lubricants included appearance: looking at the lube and assessing whether it is “dirty”, water-laden, and so forth, based on its color and other visual properties. Since the advent of detergent/dispersant lubricants, appearance has less value because the lubricant plays additional roles to its traditional friction-reducing and cooling functions, such as cleaning and scavenging. Thus a detergent removes particles which have accumulated on surfaces within the engine, while dispersants maintain these particles in suspension long enough for them to be trapped by filtration or drained with the lubricant. Virtually every diesel lubricant turns back from fuel “soot” within hours after a fresh lube charge (other than that used in marine-type crosshead piston diesels). One would have little success using appearance as a primary criterion for today’s used diesel engine lube evaluation. Another practical tool used in early oil testing was the blotter test, where a drop of used oil was placed on blotter material. A crude “soot” indication (darkness of the blot) and “dispersion” indication (the rough diameter of the dark material) were derived. Adding a drop of pH indicator fluid might indicate strong acid development, provided a potential color change on the blotter could be discerned in spite of the soot’s opacity interference. Similarly, colorimetric glycol-reacting reagents were also drop-added to detect possible coolant contamination. In the late 1940s, the railroad industry foresaw an advantage to analyzing used lubricants for various metals found in specific components of the engine. By observing changes in wear metals concentration from one sample to the next (routine periodic sampling is a must in ascertaining wear rates and trends), mechanical maintenance could be anticipated and scheduled in advance of complete component failure and resulting excessive loss of productivity from the engine. While the original emission spectrographs (which involved techniques of film exposure, developing and post-exposure density evaluation) were tedious and produced only a handful of answers in a workday, the technique proved valuable until semi-automated instrumentation became available nearly a decade later. Today spectrometric analysis for as many as 20 or more individual elements is the backbone of diesel engine used oil analysis, and the process takes no more than a minute in high-volume production laboratories using modern optics, electronics, and computer-based processing. Having this additional information on engine wear allowed the railroads to schedule teardowns on the basis of need rather than arbitrary hours of operation. The addition of a spectrometric metals analysis gave birth to “predictive maintenance”, a vast improvement over preventive maintenance. Today virtually every segment of the American military (and many other military organizations) utilizes this technology, and most of the private industry sector employs this technique to at least a cursory extent. Beginning in the mid-1970s through today, two additional areas gained significant attention in lubricant monitoring: 1. Contamination products external to the primary lubrication environment. 2. Degradation of the lubricant as evidenced in deterioration of its inherent properties. 0-8493-3903-0/94/$0.00 + $.50
© 1994 by CRC Press, Inc. Copyright © 1994 CRC Press, LLC
33
CRC Handbook of Lubrication and Tribology
34
In the next section, analytical methods are given for monitoring contamination by fuel soot, water, glycol and fuel; deterioration of the oil itself by oxidation, nitration, shearing, and additive depletion; and accumulation of engine wear particles.
TYPICAL ANALYTICAL TESTS FOR DIESEL ENGINE OILS
Contamination Fuel Soot (Combustion Solids) Fuel soot is a natural consequence of the diesel engine combustion cycle. The material is extremely fine in nature, below 1 micrometer (micron) in particulate size in most cases and thus not yet cost-effective to filter with today’s technology. Insolubles measurement via centrifuging had proved useful, but the advent of dispersant additives rendered centrifuging difficult when certain additive packages were employed in the lubricant. Further constraints (mostly environmental), in terms of the testing chemicals that can agglomerate or pack these small particles into larger ones so that they could be centrifuged, forced the technology to change. Today thermogravimetric analysis (TGA) is recognized as an accurate means of addressing fuel soot, although it is somewhat tedious. Modern infrared spectroscopy has given rise to computer-based algorithms which have shown reasonable correlations with TGA to levels of 1.0%, and this approach appears most promising.
Thermogravimetric analysis (TGA)—This is a sensitive research-oriented test with very good precision, involving combustion and residue weight analysis. Infrared analysis (with fast Fourier transforms)—Infrared spectrometric analysis provides information concerning molecular structure of the base components, additives and contaminants within the lubricant. Dispersion algorithms are employed to estimate fuel soot. This approach is far less time consuming than TGA, and therefore more practical for routine monitoring via scheduled sampling.
Water Water is readily detected in a standard hydrocarbon lubricant because it will not appreciably mix with it. Water levels above minimal condensation levels are potential hazards to adequate lubrication. Original qualitative methods included simply dropping a small amount of oil on a hot surface to note whether tell-tale sputtering would occur. Today infrared analysis and/or water-specific titrations provide a more quantitative approach to measuring this contaminant. Infrared analysis (with fast Fourier transforms)—Water is detectable at moderate levels (>1000 ppm) with this technique. Karl-Fischer titration—Very good sensitivity at parts-per-million levels (1 ppm = 0.0001%); not required nor commonly used, however, for traditional used diesel lube inspection.
Glycol Infrared analysis (with fast Fourier transforms)—Glycol’s functional molecular grouping can sometimes be isolated with this technique. Coolant additive metals as indicators (from spectrometric analysis below)—Elements such as sodium, potassium and boron are often telltale indicators of coolant in the lubricant, but comparison with fresh oil should be accomplished so as not to confuse these elements with normal oil additives. Colorimetric methods (e.g., potassium periodate)—A simple field-oriented test can be made to ascertain if glycol is present by color indicator change; however, care must be taken Copyright © 1994 CRC Press, LLC
Volume III
35
not to mistake small amounts of harmless glycols in some fresh lubes as evidence of a coolant leak.
Fuel Raw fuel can enter a diesel crankcase through a number of possibilities such as: worn rings and pistons, (as applicable per fuel system type) leaking injector seals, jumper lines or pumps, or poor injector spray patterns. ASTM D322 steam distillation, while originally designed for gasoline engines (and now retired as a test method), was adaptable to diesel engine lube testing, producing acceptable results until modern instrumentation precluded the practicality of the method. Today, gas chromatography and, on a more limited basis, infrared absorption spectroscopy offer quantitative to semiquantitative approaches to fuel contamination measurement. Flash point—Flash point is among the oldest methods for detecting and estimating fuel contamination. The apparatus consists of a metal cup into which the sample is placed. The cup is heated in accordance with a prescribed, programmed temperature gradient. A small flame is periodically passed over the cup as it is heating until a brief flash occurs. Lower flash temperatures than the fresh lubricating oil suggest fuel contamination. Sensitivity of this test is limited to levels in the range of 3 to 4% and greater. Infrared analysis (with fast Fourier transforms)—Not surprisingly, fuel molecules are quite similar to lube molecules, only smaller in size. This similarity makes it extremely difficult for infrared analysis, which is chemical-group sensitive, to discern minute amounts of fuel in a lubricant. Lower detection limit is at best 3 to 4%. Gas chromatography—This technique is the equivalent of a microdistillation, and is probably the most sensitive method available at present. Levels as low as 1.5% by volume are usually discernable. Some diesel engine mechanics have reported the ability to identify fuel leak sources at levels of contamination this low. Degradation Viscosity The viscosity of a lubricant is always of fundamental interest in an inspection, first to verify that the correct product grade is in service, then to ensure that no significant deviation has occurred during service life. Since viscosity is flow resistance with respect to temperature, the inspection methods utilize instrumentation with known, calibrated flow paths and constanttemperature heating devices. Today the technology can be computer-driven, minimizing error sources. Thickening precursors
Lube oxidation Lube nitration Incorrect product added as makeup
Thinning precursors
Fuel contamination Lube additive shearing—Multigrade lubricants contain a high molecular weight polymer additive to minimize viscosity change with temperature (i.e., improve its viscosity index). During engine operation, this additive is sheared into smaller molecular structures, causing tangible loss of viscosity at engine operating temperatures. If the polymer becomes fully sheared, the viscosity ultimately returns to that of the original base stock. Incorrect product added as makeup
Copyright © 1994 CRC Press, LLC
36
CRC Handbook of Lubrication and Tribology
Infrared Analysis Lube oxidation—Reaction of oxygen with the molecular structure of the lubricant is usually accompanied by significant viscosity increase and resulting loss of lubricating effectiveness. Lube nitration —Same as oxidation, but involving reaction with nitrogen instead of oxygen (nitrogen is, of course, an element available from air). Additive Depletion/Degradation Additives play an essential role in effective lubrication of today’s diesel engines. If their active level or functionality is reduced or impaired substantially, the lubricant’s ability to reduce friction and control wear is at risk, and so is the engine. Reduction in antioxidant, antiwear, and dispersant additives; and additive precipitation due to hydrolysis from water contamination are all forms of additive depletion/degradation. However these parameters are very difficult, if not impossible, to effectively monitor with simple laboratory testing. One exception is the depletion of total base number (TBN), the additive property specifically needed in the lube for neutralization and control of sulfur acids from normal fuel combustion. A routine titration provides effective measurement of this property, and the process can be semi-automated for relatively inexpensive throughput. While TBN is important, today’s environmental regulations have severely limited the amount of fuel that diesel oil may contain, greatly minimizing the need for a significant TBN property. Wear Particles Spectrometric Analysis The “backbone” of used lube testing for diesel engines, it is still one of the most important tests for diesel engine diagnostics. The ability to analyze for perhaps two dozen elements simultaneously and at relatively little expense makes this the focal point of the testing sequence. Spectrometric testing must be based on good sampling techniques and competent interpretation. It alone should rarely be the basis for engine disassembly. While railroad industry efforts in developing spectrometric analysis for metals was a major milestone, there were failures that did not “show” in the metals analysis. This lack of correlation seemed to be more telling on engines featuring lower BMEP (brake mean effective pressure) ratings and/or engines operating at relatively low rpms. Additional research established that the emission (or absorption) spectrometric technique had a significant blind spot once particle size reached levels much above a few microns; at 10 µm the spectrometers were virtually unable to detect metallic particles, owing to their inability to properly vaporize the larger particles, a process necessary to detection. While the actual detection size of an element depended on its specific state (whether in a compound, an alloy, or elemental state), the general physical properties of the element, and overall size of the particle involved, it became clear that spectrometric elemental analysis alone was not sufficient for detecting abnormal wear conditions. Ferrography, a relatively new technology for evaluating wear particles in lubricants, offers insights into particles larger than 10 µm. Figure 1 compares the general range of utility for various methods of analyzing particulates. Direct Reading (DR) Ferrography An abbreviated version of analytical ferrography, DR ferrography develops a ratio of “L” or “large” (roughly >5 µm in size) to “S” or “small” (< 5 µm in size) ferrous particles to differentiate fatigue or catastrophic wear from normal (expected) rubbing wear. It is not Copyright © 1994 CRC Press, LLC
Volume III
37
FIGURE 1. Comparison of particulate study methods.
unusual for DR ferrography to fail to correlate with iron readings from spectrometric analysis, again owing to spectrometer inability to detect particles in the larger domain. Analytical Ferrography
One of the most powerful and incisive tests for detecting and evaluating wear particles, the technique magnetically “combs out” (on a microscope slide) ferrous particles from the oil in an ordered large-to-small fashion. Nonmagnetic metals and debris randomly precipitate as well, based on size and density. The slide is then scanned carefully, using a specially designed microscope. The principal advantage is the ability to view directly the particles of interest, noting their morphology and size.
SAMPLING
An oil analysis result is only as good as the sample taken. Good sampling technique incorporates the following three factors:
1.
Location of the sampling point, as shown in Table 1 - The oil should be warm and well circulated prior to sampling.
Copyright © 1994 CRC Press, LLC
38
CRC Handbook of Lubrication and Tribology
2.
Sampling frequency—Table 2 gives recommended intervals for sampling. These generally coincide with the oil change interval; however, interim samples are also useful when a suspected problem is being tracked. Sampling should occur at regular intervals. Data recording is an essential but often neglected aspect of oil analysis—Table 3 lists the information which should be recorded. These data are necessary for interpretation of analytical results and should be provided to the laboratory and a copy retained with maintenance records.
3.
INTERPRETATION OF TEST RESULTS
Upon completion of the analysis, the data have to be related back to the engine if they are to be of value. This task may be undertaken by the end-user or the laboratory. Interpretation is optimized when sample results can be compared to used oil analytical data from similar engines and applications. These limits may indicate the need for an oil drain or resampling for verification purpose. With the emphasis placed on computers and data processing nowadays, it is de rigueur to utilize these tools in the evaluation of test data. There are several viable approaches to data analysis, and it is often best to combine the techniques to achieve full advantage and insight. Copyright © 1994 CRC Press, LLC
Volume III
39
Limits Limits can also be established for an engine based on a regular sampling program. These limits are generally statistically based on quantities reported which are a set amount above the average. In extreme cases, the alert may require the engine to be taken from service for corrective action and/or disassembly for inspection. Limits are usually the first data traps one would choose to set. Table 4 shows statistical analyses from 9999 samples in a diesel engine application. It is useful to note there are some cross-application similarities. Naturally, one would wish to focus on specific makes and models of engines in developing a more finely tuned set of constraints. Here the important observation is that median values are relatively close to average values, suggesting the data are reasonably valid for developing tabular limits. One would opt, perhaps, to constrain a limit to one or two standard deviations from median (or average), dependent on the “tightness” of the data and other observations, including subjective experience. Statistics should not be a substitute for common sense and experience in specific instances.
Trends Trending is at least as important as limits in the evaluation of used diesel engine oils. It is possible that a change in concentration is significant, even though a limit may not have been exceeded. It is important to treat the notion of repetitive and historical data collection and evaluation as a major principle of used lube analysis consulting. Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
40
Combinations Using both limits and trends one can develop tabular systems for computerized evaluation or data flagging, as given in Table 5, for example, for the application represented in Table 4.
Statistical Database Query Once some basic statistics are collated, it is often useful to check for interrelationships of data. The graph of Figure 2, covering monthly sampling over a 1-year span, shows moderate to poor correlation between iron from spectrometric readings and DR ferrographic “L” results. While the conclusions and theories that may be drawn from this representation are complex, it is obvious that such an approach can lend more clarity to the overall evaluation of an engine’s condition on a dynamically progressive basis. “L” values suggest fatigue or chunk (abnormal) wear, whereas spectrometric testing addresses particles that are expected as a normal part of the wear process (mostly particles less than 1 µm), unless the total amount is excessive or drastically changed from me previous sample’s data. If these data had correlated well, it is likely that abnormal wear occurred over a lengthy time, eventually manifesting itself in the smaller particles as well as the larger ones; or that enough small particles had been eroded (from dirt, acid or fuel thinning of the oil, e.g.) to lead to excessive clearances and subsequent chunking wear.
FUTURE OF LUBE ANALYSIS FOR DIESELS
As long as we have diesel engines, lube analysis will be associated with their maintenance. Meanwhile productivity requirements, as well as engine cost and maintenance costs, will surely continue to soar. Lube analysis offers a proven and cost-efficient way to assess machinery and lubricant condition, because it allows continued operation while the evaluation takes place. Some refinements and developments which might be expected in the future: 1.
2.
Increasingly specialized instrumentation for lube analysis consulting; more incisive and finely focused inspections. High-speed, semi-automated testing is now available for spectrometric metals, viscometric and infrared testing. We can surely expect additional capabilities in this area. Personal computers will likely be commonplace at the maintenance site. Data can currently be received via modem from the laboratory. Resident software is available
Copyright © 1994 CRC Press, LLC
Volume III
41
FIGURE 2. Correlation study: spectrometric iron vs. direct-reading ferrography.
for screening and re-transmitting the report, as well as statistically analyzing the data. Software for integration of oil analysis data with other diagnostic disciplines, e.g., thermography and vibration analysis, is already available in the industrial sector and will find its way to the automotive sector.
On-board sensors for evaluation of lube and even some wear conditions, leading to more sophisticated laboratory testing on an exception basis, are currently receiving a lot of attention. While this development may curtail routine oil analysis in some instances, it will not likely eliminate the need for testing. What seems most plausible is that the level of testing sophistication will rise to another echelon, complementing information obtained from the engine’s sensors.
BIBLIOGRAPHY
Beck, J. W. and Johnson, J. H., The application of analytical ferrography and spectroscopy to detect normal and abnormal diesel engine wear, SAE Pap., No. 841371, October 1984. Whitham, D, and Poley, J., Equipment management: exploiting the lube analysis data base, SAE Pap., No. 880780, April 1988. Minges, S. et al., Oil analysis: how to get your money’s worth, Construction Equipment, September 1990. Poley, J., The oil analysis power user, Lubr. Eng., 46, 630, (1990).
Copyright © 1994 CRC Press, LLC
ROTATING MACHINERY VIBRATION TESTING, CONDITION MONITORING, AND PREDICTIVE MAINTENANCE William D. Marscher
INTRODUCTION
One of the most common problems in rotating machinery installations is vibration. Lateral shaft vibrations (rotor dynamic motions perpendicular to the machine rotational axis) are those most often discussed. Troublesome vibration can also occur in the axial direction, in torsional oscillation, and in the stationary components and supporting structure. Other parameters besides vibration are indicators of machinery health. Bearing temperature, rate of lubricant oxidation or contamination by wear debris, and leakage flow are useful in this regard and can be more sensitive than vibration to certain problems. This chapter gives guidelines for monitoring all of these parameters and for including them in a predictive maintenance program. The strong emphasis given to vibration testing and analysis reflects the consensus that vibration monitoring is a key element in any general purpose rotating machinery predictive maintenance program. Further details on instruments and procedures for monitoring lubrication and other parameters are provided in companion Handbook chapters in this section on Monitoring and Maintenance.
CONDITION MONITORING
Testing of machines for vibration levels and other critical parameters (for example, machine efficiency) according to a consistent, repetitive schedule is called “condition monitoring”. Generally, condition monitoring standards should be based on good and bad experience with a given type of machine. Various general purpose specifications, which should be used only as guides, are available from the American Petroleum Institute (API), the American National Standards Institute (ANSI), the Hydraulic Institute (HI), the German Engineering Standards Society (VDI), the International Standards Organization (ISO), and the National Electrical Equipment Manufacturer’s Association (NEMA), among others. However, because detailed data on failure rate are limited (especially for a “new” or custom-designed machine), it is necessary to extrapolate existing experience through analysis. To avoid taking extrapolations too far, their objective should be limited to answering two questions: 1.
2.
What is the likelihood of damage? What will high vibration at a given location, frequency, load, etc. damage first in the machine? At what levels is this possible, likely, or certain? Do the observations suggest something is wrong? Above normal vibration can be signs of excessive forces, used up clearances, serious imbalance, misalignment, or damaged components.
Answering these questions has led to various “safe”, “alarm”, and “shutdown” criteria. In terms of vibration displacement, these criteria are generally inversely proportional to running speed, and at the common speed of 3600 rpm could reasonably be set for most machines at 2.5 mils (63 µm [microns]), 4.0 mils (102 µm), and 6.0 mils (153 µm), respectively. It is advised that user-established guesses for safe, alarm, and shut-down levels for new machines be based on manufacturer recommendations first, and general purpose specifications second.
0-8493-3903-0/94/$0.00-$0.50 © 1994 by CRC press.Inc
Copyright © 1994 CRC Press, LLC
43
44
CRC Handbook of Lubrication and Tribology
Some condition-monitoring systems have alarm level vs. frequency curves that check for violation by the machine’s current spectrum across the frequency range, where location, direction, and operating condition are taken into account. The more sophisticated of these systems track the test spectra statistically over time, and adaptively revise the alarm and shutdown limits based on what is “learned” over time about the vibration vs. danger in that particular machine or class of machine, in that particular service. Manual trending by maintenance personnel can reach the same objective of establishing sensible vibration limits that do not fail good equipment or pass bad equipment.
PREDICTIVE MAINTENANCE
Vibration signature analysis and observance of other system parameters can be formalized to extrapolate current machine behavior and project when certain components will wear out or require adjustment. This type of effort, known as predictive maintenance, has as its objectives: (1) maintaining thermodynamic performance and efficiency, (2) minimizing downtime and repair expense, and (3) planning shut-downs for repairs. Relevant parameters which should be monitored in pursuit of these objectives are: 1. 2. 3. 4. 5.
Vibration level Process flow, pressure, and power Sealing cavity leakage Lubricant quality, contamination, and temperature Machinery noise
1. 2. 3. 4.
Wear rate of close-running clearances1-4 Bearing and sealing surface deterioration or wear Occurrence of metal fatigue Process fluid flow path changes through erosion
The purpose of this monitoring is to assess machinery degradation rate due to the following:
ESTABLISHING A PREDICTIVE MAINTENANCE PROGRAM
Many companies have had difficulty implementing predictive maintenance systems.1–4 In fact, many such systems have actually increased machine downtime and the total number of repairs, generally by flagging apparent problems that previously would have gone unnoticed and without operating difficulties. A plant in the Near East comes to mind in which the newly installed predictive maintenance system was set to “red flag” all indications of potential problems. The system promptly called for every instrumented machine to be shut down for repairs. The best approach is to follow some simple rules in setting up a predictive maintenance program: 1.
2. 3.
Make it easy. Select a system that is easy to use, that has output which is easy to access and is easy to understand. Make it simple. Include predictions only for those failures based on actual equipment experience (e.g., if you have never experienced a blade loss, do not include blade loss assessment.) Make it fit. Tailor it to your particular machinery, personnel, and conditions. Work with the predictive maintenance system vendor to fit system capabilities to your needs with minimum dislocation. Modify existing procedures as little as possible at first, adding capabilities a step at a time.
Copyright © 1994 CRC Press, LLC
Volume III
4. 5.
6.
45
Make it understood. Fully train the personnel who will use it, so that they thoroughly understand all of the system functions that they will be responsible for, and the background behind any interpretations. Make it routine. Make the measurement locations, test intervals, and maintenance decisions consistent and, as much as possible, automatic. Make it effective. Splice it into your maintenance action procedures and machine inventory lists.
Important questions which should be answered in establishing the maintenance decision rules are: 1. 2.
3.
What failure types have been most common or costly? What criteria are the most trustworthy in diagnosing these failures? The criteria should include bearing housing vibration values in the vertical, horizontal, and axial directions. For high reliability, other measurements should include machine performance, bearing temperatures, lubricant condition, shaft vs. housing vibration levels, and far-field noise. What the limits of acceptability associated with the criteria? Standards organizations like ANSI or ISO52 offer a place to start in establishing vibration specifications and acceptability limits. Fluid dynamic performance, noise, bearing temperature acceptability ranges, and trend interpretations are specific to a given machine and should be defined under the advisement of the manufacturer, tempered by consultant advice and user experience.
PREDICTIVE MAINTENANCE SYSTEM SELECTION
Several major options exist. For example, should the system monitor continuously or only on a transient basis? Transient monitoring is often preferred to avoid the rapid accumulation of vast quantities of data. Examples of transient monitoring are the gathering of a regular weekly or monthly “burst” of data or the triggering of a data accumulation burst “window” if machine vibration or other performance parameters surpass alarm limits. Choice between hard-wired stationary vs. portable probes and monitors involves a tradeoff of consistency and convenient availability for long periods vs. the low cost and flexibility of portable cart-based or hand-held instruments. Convenient periodic “logging” of data with portable equipment requires that the machine be at a point on its performance map close to that of past data-logged readings so that trending can be accomplished. Diagnostics can be made without trending, instead based on machinery vibration-acceptability charts, but this practice is not recommended. An upward trend in vibration readings at any given frequency is, except in extreme cases, more important than the absolute vibration level in assessing machine current condition and impending problems.1,2,5,6 Most predictive maintenance systems involve feeding a central database with information from various hard-wired “satellite” stations located at the plant’s machines, or obtained from roving hand-held data collection units used by maintenance personnel on machines at many locations. With either system, a decision must be made concerning the type of central processor (PC or minicomputer workstation) and its operating system. Generally for PCs this is either DOS (probably together with a local area network or “LAN” to tie individual PCs in your system together) or UNIX. Minicomputers probably will use UNIX, but may use some vendorwritten proprietary system. The choice should be compatible with any systems already available to the maintenance department. At the time of this writing, an advantage of PCbased systems using DOS or DOS/Windows is typically lower cost and that their operating systems need little maintenance. Minicomputers and UNIX generally require at least one Copyright © 1994 CRC Press, LLC
46
CRC Handbook of Lubrication and Tribology
“computer systems” person to troubleshoot and maintain the computer system, and hardware and software cost is typically three to four times that of PC-based systems. Their advantage is that they operate somewhat faster (especially RISC-based systems) and can handle a large number of users simultaneously. Another decision is whether to observe vibration readings “filtered” in a narrow frequency range, “unfiltered” and therefore representative of combined vibration at all frequencies, or as a “spectrum” plot of all vibrations at all frequencies of interest. There are places for each approach. A “short list” of filtered readings at, for example, 1x, 2x, and blade pass frequencies provides quick oversight of most common problems and can be easily tabulated for trending. However, filtered lists cannot be long enough to be all inclusive, and setting of filtered acceptance values independent of frequency, as many specifications do with vibration velocity, places severe and many times unwarranted constraints on vibration at some frequencies, and not nearly enough at others. A vibration velocity of 0.3 inches per second (ips) or 7.6 mm/s, for example, is a good acceptance value at 1x or 2x running speed, is fairly severe for many industrial machines at blade pass frequency, and limits vibration unnecessarily to millionths of an inch at frequencies well above vane pass, causing occasional rejection of machines based on harmless piping acoustics. Conversely, at subsynchronous frequencies, such velocity levels might allow rubbing at close running clearances in the machine. Even at high frequencies, 0.3 ips (7.6 mm/s) may not be low enough to red-flag ball-pass frequency increases which suggest imminent rolling element bearing failure.9 Overall or “unfiltered” vibration has the disadvantage of not providing any clue as to the nature of the problem. However, it has the advantage of simplicity, and in the case where “true peak” shaft displacement is tracked (as opposed to root-mean-square (RMS)-extrapolated “average peak”) has a direct physical correspondence to clearance utilization, at least at the location measured, as pointed out by the excellent German specification VDI-2059. It also can be compared to the vast database of overall vibration vs. running speed for millions of machines readily available in the literature. On the other hand, filtered vibration vs. excitation frequency acceptability data which exist in the literature tend to be relatively specific to machine type, and even to machine manufacturer and to the specific application of the machine. For detailed vibration trouble-shooting of a difficult-to-diagnose problem, however, it is foolish to use anything other than comprehensive spectral data, even if the analysis equipment must be rented or used and interpreted by a consultant. Drawbacks are that spectral data are much more bulky to store, can make trending of specific frequencies complicated because of the extra information present, and can be very confusing to personnel uneducated in their use. Fortunately, most vibration problems are caused by imbalance or misalignment, and for these cases filtered and sometimes even unfiltered data can be used effectively to identify the problem. With the general availability today of personal computers, thought should also be given to at least partially automated identification of vibration problems by “artificial intelligence” computer programs. Some specialized “stand-alone boxes” are available where at least some “expert advice” is given depending upon vibration levels at various potentially meaningful frequencies.8 In some cases these systems are hand-held, and/or generate printed records of their advice together with a brief list of reasons for the diagnosis.9 They should be used as an initial guide only. There is no substitute for testing and evaluation of a rotating machine by an experienced person with access to a variety of tools and techniques, and able to take individual machinery characteristics and operating conditions into account. In selecting a vendor for predictive maintenance hardware and software, first make the choices discussed above. From a list of vendors making this equipment, final selection can be made based on cost vs. various specific features that fit your needs: Copyright © 1994 CRC Press, LLC
Volume III
1. 2. 3.
4.
5. 6. 7.
8. 9.
47
Is the database of sufficient size that your system can be expanded to fit future needs? Does it splice into standard databases, so data can be down-loaded to general purpose software like word processors and relational database programs? A “keyword” to look for is SQL compatibility, on which many database programs are standardizing. Low level systems should at least be able to dump to ASCII or DB4 format. Is it on-screen menu-driven with pop-down menus and easily accessed “help files” available? If portable, does the data logger include at least a low-resolution screen or immediate printout to check that the data look reasonable? Is vibration printed out in the form and units preferred? Does the system provide permanent data records, preferably with a time and date stamp, and listing of the machine and its operating condition? Does it do defect analysis and give expert advice? Are there modules for preventative maintenance work scheduling, spare parts inventory tracking, replacement part purchase tracking, and comprehensive and customizable report writing? Does the initial and upgrade/maintenance cost fit within the allocated budget? If not, a local company specializing in predictive maintenance may perform it for you as a continuing service. Some equipment manufacturers provide this service. Is the system architecture “open”, i.e., not vendor proprietary, so that inexpensive offthe-shelf peripherals can be used and the system maintained even if the vendor goes out of the business?
Also of key importance are certain “comfort” factors:
1.
2. 3.
How long has the company been in business? Will support and parts be available for the system during its expected life (probably about 5 years)? Does the system appear rugged enough to stand up to a maintenance environment? Is it easy enough to use properly? Will the operating personnel have enough background and time to learn to use it reliably? What do other users comment about its reliability and ease-of-use?
PREVENTATIVE MAINTENANCE BY DESIGN AND OPERATION
Bloch5 suggests that higher first-cost will be offset by reduced operating and maintenance costs if machines are designed with more rugged rotor systems and with appropriate monitoring instrumentation. Maintenance costs will be further reduced, with justifiable increase in first-cost, if the manufacturer has sophisticated staff and advanced tools, such as experimental modal analysis, to determine quickly the cause and corrective measures for vibration problems. Regardless of first-cost/operating cost compromises it is very common to select a machine that has excess pressure and flow capacity in order to meet future needs. Operating at flows well below the manufacturer’s “design condition” can, however, lead to taxing vibrations several times greater than those for a machine sized close to actual system requirements, as illustrated in Figure 1. Any gain in future years in having avoided the purchase of a new machine will often have been lost many times over in increased maintenance costs during the intervening time.10–14 Vibration changes should also be predicted in advance of substantial machinery “re-rates” or “revamps” to modernize them or to change their performance. For example, exchanging a mechanical seal for a shaft packing can be disastrous if removal of the extra damping provided to the rotor by the packing causes a severe critical speed to blossom in the running speed range.15 Copyright © 1994 CRC Press, LLC
48
CRC Handbook of Lubrication and Tribology
FIGURE 1. Operation below best efficiency point (BEP) of machine gives flow/blade angle mismatch and increased vibration
VIBRATION MEASUREMENTS
Economics dictates that the number of probes be minimized. To be practical, they must be readily accessible for installation and replacement, and they must not interfere with functioning of existing machinery. Major specifications, such as API, VDI, ISO, ANSI, NEMA, and the Hydraulic Institute standards, suggest measurements in the vicinity of the bearings, where vibration forces between the rotating and static components are assumed to primarily react. However, often substantial load-bearing by annular seals in hydraulic machinery such as pumps (Lomakin effect, as discussed in References 16 to 19) suggests that shaft and housing vibrations at the bearings alone are not globally representative of rotor/casing relative vibrations. In some cases, neither accelerometer nor proximity probe measurement at bearings represents the vibration levels in the central portion of the machine, such as in the case of unusually stiff bearing/bearing housing combinations. The only relatively sure solution to selection of a measurement type and location that will truly represent the probability of machine damage is a rotor dynamics analysis with flexible supports.53 A simple three-degree-of-freedom analysis considering the shaft, bearing film, and bearing housing as independent springs in series can qualitatively represent machine rotor and housing vibration sensitivity to bearing and housing stiffness. In such a model, by increasing bearing housing stiffness such that it is much greater than the shaft and bearing stiffnesses, housing vibration levels can be reduced in the limit to those of the foundation. Therefore, specifications which rely on bearing housing readings uncalibrated with respect to the ratio of bearing housing to shaft/bearing stiffness are open to interpretation error. This model also shows that use of shaft vibration relative to the housing is no less subject to abuse, since in principle bearing stiffness can be increased and/or bearing housing stiffness decreased to reduce the readings of proximity probes near the bearing as low as desired. Typical vibration test locations include the following, in order of importance: 1.
Driven-end (“inboard”) and nondriven-end (“outboard”) bearing housings, in the vertical, horizontal, and axial directions for horizontal machines, and in the perpendicular-to-discharge, parallel-to-discharge, and vertical (axial) directions at the top of the machine for vertically mounted machines. This provides an indication of
Copyright © 1994 CRC Press, LLC
Volume III
2. 3.
4.
5.
6.
49
nearly any type of vibration within the machine, although if bearing housings are unusually stiff or with long flexible machine casings (like vertical pumps), problems at one end may not be transmitted strongly enough to the other end to allow an indication of even severe problems. Effort should be made to measure both ends of any machine, but particularly those with flexible casings.20 The shaft relative to the bearing housings at the inboard and outboard bearings, and if possible on each side of the motor or turbine coupling, to observe rotor imbalance and misalignment, and to provide a cross-check on rotor/stator vibration in case bearing housing stiffness accidentally masks these effects. Thick-walled sections of the machine casing, particularly at midspan, to observe casing flexure. The baseplate and/or machine foundation near the machine attachment bolts, to check for a “soft foot” due to a improper shimming, a cracked foundation, or grout voids. The floor or foundation adjacent to the machine vs. that at a distance from the machine equal to the height of the machine, to check for foundation stiffness.21 Piping near the machine flanges and at several points in the piping assembly, to check for piping-induced vibration, particularly due to piping acoustics or piping structural resonance.
VIBRATION PROBE CHOICE: FREQUENCY RANGE, COST, CONVENIENCE
The frequency range over which vibrations should be checked should span the normal range of significant excitation force frequencies within the machine, typically several Hz up to about twice the blade pass (or, in an electric motor, slot pass) frequency. Probes and transducers which are useful in obtaining vibration data are: 1.
2.
3.
Accelerometers. Fragile but accurate in measuring acceleration over the full frequency range. Velocity probes. Rugged but accurate at directly measuring vibration velocity over a relatively narrow range of about 10 to 200 Hz. Proximity probes. Noncontacting and difficult to mount unless threaded into the bearing housing. Accurate at measuring shaft-to-housing displacement over a frequency range of DC to 500 Hz.
VIBRATION DATA ANALYSIS EQUIPMENT
The output of each of the above probes can be fed into oscilloscopes or fast Fourier transform (FFT) spectrum analyzers to obtain vibration vs. time or vibration vs. frequency traces, respectively. Frequency-based FFT data plotting has replaced time-based vibration plotting as the “workhorse” in modern vibration testing. Unfortunately, exclusive use of vibration vs. frequency data loses the special usefulness of time traces in observing certain problems. It is best to view vibration data in both the frequency and time domains.2 Oscilloscopes can be two channel such that shaft motion within clearances can be plotted as an “orbit”, the size and shape of which can identify specific problems. One channel plots vertical vibration and the other horizontal. “Wiggles” in such orbits may suggest rubs, and occasional but severe transient bursts in the linear vibration vs. time traces might suggest cavitation or surge. New ways of using such time-based data to provide more physical insight are available,23 such as the “Prony series”.22 Time-related methods where frequency data are converted to time (rather than the other way around as is more common) are also useful in understanding amplitude and frequency Copyright © 1994 CRC Press, LLC
50
CRC Handbook of Lubrication and Tribology
modulated phenomena, such as occur in gear boxes24 and in turbomachinery during rotating stall.13 The best known of these techniques is the “Cepstrum”.25 FFT analyzers can also be two channel. The feedback from a calibrated shaker or impact excitation hammer can be sent to one channel, and the resulting vibration response sent from a probe to the other channel, in order to determine the inherent frequency response characteristics of the machine (natural frequencies, damping, and vibration mode shapes). Until the advent of the FFT, vibration was presented either as a time-based oscilloscope plot, or an overall value or filtered (i.e., only vibration close to a given frequency) value on a meter. FFT frequency spectra generally are much easier to interpret than is vibration vs. time. Time traces are “noisy” and confusing when due to several strong frequency components or due to “broadband” (i.e., broad in frequency range rather than at a specific or “narrowband” frequency) phenomena such as fluid excitations. In the frequency domain, amplitude of various frequency “components” can be related to physical machinery parts or events inside the machine, such as the rate at which an impeller blade passes a diffuser vane.26 Options to be aware of in vibration FFT analysis are: 1.
2.
3.
4.
5.
All FFTs are based on the digital Fourier transform, as presented by Cooley and Tukey.27 They translate analog vibration vs. time signals into digital (i.e., stepped in frequency) vibration vs. frequency spectra. The number of steps in a spectrum determines its resolution and is a function of the quality of an FFT analyzer. The steps are typically listed in terms of “lines” of horizontal screen resolution from zero to maximum frequency. A high quality FFT analyzer has between 800 and 1600 lines. Less than 400 lines resolution is insufficient for general vibration analysis. Mathematical weighting functions or “windowing” is generally available. Hanning windows look like the upper half of a sine wave, decaying to zero at each end. They eliminate apparent vibration vs. time discontinuities when cross-multiplied with a data signal whose sine waves are chopped off at the beginning and/or ending of the datataking span, which is typical. Without the Hanning window, narrow-band signals in the vibration vs. frequency spectrum do not form thin lines, but exhibit false “skirts”, called “leakage”. In impact modal testing, discussed below, transient or exponentially decaying windows can be used to cross-multiply the excitation and response data to de-emphasize the noise or natural excitation response which dominate the signal after the machine has “rung-down” from the impact. Anti-aliasing filters avoid the misinterpretation, as lower frequency waveforms, of very high frequency data beyond the 2.56x maximum FFT frequency “Nyquist sampling” limit. Nearly total elimination of aliasing requires expensive analog filters, and is one reason for the higher price of quality FFT equipment vs. lower cost units, particularly hand-held ones Various sample-averaging techniques are available to process multiple data samples and thereby emphasize steady state vibrations. Frequency averaging each rms vibration value at each digitized frequency, sample-by-sample, results in a true rms average over the time of the sampling. Synchronous or time averaging can be used to emphasize the portion of the vibration spectrum which directly correlates with (and therefore is probably caused by, or at least has a common cause with) the signal used for synchronization. Peak averaging is useful to check the maximum vibration value at each digitized frequency. Critical speeds are evident, for example, during a run-up or coast-down in machine speed. “Zooming” allows the analyzer to “blow up” a certain portion of the frequency spectrum
Copyright © 1994 CRC Press, LLC
Volume III
6.
51
Coherence, correlation, and signal-to-noise ratio options allow the user to check how well one signal correlates with another, for evaluating cause and effect.
In cases when the extra sophistication of oscilloscopes and FFT analyzers is not needed, or is too costly or inconvenient, use of vibration meters may be appropriate. “Hand-held” inexpensive sensor/meter packages consist of a small battery-operated box with a stick-like probe built into it or attached by a short cord. The probe is usually a velocity sensor probe, which works by sensing the current flow due to the motion between two internal electromagnets, one rigidly attached and one flexibly attached to the casing. The box is either (1) a meter which typically has a tunable filter to optionally read only the vibration in a certain frequency range (with this option off, the meter or digital display represents all frequencies), or (2) a low-resolution spectrum analyzer able to plot approximate vibration level vs. frequency. Many hand-held meters also include simple integration and differentiation circuits, which allow the original form output by the probe (e.g., velocity) to be translated into displacement or acceleration. Some hand-held units have a useful feature called spike energy analysis.28 This involves logging of the square of the vibration acceleration, emphasizing very high frequencies (on the order of 10 kHz) where rolling element bearing defects tend to exhibit high accelerations, while most other phenomena do not. However, beware that spike energy values are also sensitive to minor valve cavitation, piping acoustics, and load variations on the bearing due to changes in machine operating conditions. Trends of spike energy taken on the bearing housings at identical loads can help identify bearing problems, but absolute values do not necessarily have much significance in the absence of a trend. The disadvantages of hand-held meters are that: 1.
2. 3.
Vibration vs. frequency information is “fuzzy”. It is difficult to distinguish hydraulic forces and mechanical resonances from excitation harmonics of running speed due to imbalance and misalignment. Only one probe at a time can be used, and the powerful “modal” testing discussed below cannot be performed. Some two-channel capability is restored in systems which allow combination of the probe with a strobe light to check machine rpm and probe/strobe phasing. Velocity probes lose accuracy rapidly outside of the range 600 to 10,000 cpm.
Therefore, this type of equipment is useful for determining that a vibration problem exists and for approximately determining its frequency. It cannot give much diagnostic information about why the problem exists, and usually provides only a fuzzy indication of how to solve it. Another form of equipment is represented by data recorders. These can be either digital, such as the common portable “data loggers”, or analog. If the former, the analog signal from a vibration sensor is promptly digitized for introduction into a computer subroutine for data storage or immediate vibration analysis. Digital data storage is relatively limited, and data can only be resolved to about half the digitization rate set during data acquisition. If analog data are stored on magnetic tape, they can be resolved later up to the resolution of the tape recorder. Direct AM tape recorders have accurate signal recording from 50 Hz to as high as 60 kHz, while FM tape recorders can record faithfully from DC to about 1 kHz. Both types of analog recorders suffer from a mediocre dynamic range (the largest vs. smallest signal that the recorder can deal with at one amplitude setting) of only about 50 dB, while modern digital recorders have 120 dB or more. To be sure of getting sufficient signal resolution for complete data analysis, vibration data should be taken over a range of 60 dB or more. Copyright © 1994 CRC Press, LLC
52
CRC Handbook of Lubrication and Tribology
Other potential pitfalls in using recorders are that bad data may be taken without causing notice, and head alignment from one recorder channel to another can cause false phase shifts between channels that were recorded simultaneously, causing erroneous interpretations based on the apparent time lag.
VIBRATION DATA PRESENTATION ALTERNATIVES
1. 2.
3. 4.
Vibration test data are generally plotted in four different forms:
Cartesian plots of vibration amplitude vs. frequency (“signature plots” or “spectra”). Sometimes this may be a “Bode plot” formed by combining this with a plot of the phase “lag” angle between the exciting force and responding motion vs. frequency. Polar plots of vibration amplitude vs. excitation/response phase angle for all tested frequencies (“Nyquist” plot), where vibration level is the radial vector, and phase is the angle. As frequency is increased gradually, a line is formed on the polar plot because amplitude and phase angle both change. Since natural frequencies plot as near circles, closely spaced modes are more easily identified and separated using the Nyquist plot rather than the Bode plot. Cartesian plots of vibration amplitude vs. time, similar to a typical oscilloscope trace. Polar “orbit” plots of vibration vs. time in a plane perpendicular to the shaft axis.
Amplitude scales may be linear or dB. dB (i.e., base 10 logarithmic) scales are often used to improve the resolution of natural frequency peaks and to display natural frequency broadband peaks excited at low levels by turbulence and other “white noise” (i.e., a roughly even level vs. frequency) in the machine, as discussed by Marscher.6 Vibration measurements are made in terms of displacement, velocity, or acceleration. Displacement refers to the actual extent of the vibratory motion as it takes up, for example, a certain fraction of a running clearance. Velocity is the maximum rate of change of this displacement, the vibrating component’s peak speed as it speeds up and slows down due to the vibration. Acceleration, similarly, is the rate of change of the velocity: the maximum “G” level a person would feel if he were riding on a structure motionless except for its vibration. All three measurements directly relate back to the same motion, with velocity equal to displacement times the vibration frequency (times a constant to convert units), and with acceleration equal to displacement times frequency squared (times another units constant). Therefore, velocity emphasizes vibration at higher frequencies more than displacement does, while acceleration emphasizes high frequency vibration even further, to the point that the running speed contribution to unfiltered acceleration is often minimal. Velocity and acceleration have the advantage of being more sensitive than displacement to problem phenomena at high frequencies (like ball bearing defects and pump cavitation), but have the disadvantage of emphasizing meaningless structure-borne acoustic waves that reflect piping noise, for instance, in vibration on the order of millionths of an inch (a micron) within the machine. Velocity and acceleration also have the disadvantage that they de-emphasize potentially catastrophic subsynchronous phenomena such as rotor dynamic instability and severe rotating stall in impellers or diffusers. A typical relationship between acceptable vibration displacement, as based on machine wear and running speed is shown in Figure 2. The reason the plot takes on this shape is that once machines get past a certain running speed, commonly about 1800 rpm, machine rotor diameter is inversely proportional to speed for manageable rotor stress. In order to maintain reasonable flow efficiency as the rotor shrinks, running clearances must shrink as well. This provides less room for components to vibrate relative to each other.
Copyright © 1994 CRC Press, LLC
Volume III
53
FIGURE 2. Dependence of wear on vibration at various running speeds and frequencies.
Experience indicates that allowable machinery vibration is inversely proportional to running speed. Equivalently, vibration velocity at 1x running speed should not exceed a certain number, generally agreed to be order of 0.3 inches per second (7.6 mm/s) peak, where this number is reflective of the structural material capabilities combined with a sensible compromise of manageable clearance vs. leakage rate. This relationship has been redefined by many standards organizations to relate acceptable vibration to excitation frequency, which does not necessarily follow the same relationship. This is currently an area of unresolved controversy, with API 610, 7th edition29 and HI Standards, 15th edition30 changing the scale of their vibration-specification plots from frequency to running speed, the basis for most historical data upon which the acceptance lines on the plots were based. Some compressor and turbine standards now state, on the basis of field experience, that allowable vibration amplitude is inversely proportional, not to frequency, but to the square root of frequency, following a “middle road” as shown by the middle dotted line on the right of Figure 2. The complex issues involved in this subject are provided independently by Maxwell31 and Marscher.32 Historical tracking of vibration specification development would review references by Rathbone,33 Blake,34 Baxter and Bernhard,35 and the various revisions of the standards organizations, VDI and ISO in Europe, and API, ANSI, and HI in North America. Vibration data are quoted as “rms”, “peak” (or “zero-to-peak”), or “peak-to-peak”. “Room mean square” (rms) is the time-averaged value of the vibration, where the direction of the vibration at any instant in time (positive or negative) is ignored. Peak stands for the vibration amplitude in terms of the maximum absolute value vs. the time-averaged eccentricity, and is generally applied only to vibration velocity or acceleration. Peak-to-peak is the difference between the maximum vibration motion in one direction vs. the maximum motion in the opposite direction, and is generally applied only to vibration displacement. Most FFT analyzers plot rms vibration. To convert rms values at a specific frequency to peak, multiply by 1.41, and to convert to peak-to-peak, multiply by 2.83. Peaks determined from rms values in this way are time-averaged values of the actual peaks, which may vary Copyright © 1994 CRC Press, LLC
54
CRC Handbook of Lubrication and Tribology
considerably from cycle-to-cycle. The actual maximum peak vibration as determined by observing an oscilloscope trace for a reasonable time, for example, is called “true peak”. In machines prone to wear-out by transient events, it probably correlates better with failure rate than does an extrapolation of the spectrum analyzer-plotted rms value, which is calculated and not a true peak. The rotational speeds at which vibration reaches a peak and is severe enough to cause reliability problems are called “critical speeds”. Critical speeds are commonly determined by “cascade plotting” vibration amplitude vs. frequency spectra during acceleration or deceleration between the static and maximum speed operation of the machine. In such graphs, individual vibration vs. frequency spectra at progressively higher running speeds are plotted directly behind each other so that 1x running speed, 2x running speed, etc. form straight line “backbones” in a diagram that gives the impression of a “waterfall” or a “cascade”. Critical speeds are evident on the running speed harmonic “backbones” where the spectra locally spike up to high values. A good example of cascade plotting is provided by Kirk.36 Sometimes the “backbones” or cascades of running speed harmonics are emphasized further by filtering out vibration between harmonics with “tracking filters”. These use a running speed signal from a tachometer or “keyphasor” as a reference for synchronous averaging of the harmonics.
EXPERIMENTAL MODAL ANALYSIS Experimental modal analysis (EMA) is a test method in which a known force (often constant at all frequencies within the test range) is put into a machine, and the vibration response exclusively due to this force is observed and analyzed. Natural frequencies of the combined casing, piping, and supporting structure can be obtained, and if special data collection procedures are used, EMA can also determine the rotor natural frequencies at the machine operating conditions. Separately, the frequencies of strong excitation forces within the machine can be determined by comparing the vibration vs. frequency spectrum of the machine’s EMA artificial force response to the signature analysis spectrum of the machine’s response to the naturally occurring forces from within the machine and from its attached system and environment. The main tools required to do EMA are a two-channel FFT frequency analyzer, a microcomputer with special software, a set of vibration response probes, and an impact hammer. During an EMA test, the signal from the hammer input force accelerometer is sent to one channel of the spectrum analyzer, and the signal from the vibration response probe is sent to the second channel. Dividing, at each frequency, the second channel by the first channel gives the “frequency response function” (FRF) of the machine and its attached system. The peaks of the FRF are the noncritically damped natural frequencies, and the width and height of the peaks indicate the damping of each natural frequency, and how sensitive vibration at the test location is to forces which occur in the vicinity of the hammer impact at frequencies near a given natural frequency. The details of classical modal analysis and its application are given by Ewins.37 A special modal analysis suited to testing operating machine structures and rotors, while the machine is operated at problem conditions, is given by Marscher.38,39
MONITORING PARAMETERS NOT RELATED TO VIBRATION The following are examples of other rotating machine parameters that should be checked as part of a monitoring or problem solving process. Copyright © 1994 CRC Press, LLC
Volume III
1. 2.
3.
4.
5.
6.
55
Lubricant temperature after it is discharged from the bearings. While sump temperature generally varies weakly with peak lubricant temperature, variations of ± 25% in the differential temperature between the sump and the environment are significant. Bearing race or shell temperature gives a more sensitive and faster-reacting indicator of bearing load and proper functioning of the bearing. A resistance temperature detector (RTD) or thermocouple is embedded in the metal for this measurement, and should be placed as close to the load-carrying part of the bearing as practical. Temperature changes of ±10°F (6°C) can be used as an indicator of distress. Lubricant sampling involves analysis of oil for composition, contamination, and degradation. The oil analysis may be as sophisticated as determination of constituency by spectroscopic analysis, to evaluation of oil color and clarity by visual inspection, so that oxidation rate and contamination level can be estimated. Viscosity and acidity checks are also useful indicators of oil deterioration, and therefore of unusually high loads at bearings or leakage at seals. Ferrous metallic contaminants in the oil can be sorted by ferrography. DR (direct reading) ferrography is the simplest, involving the gathering of a small but representative oil sample, diluting it, and drawing it by capillary action into a thin tube. The ratio of large particles coagulated at the tube inlet vs. small particles drawn into the tube center can reflect the amount of wear in the system. A more accurate, but more difficult and expensive form is analytical ferrography, in which magnets strip magnetic particles from the lubricant flowing over an inclined microscope slide, which is then viewed in a Dichromatic microscope.40 Particle type, shape, and size are then interpreted by experts to determine the location and rate of machine wear. Spectrographic analysis evaluates the absorption or emission spectra of all types of particles, ferrous and otherwise, gathered from flowing oil samples. Spectrophotometry is a variation that monitors nonmetallic compounds in lubricant oil by infrared absorption. Indiscriminant particles can be detected through the blockage of light by lubricant flowing between a light source and a photocell or can be gathered from filtration residue and analyzed under a microscope. The location and rate of wear can often be deduced from the quantity, size, shape, and composition of such particles. Improved predictions of machine condition are possible if the particles are monitored on a continuous basis. In lubricant analysis, it is important that lubricant samples are representative of the oil currently flowing through the bearings. This is unlikely to be true of small samples of oil drained from a corner of the sump, unfortunately a common practice. Visual inspection, in its simplest form, merely requires careful viewing of the machine for leakage of process fluid, cooling water, or lubricant, or for evidence of looseness, wear, or cracking. Thermography uses an infrared scanner to check for local heat generation, or for the possibility that a crack or excessive leakage is preventing normal levels of heat conduction.41 Fiber optic borescopes enable looking inside flow paths, sometimes while a machine is running. This technique has been used to investigate compressor and steam turbine blade erosion and pump vane cavitation damage. Noise can be monitored using meters or oscilloscopes and spectrum analyzers. Impeller and diffuser vane resonances often show up more strongly in narrowband noise traces than they do in bearing housing or shaft proximity readings. Significant responses in noise monitoring include large (a factor of two or more) increases in already high narrowband components, and substantial peaks in the broadband “noise floor” which generally indicate machine or piping resonances, or important fluid excitations such as stall fields and acoustic resonances.
Copyright © 1994 CRC Press, LLC
56
7.
8.
CRC Handbook of Lubrication and Tribology
A major difficulty in using noise as a condition-monitoring tool is that it can change significantly with position of the microphone and is subject to acoustic wave reflection within the room in which the machine is installed. Also, relating strong noise at certain frequencies to the component causing it can be difficult unless special “triangularization” or “tomagraphic” techniques are used.42 However, noise is the best indicator for phenomena such as pump or valve cavitation, which give strong noise signatures in frequency ranges of 20 to 80 kHz. Acoustic waves directed at operating turbine blading, which are then monitored for their reaction, have been found useful in detecting blade cracks.43 Machinery performance observation is probably the most significant of all nonvibration condition monitoring. As outlined in greater detail below, performance maps can be drawn and watched for changes in discharge pressure, flow, horsepower, speed, and net efficiency. When such maps are constructed or performance points are spot-checked, include possible effects of inlet pressure and temperature. For example, a drop in pump suction pressure below its required net positive suction head (“NPSHR”) will cause significant cavitation in the inlet, leading to a drop in pump efficiency and head. This could be mistaken for some other problem in the pump, such as excessive wear. Analogously to fluid machines, motor operating parameters should be monitored. This includes voltage, current, and phase-to-phase output. These parameters should be checked for smoothness, time-consistency, and consistency of one phase circuit to another. In addition, a frequency spectrum for each of these parameters should be evaluated vs. the criteria listed in the problem-solving section later in this chapter. Certain special purpose probes are also useful for measuring wear and erosion internal to rotating machinery and associated components such as valves. “Surface layer activation” through ion bombardment is one of these techniques. Radioactivity levels in the fluid, or residual levels at the activation site reflect the amount of lost material or remaining material, respectively, in wear-prone and erosion-prone areas. Another technique is acoustic or laser measurement of material thickness or clearance gaps at key wear locations, such as journal bearings and seals.
USE OF PERFORMANCE MAPPING
In judging acceptability of a new machine, it is advisable to plot vibration vs. running speed on some type of performance map. For example, a compressor can be mapped in terms of its discharge pressure vs. flow, and a pump by discharge head vs. “capacity” (discharge flow). In the case of compressors and gas turbines, it is important to compensate for inlet temperature, and often maps of “referred” parameters are made which represent pressure vs. flow for broad ranges of inlet conditions. Output of a fluid machine is determined by the intersection on such performance maps of a line representing pressure vs. flow for given speed and inlet conditions, and another line representing the pressure vs. flow “system resistance” characteristic curve of the system being discharged into. It is the usual practice to establish the system resistance curve first, and then superimpose on it plots of machine performance. If two or more machines may discharge into the system in parallel with each other (a common practice), the resulting map gets cluttered by families of operating lines representing summed output to the system if one machine discharges, if two do, if three do, etc. This confuses plot interpretation, especially if the several machines do not all operate at the same speed, as shown in Figure 3. Figure 4 shows an alternate approach found useful by the author, which collapses all machine characteristics into a single set of lines. Pressure seen at the discharge of all of the Copyright © 1994 CRC Press, LLC
Volume III
FIGURE 3 Performance mapping from the system point of view.
FIGURE 4. Performance mapping from the machine point of view.
Copyright © 1994 CRC Press, LLC
57
58
CRC Handbook of Lubrication and Tribology
parallel machines is plotted vs. flow from only the machine being tested. The system resistance becomes a roughly trapezoidal area, as shown, rather than a single line. The lines bounding this area are easily established, based on where the system resistance curve fixes the pressure with no other machines running vs. all other machines running at full speed, unthrottled. Operation of other machines in parallel (whatever their speed and condition) is spotted on the map exactly as they affect the tested machine in service, i.e., as an increase in the effective system resistance, because the tested machine must compete with flow from the other machines through the same discharge system. On such “solitary performance” maps, condition monitoring contour lines can be drawn representing bearing temperature changes or vibration amplitude, for instance, regardless of the speed and flow of parallel machines. When vibration is plotted in this manner, a “contour map” of vibration vs. running condition quickly makes evident “hot spots” of high vibration which can be avoided to maximize machine time to the next overhaul. The gradual spreading with time of such contours to take up increasing area on the map can be used as a direct indicator of machine health. Some phenomena which plot as clear contours on solitary performance maps are rotating stall, compressor surge lines, pump suction recirculation, pump cavitation, and acoustic resonances. Efficiency contours can also be placed on solitary performance maps. Generally, combined thermodynamic and mechanical efficiency is easily established, and its decrease provides an excellent indicator of machine deterioration. The basic formula for efficiency of a compressor or pump is that efficiency equals a constant x average fluid density x system discharge head x machine discharge flow/horsepower consumption.
IDENTIFYING VIBRATION PROBLEMS
Probably 90% of all machine vibration problems can be solved by careful balancing, alignment of the coupling when the system is at its rated conditions (especially if it is hot), and running the machine within the bounds of its specified operational limits (such as minimum and maximum flow constraints). Remaining machine vibration problems are generally due to resonance of a system natural frequency with one of the excitation forces common to all machines, such as residual imbalance. In the case of resonance, simple trial-and-error field fixes, such as addition of clamped weights or brace-stiffening of the apparently offending portions of the structure, can often be effective. However, if such “fixes” only shift the problem to a slightly different frequency and cannot be made effective within a reasonable period of time (on the order of hours, not weeks), it is important to answer the following questions: 1. 2. 3.
4.
5.
Are the excitation forces within normally acceptable limits? What are the natural frequencies of the rotating and stationary parts of the machine and attached structures when they are assembled as a system? What are the excitation force frequencies in the actual installation, and do they coincide with any of the assembled system natural frequencies? Is the vibration increasing unstably relative to the apparent forces in the system, such mat the vibrating component seems to be exciting itself, following initial perturbation? If there is a vibration problem due to excessive excitation force, a resonance, or an instability, how can the machine stationary or rotating structure or the attached piping and foundation system be changed most easily to solve the problem?
In new machines, high vibration levels are generally caused by other factors than shop balance and cold alignment if they are within the levels required by generally accepted specifications: imbalance in ounce-inches is less than 4 rotor weight (lb)/speed (rpm), and Copyright © 1994 CRC Press, LLC
Volume III
59
coupling alignment is within 2 mils (52 µm) total indicator reading. In actual use, imbalance and misalignment may be itself a symptom of dominate vibration due to uneven wear, shaft bow from residual stresses, and alignment offset caused by fluid side-loads, piping loads, and thermal growth differential between the driving and driven machines. In general, imbalance-induced vibration increases with the square of running speed, and misalignment-induced vibration increases little with running speed. Fluid-induced vibration in theory increases with the discharge pressure and therefore with speed squared, but this relationship gets complicated by discharge pressure following “flat” system resistance curves rather than the machine’s “square law” characteristic, and by dramatic increases in fluidinduced vibration at part-load due to flow angle/blade angle mismatch.
EXCITATION FORCE, FREQUENCY, AND EXCITATION ORDERS
Usually when a component of a turbomachine vibrates, the amplitude is close to the ratio of a time-varying force, acting on the component, divided by the stiffness of the component and its supports. If the peak value of the time-varying force is applied slowly, the component deflects the same amount as it vibrates. Exceptions include resonance and dynamic instability, discussed below. In any event, the vibration-driving mechanism is known as the “excitation force”, and in machinery it usually operates at a characteristic “forcing frequency” which is a multiple, or “harmonic”, of running speed, such as imbalance acting at lx running speed. Common excitation forces are imbalance and misalignment. One of the strongest and most difficult-to-minimize excitation forces is the vane or blade passing frequency, equal to the number of impeller blades times running speed. Since the blade pass pressure vs. angle distribution is cyclic but not sinusoidal, reasonably strong Fourier coefficients exist at 2, 3, and possibly 4 times the blade pass frequency, and these are strong forcing functions themselves. The operating point of the machine is important because at off-design points the inlet angle of incoming flow is not well matched to the blade angle. This situation can become so severe that stalling of the blades occurs at the entrance and/or exit. This dramatically increases vane pass vibration and can also cause subsynchronous (less than running speed) excitation frequencies equal to the frequency of rotating stall cells in the impeller or stationary guide vanes, as discussed by Pampreen.13 Figure 1 gives a typical example of the degree to which operating at off-design influences overall vibration. Makay and Barrett12 introduced the concepts of: (1) axial pressure pulsations on the surfaces of the impeller shrouds due to large minimum clearance between the rotating shrouds and stationary casing walls; and (2) sometimes dramatic increases in impeller vane pass pulsations due to impeller vane vs. diffuser or volute vane gap of less than 4 to 6% of the impeller diameter. Although these concepts originally referred to centrifugal pumps, they apply to all radial and mixed-flow machines.
NATURAL FREQUENCY AND RESONANCE
Natural frequencies are vitally important because vibration problems normally occur only in two circumstances:
1.
2.
The exciting force, usually imbalance or misalignment, is very large. This is why users carefully balance and align rotating machinery. The frequency of the exciting force is close to a natural frequency, allowing the force and the vibration due to that force to fully synchronize with each other. If the excitation force frequency and the natural frequency are within a few percent of each other, vibration energy from the last “hit” of the force is fully stored up when the next hit
Copyright © 1994 CRC Press, LLC
60
CRC Handbook of Lubrication and Tribology
FIGURE 5. Vibration versus frequency “magnification factor” Q = P/S.
FIGURE 6. Illustration of the phase angle by which displacement lags the force in the 360° vibration cycle
takes place. The vibration in the next cycle will then include the effects of both hits and be higher than it would be for one hit alone. This resonant vibration in machines can commonly build to between ten and thirty times the amplitude which would occur if the same exciting force were applied statically, as shown for a natural frequency in Figure 5.
Along with vibration amplification, another key parameter associated with resonance is phase angle. Its use in relating the occurrence of an excitation force to the timing of the vibration displacement which occurs in response to it is shown in Figure 6. Figure 7 shows the manner in which phase angle starts at zero, goes through a 90° change at resonance, and continues to 180° total change at frequencies well beyond resonance. For a detailed discussion of resonant phase change and the physical reasons behind it, refer to Marscher.38
ROTOR DYNAMIC INSTABILITY AND SUB SYNCHRONOUS WHIRL
Rotor dynamic instability refers to phenomena whereby the rotor and its system of reactive support forces become self-excited, leading to potentially catastrophic vibration levels even if the original excitation forces are quite low. Shaft part-speed whirl is a forced response at a frequency below the first noncritically damped shaft bending natural frequency. The most common cause of whirl is fluid rotation Copyright © 1994 CRC Press, LLC
Volume III
61
FIGURE 7. How phase angle θ relates to natural frequency (f/fn = 1.0).
around the impeller shrouds or in journal bearing clearances. Such fluid rotation is typically about 45% of running speed: fluid is stationary at the stator wall and rotating at the rotor velocity at the rotor surface, such that a roughly half speed Couette flow distribution is established in the running clearance. The pressure distribution which drives this whirl is generally skewed such that the cross-coupled component (i.e., the component of film pressure perpendicular to the motion) is positive and potentially strong. If somehow clearance is decreased on one side of the gap, due to eccentricity, for example, the resulting cross-coupled force increases further and may exceed the damping force which acts vectorially opposite it. If the cross-coupled force does dominate the damping force, it acts in the direction of the shaft whirl, perpendicular to the clearance closure. If the roughly half speed whirling frequency of the cross-coupled force and minimum clearance becomes equal to a natural frequency, a 90° phase shift occurs, with the motion in response to the cross-coupling force being delayed from acting for 90° worth of rotation. By the time it acts, therefore, the cross-coupled force tends to further close the minimum gap. As the gap closes, the cross-coupled force which is inversely proportional to this gap increases further. The cycle continues until all gap is used up, with severe rotor vibration and rubbing. This self-excited instability is called shaft whip. Since shaft whip occurs at the bending natural frequency of the shaft, the vibration response frequency “locks on” to the natural frequency. Whip begins when whirl is close to half the running speed and is equal to the shaft natural frequency. Interaction occurs between these two frequencies, so the normally circular shaft orbit now shows a loop reflecting orbit pulsation every other revolution. For an example of a rotor dynamic stability problem involving shaft whip, refer to Kirk.36
AEROELASTIC INSTABILITY AND FLUTTER
In axial fluid machines, long thin blades can have a natural frequency (usually blade torsion) close to a blade passage fluid pulsation frequency, such as trailing edge vortex shedding. This could lead to a resonance of the blade, which if strong enough could fatigue the blade near its root. Even more powerful vibration, almost certain to lead to rapid fatigue, will occur if along with the resonance there is feedback between the blade vibration and the Copyright © 1994 CRC Press, LLC
62
CRC Handbook of Lubrication and Tribology
exciting fluid pulsation. This “flutter” has been observed in axial compressors, axial gas turbines, and reaction steam turbines, as discussed by Pampreen.13 Flutter characteristics are a sharp peak in the structure- or air-borne noise vs. frequency spectrum, at several kHz (the blade torsional natural frequency), which appears at some percentage above or below the design point. This vibration does not get any lower or significantly change its frequency as the operating point moves farther in the same direction from the design point due to speed change or throttling. The peak often seems to merge discontinuously with the floor of the noise spectrum, and is not narrowband but very sharp compared to other broadband peaks in the spectrum.
PARAMETRIC RESONANCE AND FRACTIONAL FREQUENCY
Certain common types of nonlinear vibration response fall into the category of parametric resonances. They can result in large vibration in spite of relatively low driving force. Typically, such resonances are caused by bearing support looseness or a rub at a bearing, seal, blade tip, or other running clearance. The symptoms are a pulsating orbit, with a large amount of vibration at exact whole fractions of running speed, such as 1/2, 1/4, etc.44
FALSE VIBRATION INDICATIONS
In performing vibration testing, beware of false indications of vibration, called “runout”. Electrical runout on the order of 1/4 mil (6 µm) is typical for shaft proximity probes based on eddy current fluctuations and is caused by variations in the electrical characteristics in the shaft surface. The electrical runout is particularly severe in chrome-plated shafts, especially if the plating was used to repair a rub. Mechanical runout is typically also about 1/4 mil (6 µm) and is caused by scratches or poor surface finish on the shaft under the probe. This changes the probe gap and is misinterpreted by the probe as vibration displacement. Both electrical and mechanical runouts are generally obvious as very sudden increases and decreases in an otherwise relatively smooth vibration orbit trace. A scratch, for example, shows up as a sharp spike in such a trace. In a frequency spectrum, runout is not nearly as obvious, although a scratch tends to show up as hyberbolic 2x at 1/2 the 1x level, 3x at 1/3, etc. Often runout is quantified by observing vibration reading when the shaft is slowly rotated by hand or during machine coast-down. Another common false vibration reading in proximity probes results from the shaft moving out of the probe’s linear range. On the time trace or shaft orbit, this causes flat spots. With accelerometers and velocity probes, a false vibration reading of high harmonic content, as well as broadband low-frequency apparent response, is rattling of the probe if a magnetic mount is used on a surface of the machine that is not extremely flat. Magnetic mounts are common because they are so convenient, but because of this problem the author does not recommend them. Accelerometer wax (providing faithful vibration transfer from 0 to 2 kHz), fast-setting dental cement, and permanently attached threaded mounts are far more reliable. Experience has shown that even hand-holding of probes firmly against test surfaces is less likely to produce erroneous results than are magnetic bases.45
INTERPRETATION OF MACHINERY WAVEFORMS AND SPECTRA
The predictive maintenance and trouble-shooting list of Table 1 can be used for interpreting many common vibrations and could be the kernal of any home-written predictive maintenance software. It is not meant to be all inclusive and is in the order of the frequency value observed, not in order of likelihood or importance to reliability. For pictorial examples of vibration frequency spectra and shaft orbits typically associated with some of these problems, see Figure 8 and Reference 54. Copyright © 1994 CRC Press, LLC
Volume III
Copyright © 1994 CRC Press, LLC
63
64
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
Volume III
FIGURE 8. Important vibration problems in turbomachinery.
Copyright © 1994 CRC Press, LLC
65
66
FIGURE 8 (continued).
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
Volume III
INTERPRETATION OF MOTOR ELECTRICAL WAVEFORMS AND SPECTRA
67
Electric motor vibration spectrum interpretation involves focus on electromagnetically significant frequencies as well as frequencies representing mechanical phenomena. As in fluid machinery, motor mechanical vibrations can occur at lx running speed due to imbalance, at 1x and 2x due to misalignment, as discussed by Wu,46 at exactly 1/2 times running speed and possibly strong harmonic response at 1x, 2x, 3x, etc., due to rubbing or loose bearing retention, and at 42 to 49% of rubbing speed due to subsynchronous rotor instability, as discussed by Campbell.47 In addition, rolling element bearing cage rotation frequency at 35 to 45% of running speed has been observed if ball bearings are worn, and other rolling element bearing frequencies are commonly evident, such as “ball passing” frequencies.48 Electromagnetic frequencies can be sensed as radiofrequency waves as well as noise and vibration, as pointed out by Jonas.43 Common electromagnetic frequencies are unbalanced magnetic pull due to uneven air gap between the motor rotating and stator components, which can show up as lx running speed due to shaft bow or nonaxisymmetry in the rotor (e.g., a broken bar), or 1x and 2x due to nonaxisymmetry in the stator, causing a constant direction side-load. Broken or cracked rotor bars also tend to emphasize 2x the slip frequency in induction motors, as well as the ±2x slip frequency sidebands around lx running speed and possibly around 2x the line frequency. The physical reason for sidebanding is amplitude modulation of the primary vibration, as illustrated by the gear box example in Figure 9. Also important is the line frequency (typically 60 Hz in North America and 50 Hz elsewhere), and two times the line frequency which is caused by counterrotating induced magnetic field components in AC motors. The two times line frequency rotates in the same direction as the rotor, and if it is excessive, it can signify rotor/stator eccentricity (due to misalignment or thermal bowing of the stator, perhaps because of shorted laminations or stator turns in a local area), or unequal coil size or phase resistance, as discussed by Maxwell.49 Problems caused by local heating and thermal bowing of the rotor or stator are sometimes recognized because the problem gets worse as the motor heats. Infrared scanning devices have been used to locate flaws in electromagnetic circuits, such as cracked rotor bars or shorted stator laminations, due to the temperature increase in such areas.50 Coil resistance, core loss, and dye penetrant checks on disassembled components also can be useful. In three-phase machines, 3x running speed harmonics and 3x and 6x line frequency harmonics may also appear if two or more of the phase circuits are flawed by an open circuit, by a short circuit, or by a cracked metal component which normally carries a significant proportion of the magnetic flux. This situation might occur if the motor has been severely overheated but is still marginally functional. Some useful tests discussed by Maxwell49 to determine if a problem is mechanical or electromagnetic is to observe beating (residual imbalance at running speed vs. line frequency, or 2x running speed vs. 2x line frequency will cause a beat at slip or 2x slip frequency), to check for gradual (mechanical) or immediate (electromagnetic) drop-off in vibration when the power to the motor is tripped, and motor noise and vibration when it is uncoupled from its mechanical load (e.g., a compressor). The latter test is not as telling as would be guessed if the problem is electromagnetic, because electromagnetic problems tend to show up strongly only when the motor current is high, in other words, when it is under load. In variable frequency drive/motor systems, other important excitation frequencies show up at harmonics of the inverter frequency, as discussed by Carbone.51 Six-step inverters (the most common) produce six torsional pulses times the line frequency, i.e., six torsional pulses per revolution in two pole motors (and six radial pulses, through the effects of torsional/ lateral coupling, particularly in flexible rotors). Because the inverter pulses are not sine waves, the Copyright © 1994 CRC Press, LLC
68
CRC Handbook of Lubrication and Tribology
FIGURE 9. Source of sidebanding in a gear box.
2, 3, and higher harmonics of 6x are also potentially strong, providing excitations at 12x, 18x, and possibly 24x running speed.
CLOSURE
Vibration analysis is a critical component of condition monitoring and predictive maintenance. Those involved in predictive maintenance are urged to include as well the machine operating condition parameters such as discharge pressure and flow, machine efficiency, bearing temperature, and lubricant condition and contamination rate. In establishing a predictive maintenance system, “keep it short and simple”. At first, include only machinery that has either been unreliable or is critically important, and track only parameters and vibration frequencies that have caused problems. As experience increases, more machines and parameters can be added, but avoid the over-eager overload that has plagued many predictive maintenance programs. Also, establish the program in such a way that measurements and maintenance decisions are cast into a timely routine, and are permanently logged in an easily accessible and easily interpreted fashion. This is best accomplished if the new predictive maintenance “system’“ blends as seamlessly as possible with previous procedures, and if the older procedures are discarded only as they are no longer needed. The new procedures should be based at least as much on machine history and “trending” (rates of change) of key parameters as they are based on absolute values. Although outside vibration standards such as API and ISO can be used as a start, as quickly as your experience allows, develop your own in-house standards, based on your own equipment, operated the way that you use it. If possible, interface with plant planning and with purchasing personnel prior to machine specification to ensure that a machine is selected from a quality manufacturer and will run as much of the time as possible near its design point. Copyright © 1994 CRC Press, LLC
Volume III
1. 2.
3.
4. 5.
6.
7.
8.
9. 10. 11.
12.
13. 14. 15.
16.
17.
18. 19.
20.
21.
22. 23.
24. 25.
26. 27.
28.
29. 30. 31.
32. 33. 34.
35.
REFERENCES
69
Mitchell, J. S. et al., Applications of spectrum analysis to onstream condition monitoring and malfunction diagnosis of process machinery, 1st Turbomachinery Symp., Texas A &M University, College Station, TX, 1972. Frarey, J. L. et al., Vibration signature analysis at Philadelphia Electric, Proc. EPRI Machine Symp., Cherry Hill, NJ, 1982. Stewart, R. M., The way ahead for machinery health monitoring as a subset of plant control, Noise Vib. Control, 16(2), Feb. 1985. Scheibel, J. R. and Colsher, R. J., Predictive maintenance for electric utilities, Sound Vib., May 1991. Bloch, H., Practical Machinery Management for Process Plants, Vol. 1 to 4, Gulf Publishing, Houston, c. 1982. Marscher, W. D., State-of-the-art vibration test technology, 6th Tech. Conf., Seals and Vib. Reliability of Centrifugal Machinery, Sumy Institute, Sumy, Ukraine, Sept. 1991. Mitchell, J. S., Bearing diagnostics: an overview, Proc. ASME Winter Annu. Meet., San Francisco, Dec. 1978, 15. Fritsch, T. et al., On line real time expert systems: present applications and future potential, Proc. EPRI Conf. on Expert System Applications for the Electric Power Industry, Boston, Sept. 1991. Piety, K. R., Practical experience using an automated diagnostic system, Sound Vib., 27, 14, Feb. 1993. Agostinelli, A., Nobles, D., and Mockridge, C. R., An experimental investigation of radial thrust in centrifugal pumps, J. Eng. Power, Trans. ASME, 82, 1960. Fraser, W. H., Centrifugal machine hydraulic performance and diagnostics, Pump Handbook, McGraw-Hill, New York, 1985. Makay, E. and Barrett, J. A., Field experience brings help to embattled pump users, Power Magazine, July 1987. Pampreen, R. C., Compressor Surge and Stall, Concepts ETI, Norwich, VT, 1993. Marscher, W. D., Wear of Pumps Chapter, ASM Handbook, Vol. 18, ASM International, 1992, 593. Jen, C.-W., Comparison of the rotor dynamic coefficients of packing and mechanical seals in centrifugal pumps, Lubr. Eng., 47(8), 616, 1991. Black, H. F., Effects of fluid-filled clearance spaces on centrifugal pump vibrations, 8th Turbomachinery Symp., Texas A &M University, College Station, TX, 1979. Allaire, P. E. et al., Dynamics of short eccentric plain seals with high axial Reynolds number, J. Spacecr. AIAA, v.15, n.6, 1978. Marscher, W. D., Analysis and test of pump “wet” critical speeds, STLE Trans., 34(3), 445, 1991. Childs, D. W., Finite length solutions for rotordynamic coefficients of turb. annular seals, ASME 82-LUB-42, 1982. Walter, T. J. et al., Detection of incipient failure in vertical pumps, EPRI Prof. Rep., No. RP2338–1, Electrical Power Research Institute, 1986. Marscher, W. D., Determination of pump rotor critical speeds during operation through use of modal analysis, Proc. ASME 1986 WAM Symp. Troubleshooting Methods and Technology, Anaheim, CA, Dec. 1986. Davies, P., A recursive approach to prony parameter estimation, J. Sound Vib., 89(4), 571, 1983. Rice, D. A., New techniques for vibration analysis, Proc. 11th Annu. Meet. Vibration Institute, St. Louis, June 9, 1987. Smith, J. D., Gears and Their Vibration, Marcel Dekker, New York, Pub, 1983. Randall, R. B., Separating excitation and structural response effects in gearboxes, IMechE Prepr., No. C305/84, 1984. Bolleter, U., Blade passage tones of centrifugal pumps, Vibrations, 4(3), Sept. 1988. Cooley, J. and Tukey, J., An algorithm for the machine calculation of complex Fourier series, Math. Comput., 19(90), 297, 1965. Pritchard, J. W., Vibration measurement and analysis, Plant Services Handbook of CMM and Predictive Maintenance, Pulman Publishing, Chicago, Oct. 1989, 68. API 610, 7th ed., American Petroleum Institute, Washington, D.C., Jan. 1986. HI Standards, 14th ed., The Hydraulic Institute, Cleveland, Jan. 1983. Maxwell, A. S., Experience with use of vibration standards, 6th Machinery Dynamics Seminar, Toronto, Canada, Sept. 1980. Marscher, W. D., The relationship between pump rotor system tribology and appropriate vibration specifications for centrifugal pumps, Proc. Inst. Mech. Eng. 3rd Eur. Congr. Fluid Machinery for the Oil and Petrochemical Industries, May 1987. Rathbone, T., Vibration tolerance, Power Plant Eng., 43, November 1939. Blake, M., New vibration standards for maintenance, Hydrocarbon Processing and Petroleum Refining, Jan. 1964. Baxter, R. L. and Berahard, D. L., Vibration tolerances for industry, ASME 67-PET-14, 1967.
Copyright © 1994 CRC Press, LLC
70 36.
37. 38.
39.
40.
41.
42. 43. 44.
45.
46.
47. 48.
49. 50.
51. 52. 53.
54.
CRC Handbook of Lubrication and Tribology Kirk, R. G., Evaluation of aerodynamic instability mechanisms for centr. compressors, ASME 85-DET-147, 1985. Ewins, D. J., Modal Testing: Theory and Practice, Research Studies Press, Wiley, New York, c. 1984. Marscher, W. D., Structural design and analysts of modern turbomachinery systems, Sawyer’s Gas Turbine Engineering Handbook, 3rd ed., vol. 1, Turbomachinery Publications, Norwalk, CT, 1985. Marscher, W. D., How to use impact testing to solve pump vibration problems, EPRI Power Plant Pumps Symp., Tampa, FL, June 1991. Kondos, J., Ferrographic particle analysis for machine condition monitoring, Plant Services Handbook of CMM and Predictive Maintenance, Pulman Publishing, Chicago, 1989, 82. Dresser, D., Thermography: temperature measurement detects problems, Plant Services Handbook of CMM and Predictive Maintenance, Pulman Publishing, Chicago, 1989, 86. Rasmussen, G. et al., Gated analysis of time varying signals, ASME Pap., No. 86-WA/NCA-21, Dec. 1986. Jonas, O., Diagnostic monitoring—an overview, Power Magazine, McGraw-Hill, p. 61, Jan. 1992. Ehrich, F. F., Spontaneous sidebanding in high speed rotordynamics, ASME Trans. J. Vibr. Acoustics, 114, 498, 1992. Bowers, S. V. et al., Real world mounting of accelerometers for machinery monitoring, Sound Vibr., 25, 18, Feb. 1991. Wu, J. J., Removing the mystique from mechanical drive rotating analysis, IEEE Pet. Chem. Ind. Conf, 27th Annu. Meet., Houston, TX, Sept. 1980, 59. Campbell, W. R., Diagnosing alternating current electric motor problems, Vibrations, 1(3), 12, 1985. Schlitz, R. L., Forcing frequency identification of rolling element bearings, Sound Vibr., 24, 14, 16–19, May 1990. Maxwell, J. H., Signature analysis of motors, Proc. Vibr. Instit., Houston, TX, April 1983, 39. Mondy, R. E. et al., Machine modifications solve complex vibration problems, Power, Feb. 1985 and March 1985. Carbone, H. M., Sidestepping traps in AC drive selection, Machine Design, 48, 167, Feb 12, 1987. ISO 2372 Mechanical Vibration of machines International Standards Organization, Versailles, 1974. Nicholas, J. C., The effect of bearing support flexibility on critical speed prediction, ASLE 85-AM-2E-1, 1985. Ehrich, F. F., Handbook of Rotordynamics, McGraw-Hill, New York, 1992.
Copyright © 1994 CRC Press, LLC
FILTRATION
William N. Needelman
INTRODUCTION
Filters have one principal purpose, to protect components from contamination. Mechanical components requiring protection include rotating and sliding elements, flow passages, nozzles, and heat exchange surfaces. The host fluid, providing lubrication and perhaps hydraulic power transfer, is also a component requiring protection from the contaminants it contains. Contaminants are operationally defined as foreign materials in the host fluid capable of degrading the performance of one or more components of a system. As inventoried in Table 1, they may be conveniently divided into two broad classes, particulate and chemical.
CHEMICAL CONTAMINATION
Chemicals may be immiscible or dissolved to some extent in the host fluid. They include tramp fluids (including oils in water-based fluids), solvents, and fluid breakdown products, especially acids. Water and acids are closely associated with corrosion. Acids are usually the end-product of oil oxidation reactions, which are themselves catalyzed by contaminants. Water, the most common chemical contaminant (in oil-based systems), may be present in sufficient quantity to saturate the oil. Above about 0.1 to 0.2 wt% (1000 to 2000 ppm) concentration commonly forms a separate discontinuous phase of free water, either emulsified into droplets or water collected in bulk.
Oil Oxidation The primary mechanism of fluid breakdown is oxidation of the oil. End products of oil oxidation include acids, typically measured as total acid number (TAN), and insoluble resins designated by descriptive terms such as varnishes, gums, and gels. Oil oxidation proceeds through a series of intermediate reactions. In order to impede these reactions, oils are formulated with antioxidants to break the oxidative chain reactions.1 These competing mechanisms result in an initial induction period during which the antioxidant additive is slowly depleted, followed by rapid escalation of oxidation and accumulation of oil oxidation products. Oil oxidation is accelerated both by higher temperature and by fluid contamination. As shown in Figure 1, dry oil not in contact with fresh metal surface is significantly more stable— it has a significantly longer induction period—than either water-contaminated oil or oil in contact with metal surfaces. The combination of water and metal is the worst case. In operating systems, water either enters into the system from the environment or via internal leaks from adjacent systems. Catalytic metallic surfaces are produced during wear processes. Fine filtration has been found to increase oil life by inhibiting oil oxidation. Decreasing the overall wear rate through filtration leads to reduced production of catalytic fresh metal wear debris. In addition, wear debris retained in the filter cannot travel to hot zones in the machinery where oxidative reactions proceed most swiftly. One significant example of oil life extension through filtration is found in a gas turbine engine application. Although U.S. Army typical oil change intervals for this type of system are every 200 hours, one widely deployed engine equipped with 3-µm filtration requires no oil changes between overhaul periods of 1500 to 2000 hours. In another example, the acid accumulation in a city transit bus transmission operating for over 100,000 miles with 6-µm 0-8493-3903-0/94/$0.00 + $0.50 © 1994 by CRC Press, Inc.
Copyright © 1994 CRC Press, LLC
71
72
CRC Handbook of Lubrication and Tribology
FIGURE 1. Catalytic effect of contaminants on oil oxidation (modified D943 method).Metals and water can act as catalysts to increase the oxidation rate of an oil.2
Copyright © 1994 CRC Press, LLC
Volume III
73
FIGURE 2. Fluid purifier. Chemicals such as water and solvents removed by using a spinning disc to generate a fine mist of oil from which volatiles rapidly diffuse into a low pressure air stream.
Filters was equivalent to otherwise comparable transmission operating for less than 25,000 miles with 40-µm filters.
Chemical Contamination Removal Removal of bulk quantities of immiscible fluids may be carried out in a settling tank. Residence time to attain separation depends on density differences between contaminants and host fluid as well as host fluid viscosity. For lubration and hydraulic systems, excessive residence times lead to the frequent practice of accelerating the process by imposing a centrifugal force on the fluid, which creates the equivalent of an elevated gravitational force. Vortex generators and centrifuges are two common devices for bulk contaminant removal. Coalescers also find use for removing immiscible fluids, especially water from oil. These phase separators do not permit the removal of soluble chemical contaminants nor the removal of common chemicals such as water when present below saturation levels. Volatile chemicals can be removed from oils by forced evaporation process, where they are extracted form the oil by a low pressure gaseous carrier stream. An example of a portable device capable of removing water and other volatile chemicals to well below saturation levels is shown in Figure 2.
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
74
FIGURE 3. Representative particle contamination as function of oil filtration mechanical systems. (From Needelman, W. M. and Zaretsky, E. V., Power Transm. Des., 33, 65, 1991.)
PARTICULATE CONTAMINATION AND WEAR
Contaminant particles are discrete objects ranging in size from submicron to well above 100 µm. The hardness of these particles ranges from abrasive grits such as silica sand and oxides to rather amorphous gelatinous masses. A compilation of typical particle size distributions is shown in Figure 3. Two noteworthy conclusions can be deduced from these data: (1) finer filters maintain significantly cleaner fluid systems, and (2) in any fluid system, there are several orders of magnitude more particles smaller than 5 µm than particles larger than 25 µm. Both conclusions directly affect component wear and its control. Mechanical deterioration caused by contaminant particles involves several mechanisms by which these particles interact with and degrade components. The primary forms of damage are produced during three-body contact of a particle with two opposing surfaces, as depicted in Figure 4. For surface damage to occur in this process, particle sizes must be on the order of the dynamic fluid film thickness separating the surfaces. A particle then braces against one surface and invades the opposing surface. Table 2 lists representative dynamic films for a variety of mechanical elements. Larger particles may be excluded from the contact zone, especially if a sliding contact does not enlarge during the duty cycle; alternatively, rolling elements and gearing may engulf particles larger than the dynamic film. Particles smaller than the dynamic fluid film can pass through the contact zone without seriously impacting either surface. Furthermore, in order to produce tangible damage a particle needs to be nearly as hard as or harder than at least one of the opposing surfaces. The hardness of many microscopic particles is difficult to ascertain. Whereas hardness of synthetic grits and mineral particles is tabulated, the hardness of metallic wear debris is complicated by metallurgical transformations in the contact zone (“wear hardening”) and by oxide formation. Copyright © 1994 CRC Press, LLC
75
Volume III
FIGURE 4. Three-body contact of a contaminant particle.
Depending on the relative motion between opposing surfaces, three-body contact produces either sliding contact abrasive wear or rolling contact fatigue. During sliding contact, particles plow through the surface removing material in the form of microscopic chips and leaving furrows. Loss of material results in roughening, internal leakage, and misalignment. In rolling contact, depicted in Figure 5, particles create microdents and roughened surfaces, both factors contributing to reduced fatigue life and ultimately the deeply cratered surface of fatigue spalling. Other forms of wear attributed to contaminant have been tabulated in Table 3. Material lost from component surfaces via any wear mechanism participate in the wear of other system components. The uncontrolled process in which wear debris produces more
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
76
FIGURE 5. Particle-induced rolling contact fatigue spalling.
component wear begetting even more debris, in an escalating avalanche of debris and wear, is termed the chain-reaction-of-wear. This chain-reaction can be broken with filters.
FILTER FUNDAMENTALS
Filters are devices for separating contaminants from the host fluid. The majority of filtration devices used by industry consist of passive porous structures. On the coarse end of the scale are woven wire screens used to remove large and potentially devastating chips. In these units the wire cloth should be sintered in order to maintain strength and stability over extended service intervals typical for these devices. Most oil filters have a three-dimensional porous structure formed by fibrous materials. Relatively coarse filters contained packed fibers Copyright © 1994 CRC Press, LLC
Volume III
77
FIGURE 6. A composite filter pack with (1) metal or polymer support meshes upstream and downstream for strength and flow distribution; (2) fibers bonded with inert resin into graded pore structure for maximum service life; and (3) inner core to withstand maximum ∆P.
or chips, and moderately coarse filters have cellulose or cotton fibers. Finer filters are constructed from fibers of synthetic polymers and/or glass. The overlapping fibers, retained by resin binders, form the filter medium. In order to meet high performance demands the medium is incorporated into a composite filter pack as shown in Figure 6. The filter pack, in the form of one or more layers, is corrugated into a filter element, or cartridge, in order to obtain maximum surface area within the confines of the filter envelope. The cartridge may be placed into a housing accommodating auxiliary functions, which is then permanently installed in the system. In this case only the cartridge is replaced. Alternatively, the cartridge may be incorporated into a canister (spin-on can) which is mated to a port on the system, in which case the cartridge and outer metal container are replaced as a unit when the filter is spent.
Filter Stability Filter performance should be stable over time. Glass and synthetic fibers, along with relatively inert binders such as epoxy- and fluorocarbon-based polymers, are compatible with a broad range of fluids and produce filters with long-term stability that do not degrade with extended use. Many aerospace and military and several industrial specifications incorporate techniques for simulating operating conditions, such as flow surges and high temperature heat soaks, to ensure suitable in-service performance. These methods are listed in Table 4. Many industries have suffered by not using these performance concepts. Filter Efficiency and Particle Removal Depending on the morphology of the pores and the size distribution of the particles, some portion of the particles suspended in the influent fluid challenging the filter are retained, while
Copyright © 1994 CRC Press, LLC
78
CRC Handbook of Lubrication and Tribology
the remainder pass through. In order to perform the function of protecting machinery from contaminant particles, it is necessary that the filters remove a significant number of particles equal to or greater in size than the dynamic fluid films of the system components. In order to substantiate the ability of filters to remove such particles, two parallel approaches are taken. One is to specify filters determined by laboratory testing to be capable of retaining I these particles. The second is to monitor the fluid to insure that the particles are in fact being removed. The ability of a filter to remove particles is referred to as filter efficiency. A variety of I tests have evolved to measure filter efficiency under controlled (i.e., laboratory) conditions, often associated with the needs of a specific industry. Several of the better known of these procedures are summarized in Table 5, along with strengths and limitations. Beta values, Figure 7, have the advantage of describing filter efficiency over a range of particle sizes, as shown for a variety of media in Figure 8. Regardless of the “rating method” used, the manner in which filters capture particles should be understood. Filters behave like probability machines. Any particle challenging a filter has a chance of being captured or of passing through, with the odds for capture increasing with increasing particle size. For example, a 10-µm sized particle may have a 75% chance of Copyright © 1994 CRC Press, LLC
Volume III
FIGURE 7. Filter efficiency is defined by its “beta ratio”; downstream fluid quality may vary with specific application.
79
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
80
FIGURE 8. Representative beta curves. A steep beta curve demonstrates consistent, stable pore structure and high β performance across the selected particle range. Flatter curves indicate inconsistent pore structure and lack of control over the particle range. Micron (µm) filter ratings are given for β = 200.
being retained by filter A and thus a 25% chance of passing through into the downstream fluid. The odds for a larger 20 µm particle may change to 99% retention vs. 1% passage and, similarly, 1% retention and 99% passage for a 1µm particle. If it is judged that filter A is inadequate for protecting machinery in a given application, higher efficiency filter B may be selected, with 99% removal of 1 µm particles and comparably higher values for larger particles.
Sampling and Contamination Levels Another approach for assessing filter performance, as well as the level of contamination of operating systems, is fluid sampling. Typically fluid is withdrawn from a system and collected into sample bottles. The bottles are then forwarded to a laboratory for particle counting. Particle count data is usually reduced to a “cleanliness code”; two common examples are provided in Figure 9 and 10. Particle counting may be performed by drawing the fluid through a laboratory membrane (patch) and using an optical microscope to manually count particles isolated on the surface of the membrane. Alternatively, the fluid may be passed into an automatic particle counter consisting of an electronic sensor outputting signals to a processor calibrated to provide particle sizes. Both analytical methods require precautions. However, the greatest obstacle to accurately assessing system contamination levels is the fluid sampling operation. Sources of background counts are ubiquitous, and include dust motes, sampling valves, bottle cap seals, and the walls of sample bottles (flexible plastic bottles are the worst!). If system contamination is high, such as PCC 20/18/16, then background interference is less important. However, in clean systems of PCC 16/14/12 or better, the superposition of background counts can and does mislead expert as well as novice investigators. An increasing popular solution to this dilemma is to eliminate major sources of error by attaching an automatic particle counter directly on-line with the system.
COMPARISON OF COSTS AND BENEFITS
The cost/benefit ratios of filters, as for any equipment, needs to be considered. Copyright © 1994 CRC Press, LLC
81
Volume III
FIGURE 9. SAE AS4059 aerospace contamination classification. Costs The primary costs associated with filtration are cost of the filter, replacement maintenance time, and disposal costs. Replacement and disposal costs are strongly dependent on filter service life.
Service Life Service life is defined as the time interval of filter operation from installation to removal. The useful life of filters has been significantly enhanced by recent technological advances. As illustrated in Figure 11, building filter media with thinner (and strong) fibers increase the void volume of the porous structure. This provides more open space for fluid flow and for capturing particles, decreasing initial differential pressure and increasing dirt holding capacity. These factors lead to increases in service life by up to five times compared to older (thick fiber) technology. Another enhancement is to grade the filter media from coarse to fine pores in the upstream-to-downstream direction, as shown in the photomicrographs of Figure 12. This tapered pore design enables the coarser upstream surface to act as a prefilter for larger particles and allocates the finer downstream pores to smaller particles, thus further extending service life.
Dirt Capacity Current laboratory test methods providing data on filter dirt capacity are of limited utility. Ail of these methods measure the amount of test contaminant needed to load a filter from clean to terminal differential pressure. Unfortunately, the amount of material needed to plug a filter is highly sensitive to the shape and size distribution of the particles. And test contaminants—quite necessary for controlled reproducible laboratory testing—often are not representative of variable contaminants found in real-world operating systems. Table 6 enumerates major factors to be taken into account in order to increase the service life of filters.
Additional Considerations Filters are installed to protect machinery from contaminants. Choosing a low cost filter that provides marginal protection is false economy. Similarly, the filter must be able to withstand operating conditions, including fluid compatability at temperature. Much to the Copyright © 1994 CRC Press, LLC
FIGURE 10. The Pall Cleanliness Code references the number of particles greater than 2, 5, and 15 microns in each milliliter of fluid. The results of particle counting are plotted on a graph (shown above). The corresponding range code, shown at the right of the graph, gives the cleanliness code number for each of the three particle sizes. For the example above, the cleanliness code number would be 16/14/12. (Note: Proposed extension of ISO 4406).
Copyright © 1994 CRC Press, LLC
Volume III
Copyright © 1994 CRC Press, LLC
83
CRC Handbook of Lubrication and Tribology
84
FIGURE 11. Smaller fiber size provides more void volume, higher dirt capacity, lower pressure drop, and longer service life.
FIGURE 12. Advanced fiber technology and grading
dismay of many an incautious operator, an unstable filter can transform from a particle collector into a particle source.
Benefits of Filtration Component Life and Performance Fluid contamination causes 70 to 85% of all failures and wear problems in lubricated machines.6 The primary benefits derived from controlling contamination with filtration stem from increasing the life of components as well as enhancing component and system performance. For example, one study reports a six times increase in roller bearing fatigue life by changing from 40 µm to 3 µm filters.7 Another finds that maintaining extremely clean oil leads to bearings lasting more than 40 times theoretical L10 fatigue life without failures.8 An investigator from a major bearing company asserts that rolling contact fatigue life can increase up to 500 times by upgrading from 100 µm to 3µm filtration.9 Over 70% of hydraulic system failures are attributed to fluid contamination.10 The primary wear mode in hydraulic systems is sliding contact abrasive wear. Improvements result from Copyright © 1994 CRC Press, LLC
Volume III
85
removing hard abrasive particles down to the size of 1 µm or less. Reports from the field indicate pump life increases up to 10 times by maintaining clean fluid with high efficiency filters. One well-documented investigation by the U.S. Navy finds pump wear diminished more than 13 times by upgrading from 15 µm to 1 µm filters.” Similar improvements are found for other sliding contact elements, such as seals, bushings, and tapered bearing edges. Savings Economic benefits derived from improving system performance and component life are application dependent. Factors for consideration include:
1. 2. 3. 4. 5. 6. 7.
Increased up-time reliability, and safety Decreased maintenance time Lower component repair/replacement/transportation costs Lower component inventory Fewer rejects Reduced fluid purchases and disposal costs Reduced energy consumption
RECOMMENDED CONTAMINATION LEVELS—HOW CLEAN?
This question is best approached via three avenues. The first is historical, the second modern, and the third a synthesis. With the advent of the industrial revolution came comprehension that large pieces of debris quickly and unerringly failed equipment. They still do. Contacts, clearances, and flow passages are rapidly damaged or fouled by a single large particle. These chips may enter the system from external sources, be generated by failing components, and are even created when making a fitting. Therefore, for “catastrophic failure prevention filtration”, it is recommended that sintered woven mesh screens be placed immediately upstream of critical components. These screens should be capable of capturing chips 100 µm and larger, and perhaps smaller chips for more sensitive (usually high pressure) systems. The second approach has a more recent tradition. Modern lubrication theory has delineated the presence and size of dynamic fluid films. Particles the size of and larger than these films have been found to damage surfaces. There are orders of magnitude more small particles in the fluid film size range. Each of these contributes to the accumulating wear damage, culminating in component failure. Therefore, for “wear control filtration”, the target is to filter down to the size of dynamic fluid films. This proposal leads to a dilemma, since some of these films are as small as 0.1 µm. Although such filtration is common in industries concerned with low quantities of contaminants, such as electronics and pharmaceuticals, economically justifiable oil filters are currently available only to 1 µm (i.e., β1≥ 200). Therefore, the current Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
86
FIGURE 13. Guidelines for selection of filter rating and system cleanliness in Figure 10.
Copyright © 1994 CRC Press, LLC
87
Volume III
recommendation is to use filters capable of removing particles the size of the dynamic films of system components, down to as low as 1 µm. Advances into submicron range filtration are anticipated for the future. The requirements for large particle removal to attain “catastrophic failure prevention” and for small particle removal to achieve “wear control” have been synthesized into recommended cleanliness levels. When the appropriate degree of particle control is maintained, contamination-induced mechanical wear is minimized. The optimal cleanliness level depends on the type of machinery and criticality of operation. A summary of recommended cleanliness levels and of the degree of filtration required to achieve these levels is provided in Figure 13 for a wide variety of mechanical components.
REFERENCES
Lansdown, A. R., Lubrication, A Practical Guide to Lubricant Selection, Pergamon Press, New York, 1982, 38. Abner, E., Lubricant deterioration in service, Handbook of Lubrication, Vol. I, Booser, E. R., Ed., CRC Press, Boca Raton, FL, 1983. 3. Needelman, W. M. and Zaretsky, E. V., New equations show oil filtration effect on bearing life. Power Transm. Des., 33(8), 65, 1991. 4. Needelman, W. M., Filtration for wear control, Wear Control Handbook, (ASME), Peterson, M. B. and Winer, W. O., Eds., American Society of Mechanical Engineers, New York, 1980, 507. 5. Bensch, L., The overrated filter rating factor. Machine Design, 55, June 23, 1983. 6. Godfrey, D., Clean, dry oils prolong life of lubricated machines, Lubr. Eng., 46(1), 4, 1989. 7. Macpherson, P. B., Bachu, R., and Sayles, R., The influence of filtration on rolling element bearing life, Proc. 33rd Mech. Failures Prevention Group, Shives, T. R. and Willard, W. A., Eds., U.S. Dept. of Commerce, Gaithersburg, MD, 1981, 326. 8. Dalal et al., Final report on progression of surface damage in rolling contact fatigue, U.S. Navy Contract No. N00014–73-C-0461. 9. Jacobson, B., Cleanliness is required in lubrication systems, Swedish Engineering Society Annual Technical Week, 1987, Sweden. 10. Anon., Effective Contamination Control In Fluid Power Systems, Sperry Corporation, Flint, MI, 1980. 11. Ohlson, J., Effect of contamination and filtration level on pump wear and performance, SAE Aerospace Fluid Power and Controls Technology Meeting No. 96, 1984. 1. 2.
Copyright © 1994 CRC Press, LLC
RAPID DETERMINATION OF REMAINING USEFUL LUBRICANT LIFE Robert E. Kauffman
INTRODUCTION
This chapter deals with rapid techniques capable of determining the remaining useful live of lubricants, e.g., lubricating oils, hydraulic fluids, greases, etc. For this chapter, the remaining useful life refers to the remaining oxidative stability, and consequently the remaining antioxidant capacity of the lubricant. Lubricants which become unusable due to contamination, hydrolysis, loss of corrosion inhibitors, thermal breakdown, etc. are not discussed. The techniques discussed in this chapter generally require less than 1 hour to perform and are suitable for routine use. Whether the discussed technique is suitable for routine use by the reader will depend on the equipment being monitored. For stationary equipment with long oil change intervals (steam turbine and transformer oils may exceed 30 years) techniques with analysis times less than 8 h would be considered rapid. On the other hand, fleet maintenance programs for mobile equipment (automobiles, trucks, aircraft, etc.) would consider techniques with analysis times over 10 min unsuitable for routine use. Antioxidants, natural and synthetic, are present in lubricants to increase their thermaloxidative stability. Since the antioxidants are depleted with equipment operating time, they eventually become ineffective (unless they are replenished by lubricant or additive additions), allowing large changes to occur in the physical and chemical properties of the lubricant. The length of operating time from lubricant sampling until large changes occur in physical properties is referred to as the “remaining useful life” (RUL) of the lubricant (Figure l).1.2 The plots in Figure 1 show that the commonly monitored properties such as viscosity and acidity are insensitive to changes in the RUL of the lubricant. Additionally, these physiochemical properties are formula dependent and are affected by equipment operating conditions such as fuel dilution, viscosity shear, removal of oxidation products through evaporation or corrosion, etc. Since condition monitoring techniques based on such measurements as viscosity, acidity, color, conductivity, dielectric constant, particulates, etc. have been previously described in detail3.4 and are unable to provide accurate RUL evaluations, they will not be discussed in this chapter. Techniques capable of performing rapid evaluations of remaining antioxidant concentration or capacity will be covered in four basic categories: (1) thermal-oxidative stressing, (2) chemical-oxidative stressing, (3) electrochemical, and (4) instrumental techniques. Calculations of the actual RUL of a lubricant from the remaining antioxidant concentration (or capacity) and antioxidant depletion rate are described later in this chapter.
THERMAL-OXIDATIVE STRESSING TECHNIQUES
Of the rapid techniques discussed in this chapter, the majority of the remaining antioxidant evaluation techniques are based on accelerated thermal-oxidative stressing. Long-term stability tests (Figure 1) have been used extensively to predict the performance of lubricant additive packages and basestocks under various thermal-oxidative conditions. Air flow, metal catalysts, sample size, temperature, etc. of these long-term tests are chosen in an attempt to simulate the operating conditions of specific equipment. Long-term thermal-oxidative stability tests used to determine the useful lives of lubricants have been previously described in detail.5–9 As an example of a rapid, nonroutine stability test, the Penn State microoxidation test10 exposes a one-drop sample to a temperature in the 250°C range, followed by liquid 0-8493-3903-0/94/$0.00 + $.50
© 1994 by CRC Press, Inc. Copyright © 1994 CRC Press, LLC
89
90
CRC Handbook of Lubrication and Tribology
FIGURE 1. Percent remaining useful life, percent remaining additive, viscosity (40°C), and total acid number vs. hours of stressing time and remaining useful life at 150°C (laboratory stressing test) of a typical railroad diesel engine oil (TBN = 13, single grade).
chromatography to determine changes in the lubricant sample. This test, commonly applied to study degradation mechanisms for various petroleum and synthetic oils, has also found use in simulating oil oxidation life in automotive and diesel service.9–11 The use of liquid chromatography limits the potential of the Penn State microoxidation test for routine use. The rapid, routine techniques described in this chapter use accelerated thermal and oxidative conditions in combination with sample sizes small enough to provide thin films of lubricant. These operational conditions rapidly deplete the antioxidants in the lubricant sample and then rapidly degrade the lubricant. Various methods are then used to detect the “onset time” (isothermal conditions) or “onset temperature” (ramped temperature conditions) at which the rapid degradation begins. This onset time or temperature provides a measure of the antioxidant capacity of the entire system instead of individual antioxidant concentrations. Techniques12,13 which monitor basestock degradation by the chemiluminescence of the lubricant sample during stressing are limited by equipment costs and poor reproducibility at highly accelerated rates. However, a multiple sample chemiluminescence instrument has been reported12which has the potential for routine RUL evaluations. Other rapid tests use inverse gas chromatography,14 weight loss,15 and gas evolution rate16 to determine the oxidation induction times of lubricants. The rotating bomb oxidation test (RBOT)4,9,17–19 is an accelerated technique used extensively in monitoring remaining antioxidant capacities of steam and gas turbine oils inhibited by phenolic and amine type antioxidants. The RBOT as described in ASTM D227217 is performed by placing 50 g of lubricant into an axially rotated (100 rpm) stainless steel bomb heated at 150°C with an initial oxygen pressure of 90 psi. Water and a copper catalyst coil are added to the oil to simulate the steam turbine environment. The induction time is indicated by a rapid drop in oxygen pressure. Reported induction times for fresh steam turbine oils range from 175 to 380 min4 and a minimum RBOT induction time of 200 min has been used for Copyright © 1994 CRC Press, LLC
Volume III
91
qualifying candidate steam turbine oils.18 The RBOT has also been used to evaluate the RUL of petroleum gas turbine oils8 which produce induction times in the range of 44 to 2800 min. The RBOT has been less successful in evaluating hydraulic fluids, automotive oils, and other zinc containing oils due to gaseous oxidation byproducts which have been reported8 to affect induction time. The rig to perform the RBOT is available from several commercial sources. The thin film oxygen uptake test (TFOUT)9,20,21 is a modified RBOT which also monitors oxygen pressure to determine the end of the lubricant induction time. In contrast to the RBOT, the TFOUT has been used to evaluate the RUL of lubricants containing zinc dithiophosphate additives.20 The TFOUT employs fuel and metal catalysts and a reaction temperature of 160°C. Although RBOT and TFOUT techniques are used to perform “short term” evaluations of lubricants, their induction times, equipment costs and complexity have limited their use primarily to the electric utilities and well-equipped oil analysis laboratories. Of the accelerated thermal-oxidative stressing techniques, differential thermal analysis,1,2,22–28 in particular high pressure-differential scanning calorimetry (HP-DSC),24–28 appears to be the most rapid and easiest to operate. The HP-DSC techniques use oxygen or air pressures up to 500 psi to reduce volatilization (lubricant and additives) during the accelerated thermal-oxidation stressing performed isothermally in the 150 to 250°C range. Researchers have shown that isothermal conditions are better suited for remaining antioxidant capacity determinations than ramped temperature conditions.1.2.25 Once antioxidants in the lubricant are depleted, the temperature of the sample increases due to the exothermic oxidation of the basestock. In contrast to the RBOT and TFOUT techniques, induction times of the HP-DSC techniques range from 6 to 12 min for aircraft gas turbine oils at 250°C1.2.27 and from 10 to 40 min for fresh crankcase lubricants at 175°C with soluble metal catalysts.24 Another advantage is small sample size, less than 50 mg, which allows testing of small quantities of used or candidate oils. Good agreement has been reported between the HP-DSC induction times and long-term oxidation tests for aircraft turbine oils,1,2,25.27 crankcase lubricants,24 greases,26 and polyalphaolefin oils.28 A pseudo HP-DSC technique,1.27 which uses sealed sample pans prepared under an oxygen atmosphere, has also been developed to evaluate the remaining antioxidant capacities of aircraft turbine oils without the requirement for high pressure cells.
CHEMICAL-OXIDATIVE STRESSING TECHNIQUES
Due to the safety hazards of thermal-oxidative stressing techniques (high temperatures and pressures), chemical-oxidative stressing techniques based on free radical29,30 and cumene hydroperoxide1,2,31 titrations of the lubricant’s antioxidants have been developed. The free radical titration technique29,30 was developed by Ford Research to evaluate the antioxidant capacities of new and used lubricants. The technique is performed under an oxygen atmosphere at 60°C and uses a free radical initiator to produce peroxy radicals to deplete the antioxidants present in the oil sample diluted with hexadecane and cyclohexene. Once the antioxidants are depleted, the cyclohexene oxidizes causing a rapid decrease in the oxygen pressure. Fresh crankcase lubricants produced induction times of up to 75 min.29 The addition of cumene hydroperoxide1,2,31 depletes the antioxidants present in the oil sample diluted with toluene containing a nickel complex laser dye. Once the antioxidants are depleted, the cumene hydroperoxide reacts with the laser dye, causing the green solution to turn colorless. The rapid color change is monitored using a visible spectrophotometer. The length of time before the rapid decrease in color occurs has been used to evaluate the remaining antioxidant capacities of automotive crankcase oils and aircraft turbine oils.1,2,31 Copyright © 1994 CRC Press, LLC
92
CRC Handbook of Lubrication and Tribology
ELECTROCHEMICAL TECHNIQUES
Of the numerous electrochemical procedures currently used in analytical chemistry (polarography, coulometry, etc.), cyclic voltammetric techniques1,2,32–38 require the least expensive instrumentation, require the shortest analytical time (less than 10 seconds), and are the easiest to operate in evaluating remaining antioxidant concentrations of lubricants. In contrast to the thermal-and chemical-oxidative stressing, cyclic voltammetric techniques determine individual antioxidant concentrations through current-voltage relationships at solid electrodes. The voltage of one electrode (auxiliary electrode) is increased, then decreased (cycled) linearly with time, and the current produced at a second electrode (working electrode) is recorded as a function of voltage. The lubricant is dissolved (ester oils) or suspended (hydrocarbon oils and greases) into a solvent containing an electrolyte prior to analysis. The voltage at which current flow increases and the magnitude of the current flow increase are used to identify and quantify, respectively, the antioxidant(s) present in the lubricant sample. Cycling the voltage prior to analyzing for antioxidant content has been used to study the synergistic mechanisms displayed by aircraft turbine oil antioxidants.1.32.34 In most cases, an electrolyte is added to the lubricant/electrode interface by dissolving the lubricant into an electrolyte containing solvent1.2.32–35 or into an electrolyte containing solid film.35 One cyclic voltammetric technique38 does not require electrolyte addition to the lubricant/electrode interface and can be performed in a temperature range of room temperature (diluted with a solvent) up to 300°C (inserted into the lubrication system of operating equipment). The technique has been used to perform simultaneous antioxidant depletion and oxidative degradation (condition monitoring) evaluations of used aircraft turbine oils.38 To date most cyclic voltammetric techniques have been used to quantitate and identify the antioxidants present in fresh aircraft turbine oils.1.2.32.34.37.38 Cyclic voltammetric techniques have also been used to evaluate the remaining antioxidant concentrations of used aircraft turbine oils,1.2.32.34 and of fresh and used automotive crankcase oils.33.36 A commercial “remaining useful life evaluation rig” (RULER) was specifically designed to perform cyclic voltammetric analyses of different type lubricants.1.2.34 In the case of automotive crankcase oils, depletion of zinc dithiophosphate-type additives detected by the RULER was correlated with an increase in metallic wear debris.36 In the case of aircraft oils, accelerated depletion of secondary aromatic amine antioxidants was correlated with seal damage2.38 and was used to predict engine failures prior to component damage.1.34 Steam I turbine oils; transformer oils; transmission fluids; marine, truck, and railroad diesel engine oils; hydraulic fluids; and greases have also been successfully analyzed by the RULER.38
INSTRUMENTAL TECHNIQUES
Instrumental techniques have been used to identify and quantify the antioxidants in fresh lubricants and to monitor depletion of antioxidants in used lubricants. Gas chromatography1,44,45 and liquid chromatography10,40–45 have been used to perform remaining antioxidant concentration evaluations with varying degrees of success. Gas chromatography, which separates a lubricant into its components by boiling point, is unable to quantitate the antioxidant species generated during lubricant use,4 and consequently, gas chromatographic techniques tend to underestimate the antioxidant capacities of used lubricants. Liquid chromatography (LC), which dissolves the lubricant in a solvent and then separates the lubricant components by polarity (high performance LC)40–42,44,45 or molecular weight (gel permeation LC),10.11.43 is able to quantify the original and generated antioxidant species.45 Use of liquid chromatography for routine evaluation of remaining antioxidant concentration has been limited by its expense, use of toxic solvents, and complexity of sample preparation. Copyright © 1994 CRC Press, LLC
Volume III
93
Economical thin layer chromatography (TLC),46,47 uses different solvent mixtures to dissolve and separate lubricants into their components on coated glass plates. After a developing time exceeding 1 hour, an ultraviolet lamp is used to quantitate the antioxidants present. TLC techniques have been used to monitor the depletion of antioxidants in used aircraft turbine oils46 but are limited for routine use by their long developing times. Of the different instrumental techniques used to evaluate the remaining antioxidant concentrations of lubricants, techniques based on Fourier transform infrared spectroscopy (FTIR)3,818,22,42,48–52 are the most widely used. As the oil sample is scanned in a wavelength range of 2 to 50 microns, the amount of light absorbed at each wavelength is then used to identify the type and concentration of each component present in the lubricant. In contrast to cyclic voltammetric techniques which respond only to the antioxidant species, FTIR responds to all components present in the lubricant including the basestock. Computer software is able to reduce the complex spectrum into quantitative antioxidant measurements and to evaluate lubricant contamination and oxidative deterioration (condition monitoring) of the basestock. The U.S. military services52 concluded that FTIR was more suitable for condition monitoring than for remaining antioxidant concentration evaluations of used crankcase oils. However, another study9 obtained good correlation between additive depletion and FTIR results for oxidized crankcase oils. Still, another study18 determined that the antioxidant concentrations determined by FTIR and the induction times of RBOT tests showed much better correlation for used steam turbine oils than for used turbine oils obtained from nuclear power plants. A commercial FTIR spectrophotometer4,9 has been designed specifically for lubricant analyses.
ASSESSMENT OF TECHNIQUES
In order to select the analytical technique best suited for routine RUL evaluations, numerous factors must be considered: number of samples analyzed daily, value of monitored equipment, expertise of operator, time requirements, etc. Experimental parameters and remaining antioxidant evaluation capabilities of the various techniques are assessed in Table Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
94
1. The chemical-oxidative stressing techniques require toxic chemicals, are not commercially available, and consequently were not considered for routine RUL evaluations. Table 1 indicates that for fleet operators, on-site analysis, high throughput labs, etc. where instrumental costs, analysis time, and ease of operation are most important, cyclic voltammetric techniques are best suited for development into RUL evaluations. If instrument costs are not an overriding factor and condition monitoring as well as remaining antioxidant evaluations are of interest, FTTR techniques are better suited for routine use. One limitation of the cyclic voltammetric and FTIR techniques is their inability to predict the effects of different additive combinations and basestocks on the lubricant antioxidant system. Consequently, thermal-oxidative stressing techniques are better suited when additive or basestock differences may affect remaining life evaluations. However, the most accurate and rapid remaining antioxidant evaluations are obtained using a combination of techniques. Cyclic voltammetric or FTIR techniques can be used to screen incoming batches of fresh lubricants and to monitor used oil samples to ensure that the types and concentrations of antioxidants are sufficient to inhibit oxidative degradation of the lubricant during use. When the screening process indicates that fresh samples contain new types of antioxidants or that antioxidant concentrations of used samples have decreased substantially, then thermal-oxidation techniques are required for accurate RUL evaluations. Since these screening methods require small sample sizes and are very rapid (Table 1), the combined methodology allows close monitoring of used oils with minimal expense and operator time.
CALCULATING REMAINING USEFUL LIFE
Once the technique has been chosen, studies must be performed to relate the results with the RUL evaluations of a lubricant. The studies are performed by obtaining a series of fresh and stressed lubricant samples from a long-term stability test or operating equipment of interest. The lubricant samples are then analyzed for remaining antioxidant by the selected technique(s) and are characterized by condition monitoring techniques, e.g., viscosity and total acid number. By plotting the remaining antioxidant concentration or capacity and condition monitoring measurements vs. operating time (or stressing time), a set of curves similar to Figure 1 may be produced. If the antioxidant(s) depletes linearly with increasing operating time, the depletion reaction is “zero order”. However, if the antioxidant depletion rate decreases as the antioxidant concentration or capacity decreases, the depletion reaction is “first order” and the log of the antioxidant concentration or capacity will be linear. Effects of lubricant makeup and more complex mechanisms will result in nonlinear RUL plots requiring curve fitting routines for RUL evaluations. The following equations are then used to determine the % RUL and the actual RUL (in operating time) of a lubricant sample [assuming zero or first (log values) order]:
Copyright © 1994 CRC Press, LLC
Volume III
where
95
A.C. = antioxidant concentration or capacity O.T. = operating time 0% RUL = antioxidant concentration or capacity at which condition monitoring measurements begin to accelerate (Figure 1) 100% RUL = antioxidant concentration or capacity of fresh lubricant
Since lubricant formulations vary with manufacturer, studies should be performed for each lubricant used in a particular application. For example, the % remaining antioxidant concentrations vs. hours of remaining useful life at 150°C [laboratory test9] are shown in Figure 1 for a railroad diesel engine oil. The useful life of 120 h (Cu absent) was determined from the breakpoints in the physical and chemical property curves shown in Figure 1. Although the presence of Cu decreased the useful life of the stressed oil to below 60 h,38 0% RUL still occurred at 20% remaining additive. By assigning 100% RUL to the fresh oil and 0% RUL to the sample with 20% remaining additive, % RUL values could be calculated for each stressed sample, and the actual RUL of any sample could be determined from the % RUL of the previous sample.
VALUE OF REMAINING USEFUL LIFE MEASUREMENTS
RUL evaluations have numerous advantages in determining the health of a monitored lubrication system. First, antioxidant concentrations or capacities of incoming batches of fresh lubricants can be checked to ensure they meet minimal requirements and to adjust lubricant change intervals. For instance, although two steam turbine oils with RBOT induction times of 205 and 380 min4 would pass the 200-min minimum set by one user,18 they would require different change intervals to ensure that the 380-min lubricant was not discarded prematurely. The second benefit of RUL evaluations is their ability to determine additive depletion rates in different operating equipment. The linear decrease in RUL shown in Figure 1 simulates a lubrication system which experienced consistent use without oil additions. However, a significant number of applications require periodic lubricant additions to replace losses by seal leakage, volatilization, etc. The % RUL vs. operating time plot in Figure 2 was produced during a aircraft turbine test and demonstrates the effects of oil additions. Each time oil was added to the system, the % RUL value increased, since the fresh oil had a higher concentration of antioxidants than the used oil. If the lubricant addition rate supplies antioxidant to the system at a rate equal to or greater than the antioxidant depletion rate of the lubricant system, the % RUL of the used oil will level off and become approximately steady state with increasing operating time (Figure 2). Consequently, systems with low antioxidant depletion rates and high makeup will not require lubricant changes and the lubricant will experience minimal changes in its physical and chemical properties with increasing operating times (Figure 2). However, RUL evaluations are highly valuable for relating lubricant system performance to makeup rates. Even though the viscosity and total number measurements are level throughout the test in Figure 2, the % RUL decreased to 70%. If the RUL value begins to increase with operating time, the user is alerted to an increase in the lubricant makeup rate (assuming no separate addition of antioxidant and that the fresh oil has not changed) indicating increased oil loss, e.g., worn seal. Even more important, if the % RUL begins to decrease rapidly, the user is alerted to the initiation of abnormal conditions which are accelerating antioxidant depletion prior to severe lubricant degradation and possible component damage. Examples of RUL evaluation capabilities to detect abnormal operation are shown in Figure 3 for a series of used oil samples Copyright © 1994 CRC Press, LLC
96
CRC Handbook of Lubrication and Tribology
FIGURE 2. Effects of lubricant additions on remaining useful life, viscosity (40°C), and total acid number (TAN) vs. hours of engine operation for the oils obtained from an aircraft turbine test stand (+ signifies oil addition).
FIGURE 3. Percent useful life vs. the hours since last oil change for used oil sample series obtained from normal and abnormal operating C-130 aircraft engines.
Copyright © 1994 CRC Press, LLC
Volume III
97
FIGURE 4. Percent remaining additive vs. stressing time at 175°C for an ester-based, clay-filled grease (hairline cracks, hardening and separation noted).
obtained from the engines of C-130 military transport aircraft2 During normal operation, the RUL values remain above 90%. The rapid decreases in % RUL for the C-l and C-2 aircraft engines (Figure 3) were a result of cracked seals which allowed hot air to come in contact with the oil, accelerating the antioxidant depletion. Previous research1,34 determined that for abnormally operating aircraft engines (military and commercial), the % RUL of the lubricant samples decreased rapidly prior to engine failure. Similar rapid decreases in RUL values have been determined during severe wear for other types of lubricants and greases.33,38 In the case of greases, RUL evaluations can be used to monitor the depletion of different type additives. The concentrations of two additives with antioxidant capacity decreased at different rates (Figure 4) in grease samples heated at 175°C as thin films in aluminum pans. After additive 1 in Figure 4 decreased to 15% of its original value, the heated grease samples began to harden and crack. Grease samples obtained from failed bearings showed an accelerated depletion of additive 2 compared to normally operating bearings.38 Finally, RUL evaluations can be used to extend lubricant change intervals. During its useful life, the physical and chemical properties of the lubricant remain fairly constant (Figures 1 and 2). Therefore, if monitoring techniques such as cyclic voltammetry or FTIR are used, the identified antioxidants can be added to the used lubricant in the correct proportions to restore their original concentrations prior to changes in the lubricant’s physical and chemical Copyright © 1994 CRC Press, LLC
98
CRC Handbook of Lubrication and Tribology
FIGURE 5. Percent remaining additive, viscosity (40°C), and total acid number (TAN) vs. stressing time at 150°C for a typical antiwar hydraulic fluid.
properties. For example, the concentrations of two different type antioxidants in an antiwear hydraulic fluid decrease at different rates when stressed in a laboratory oxidation test9 (Figure 5). Used lubricant samples analyzed in our laboratory have shown that the depletion rates of the individual antioxidants are dependent on the application, e.g., steam turbine, large hydraulic systems, etc. tend to accelerate the depletion of phenolic type antioxidants whereas I precision machining equipment, small hydraulic systems, etc. tend to accelerate the depletion of multifunctional (N, S containing) additives. Consequently, the drain periods of lubricants can be extended more efficiently by additive replenishment if the antioxidant undergoing accelerated depletion is identified for each application.
REFERENCES
1. Kauffman, R. E. and Rhine, W. E., Assessment of remaining lubricant life, Rep. No. AFWAL-TR-86–2024, Nov. 1986 (NTIS AD-A177 186). 2. Kauffman, R. E., Techniques to evaluate the remaining useful lubricant life of gas turbine engine lubricating oils, 44th Meet. Mechanical Failures Prevention Group Proc., Vibration Institute, Willowbrook, IL, 1990, 121. 3. Booser, E. R., Ed., Handbook of Lubrication, Vol. 1, 1983, 481–487 and 517–531.
Copyright © 1994 CRC Press, LLC
Volume III
99
4. Young, W. C. and Robertson, R. S., Eds., Turbine Oil Monitoring, STP1021, ASTM, Philadelphia, 1989. 5. Oxidation characteristics of inhibited steam-turbine oils, ASTM Meth. D-943, ASTM, Philadelphia, 1981, 1991. 6. Hsu, S. M., Review of laboratory bench tests in assessing the performance of automotive crankcase oils, Lubr. Eng., 37(12), 722, 1981. 7. Oxidation stability of lubricating greases by the oxygen bomb method, ASTM Method D-942, ASTM, Philadelphia, 1990. 8. Philadelphia, PA (1986), Ed. Stadtmiller, W. H. and Smith, A. N., Eds., Aspects of lubricant oxidation, ASTM Pub., STP1021, ASTM, Philadelphia. 9. Stauffer, R. D. and Thompson, J. L., Improved bench oxidation tests for railroad diesel engine lubricants, Lubr. Eng., 44(5), 416, 1988. 10. Gunsel, S. et al., Evaluation of some poly-alpha olefins in a pressurized Penn State microoxidation test, Lubr. Eng., 43(8), 629, 1987. 11. Perez, J. M., Kelley, F. A., Klaus, E. E., and Bagrodia, V., Development and use of Penn State microoxidation test for diesel engine oils, SAE Pap., 872028, Toronto, 1987. 12. Zlatkevich, L., New chemiluminescence apparatus and method for evaluation of thermal oxidative stability of lubricants, Lubr. Eng., 44(6), 544, 1988. 13. Pei, P. et al., Chemiluminescence instrumentation for fuel and lubricant oxidation studies, Lubr. Eng., 45(1), 9, 1989. 14. Sen, A. K. and Kumar, R., Oxidation stability of lubricants by inverse phase gas chromatography, Lubr. Eng., 47(3), 211, 1991. 15. Naga, H. H. and Samen, A. E., The effect of base stock volatility on lubricating oil oxidation stability, Lubr. Eng., 44(11), 931, 1988. 16. Ravner, H. and Wohltjen, H., The determination of the oxidative stability of several deuterated lubricants by an electronic gas sensor, Lubr. Eng., 39(11), 701. 1983. 17. Oxidation stability of steam turbine oils by rotating bomb, ASTM Meth. D-2272, ASTM, Philadelphia, 1985, 1991. 18. Yoshida, T. and Iqarashi, J., Consumption of antioxidant of turbine oil in service unit, Trib. Trans., 34(1). 51, 1991. 19. Strigner, P. L. and Brown, K. J., Some properties of Canadian steam turbine oils, Lubr. Eng., 43(4), 283, 1987. 20. Ku, C. S. and Hsu, S. M., A thin film oxygen uptake test for the evaluation of automotive crankcase lubricants, Lubr. Eng., 40(1), 75, 1984. 21. Hsu, S. M. et al., Mechanisms of additive effectiveness, Lubr. Sci., 1, 2, 165, 1991. 22. Biswas, A. K. et al., Evaluation of antioxidants in lubricating oils by differential thermal analysis and IR spectroscopy, Wear, 82(1), 45. 1982. 23. Ohibach, K. H. et al., Simultaneous thermal analysis-mass spectrometry on lubricant systems and additives, Themochim. Acta, 166, 277, 1990. 24. Hsu, S. et al., Evaluation of automotive crankcase lubricants by differential scanning calorimetry, Soc. Automot. Eng., Spec. Publ., SP-526, 1982, 127. 25. Zeman, A., Differential-Scanning kalorimetric (DSC,PDSC)-Mozlichkeiten bei der Beurteilung der thermirchoxidativeen stabilityät von synthetischen Flugturbineolen, Schmiertech. Tribol., 29, 1982, 55 (in German). 26. Petrova, L. N. et al., Micro methods for evaluation of the physiochemical properties of lubricating greases, Khim. Tekhnol. Topl. Masel, 1, 37, 1987. 27. Kauffman, R. E. and Rhine, W. E., Development of a remaining useful life of a lubricant evaluation technique. I. Differential scanning calorimetric techniques, Lubr. Eng., 44(2), 154, 1988. 28. Paige, H. L. et al., A systematic study of the oxidative stability of silahydrocarbons by pressure differential scanning calorimetry, Lubr. Eng., 46(4), 263, 1990. 29. Mahoney, L. el al., Determination of the antioxidant capacity of new and used lubricants: methods and applications, Ind. Eng. Chem. Prod. Res. Div., 17, 250, 1978. 30. Koreck, S. et al., Antioxidant consumption and oxidative degradation of lubricants, Nat. Bur. Stand. Pub. 584. 1980, 227. 31. Kauffman, R. E. and Rhine, W. E., Development of RULLET. II. Colorimetric methods, Lubr. Eng., 44(2), 162, 1988. 32. Kauffman, R. E., Method for evaluating the remaining useful life of a lubricant, U.S. Patent 4,744,870, May 17, 1988. 33. Kauffman, R. E., Method for Evaluating the Remaining Useful Life of a Hydrocarbon Oil, U.S. Patent 4,764,258, Aug. 16, 1988. 34. Kauffman, R. E., Development of a remaining useful life of a lubricant evaluation technique. HI. Cyclic voltammetric methods, Lubr. Eng., 45(11). 709, 1989. 35. Cheek, G. T. and Mowery, R., Determination of antioxidants in lubricating oils using ultramicroelectrodes,Anal. Chem., 61, 1467, 1989. 36. Kirkpatrick, J. F., Diagnostic Tools for Reciprocating Engine Systems, Rep. No. GRI-91/0041, Gas Research Institute, Chicago, Jan. 1991.
Copyright © 1994 CRC Press, LLC
100
CRC Handbook of Lubrication and Tribology
37. Price, R. J. and Clark, L. J., Chemical sensing of amine antioxidants in turbine lubricants, Analyst, 116, 1121, 1991. 38. Kauffman, R. E., Remaining useful life measurements of gas turbine engine oils, diesel engine oils, automotive oils, hydraulic fluids, and greases using cyclic voltammetric methods, in Condition Monitoring International Conference Proceedings, Joint Oil Analysis Program, Technical Support Center, Naval Air Station, Pensacola, Florida, Nov. 1992. 39. Rai, M. M. et al., GC-FTIR analysis of amine type antioxidants in petroleum products and their influence on environment, Indian J. Environ., Prot., 9(2), 124, 1989. 40. Barth, P. et al., Quality control of used synthetic aviation turbine oils by analytical methods. I. Determination of the antioxidative capacity by HPLC and GC, Fresenius Z. Anal. Chem., 314, 25, 1983. 41. Kholostova, G. G., Analysis of the quality of aviation lubricating oils by liquid and gas-liquid chromatography, Khim. Tekhnol. Top. Masel, 6, 24, 1986. 42. Combellas, C. et al., Coupling of a high performance liquid chromatograph with a Fourier transform infrared detector, J. Chromatogr., 259(2), 211, 1983. 43. Jones, W. R., Liquid chromatographic analysis of a formulated ester from a gas turbine test, Lubr. Eng., 41(1), 22, 1985. 44. Musha, K. et al., Rapid analysis of lubricating oil additives by reversed-phase high performance liquid chromatography, Bunseki Kagaku, 34(3), 26, 1985 (in Japanese). 45. Keller, M. A. and Saba, C. S., Chromatographic analysis of phenyl-1-naphthylamine and 4,4’dioctyldiphenylamine and their intermediate oxidation products in oxidized lubricants, J. Chromatogr., 409, 325, 1987. 46. Sniegoski, P. J., A kinetic study of lubricant antioxidant depletion in aircraft gas turbine engines, Lubr. Eng., 41(1), 11, 1985. 47. Kuniya, J., Separation of additives in lubricating oils by TLC, Bunseki Kagaku, 37(9), 87, 1988 (in Japanese). 48. Wooton, D. L. et al., Infrared analysis of heavy-duty diesel engine oils, Soc. Automot. Eng., Spec. Pub., SP-589, 71, 1984. 49. Coates, J. P. and Setti, L. C., Infrared spectroscopic methods for the study of lubricant oxidation products, ASLE Trans., 29(3), 394, 1986. 50. Zhang, Z. and Yang, S., Separation and identification of greases, Runhua Yu Mifeng, 1, 25, 1988 (in Chinese). 51. Ofunne, G. C. et al., Studies on the ageing characteristics of automotive crankcase oils, Trib. Int., 22(6), 401, 1989. 52. McCaa, B. B. and Coates, J. P., Evaluation of used crankcase oils using computerized infrared spectrometry, JOAP-TSC Report 84–01, June 1984, (DTIC AD-A152 993).
Copyright © 1994 CRC Press, LLC
FRICTION AND WEAR OF CERAMICS Said Jahanmir and Traugott E. Fischer
INTRODUCTION
Technical ceramics offer unique capabilities as tribe-materials in diverse applications requiring wear resistance and chemical stability at elevated temperatures. Ceramics are generally defined as inorganic nonmetallic solid materials. This definition includes not only materials such as pottery, porcelains, refractories, cements, abrasives, and glass, but also nonmetallic magnetic materials, ferroelectrics, and a variety of other new products. Renewed interest in ceramics is rooted in unique materials classified as electronic and optical ceramics, and structural ceramics.1 The outstanding mechanical properties of technical ceramics are their hardness, even at elevated temperatures, low density, and high fatigue resistance. A major drawback is thenlow toughness, which is the cause of their limited reliability. Ceramic materials possess a wide range of chemical properties. Some ceramics are in the most stable thermodynamic form, while others, for example oxides and nitrides, are highly reactive with water at room temperature. The thermal properties also span a wide range; while diamond shows the highest heat conductivity of all materials, zirconia and magnesia are among the best thermal insulators. This chapter first gives a brief review of some fundamental issues regarding friction and wear of technical ceramics, followed by a discussion of structure, processing methods, mechanical properties, and tribological behavior of these materials. Finally, typical tribologyrelated applications of ceramics are reviewed.
GENERAL GUIDELINES GOVERNING TRIBOLOGICAL BEHAVIOR
Friction and wear of materials involve both elastic and plastic deformation in metals2 and elastic deformation and microfracture in ceramics.3,4 Friction is generally described in terms of the friction coefficient—the ratio of friction force to the normal load pressing the two surfaces together.
Friction and Lubrication While unlubricated ceramics usually exhibit high friction coefficients just as metals do, low friction coefficients have been observed in short sliding tests due to lubricating oxides or contaminant layers on the surface. These layers are soon worn away and the friction coefficient assumes the high values shown in Table 1. The statement, often made, that ceramics possess inherent low friction because of their low adhesion is not borne out by the data. Other factors such as tribochemical reaction with the environment could result in low friction. Lubricants perform two essential functions: hydrodynamic lubrication at high speeds and low loads, and boundary lubrication at low speeds and high loads. Hydrodynamic lubrication is as effective with ceramics as with metals. Hydrodynamic lubrication is facilitated by running-in of bearings, which involves a controlled form of wear in early operation with the effect of decreasing surface roughness. In metals, this commonly occurs by plastic deformation of the asperities. Since little plastic deformation occurs in the wear of ceramics, running-in of ceramics occurs predominantly by fracture and tribochemistry.5 In boundary lubrication, the shear strength or adhesion between surfaces decreases with adsorption of suitable molecules (boundary lubricants) from the fluid. This is as effective with Copyright © 1994 CRC Press, LLC
103
104
CRC Handbook of Lubrication and Tribology
ceramics as with metals. Chemically inert hydrocarbons such as pure paraffins or basestock oils act as boundary lubricants, resulting in a friction coefficient as low as 0.12 with ceramic surfaces,5 unlike metals. We speculate that the same acid and base sites on ceramics that are responsible for their catalytic activity in hydrocarbon cracking and isomerization are also responsible for chemisorption of paraffins. Other schemes are being developed for lubrication at extreme temperatures: supply of lubricant additives through a gas6 and deposition of carbonaceous layers by catalytic decomposition of hydrocarbons.7 Mechanical Aspects of Wear The wear volume of many materials can be expressed by8
where K is the dimensionless wear coefficient, L the normal load pressing the surfaces together, H the hardness expressed in units of load per unit area, and S the sliding distance. This equation implies that the wear volume is proportional to the applied load and inversely proportional to the hardness of the material. These relationships are not generally true for ceramics. While Equation 1 is based on contact and wear phenomena with plastic deformation, as in metals, the hardness effect is often omitted to give the wear constant k (expressed in units of mm3/N m), where k = K/H. The wear constant, defined as wear volume per unit sliding distance per unit load, is often preferred in describing wear of ceramics. An equation for abrasive wear of ceramics derived by Evans and Marshall3 assumes a purely brittle form of wear due to the extension of lateral cracks as the indenter slides along the surface to produce a scratch. It predicts that the wear volume varies with the normal force L, hardness H, sliding distance S, and toughness KIc as where C is a constant. This equation, which agrees reasonably well with measurements, shows that hardness and toughness both contribute to abrasive wear resistance. In sliding (nonabrasive) contact, the dominant wear mechanism of ceramics is microfracture.4,9 Penetration of a harder solid into the surface of the material does not, as a rule, occur, and toughness is the main determinant of wear resistance. This was verified in the case of zirconia prepared with toughness varying from 2 to 11.5 MPa m1/2 by doping with yttria.9 Wear resistance of these materials increases with the fourth power of toughness (by a factor of 1200 when toughness increases by a factor of 6). There is evidence for some plastic deformation during wear of certain ceramics. In transformation-toughened zirconia, cleanly cut grooves on the surface worn in air attest to plastic deformation. Dislocation pileups at Copyright © 1994 CRC Press, LLC
Volume III
105
grain boundaries observed in wear tracks of alumina, however, cause intergranular fracture rather than plastic deformation. An interesting feature of most ceramic wear is a transition from mild to severe wear observed at a certain load, which depends on the sliding velocity and temperature. Severe wear is caused by relatively large-scale fractures propagating under the influence of macroscopic contact stresses.10 In the mild wear regime below the transition to severe wear, wear rates of most ceramics still vary rapidly with load or macroscopic contact stress. The wear rate of alumina increases with the fifth power of a load11 and adherent wear debris slow down wear at large sliding distances.12 The wear rate of zirconia increases almost as fast in the mild wear regime.13 At low contact pressures, ceramics present a wear resistance three to four orders of magnitude better than that of metals; this advantage is lost at high contact pressures.14 In rolling contact, silicon nitride performs much better than other ceramics. High fracture strength and outstanding fatigue resistance allow the material to sustain high contact stresses without damage. Wear rate of a number of ceramics in rolling wear could be expressed by the formula15
where w is wear volume per unit sliding distance, Pm is the maximum macroscopic Hertzian contact stress, Rmax the average maximum surface roughness, and KIc the toughness of the material. Inclusion of the surface roughness illustrates its additive effect on contact stress in the wear process. When ceramics are used in contact with metals, the latter deform plastically in the contact and do not generate sufficient contact stresses to induce wear by fracture in the ceramic. In unlubricated service, the metal is worn and its wear debris transfers to the ceramic. The latter wears either by fatigue or, more often, by tribochemical reaction with the metal and environment. An illustration of this phenomenon is the dissolution of diamond when used to cut steels.
Tribochemical Aspects Despite the outstanding corrosion resistance and chemical inertness of ceramics, early researchers uncovered a strong influence of chemical environment, especially the humidity, on the wear of ceramics, even at room temperature.16 The wear rate of silicon nitride decreases by two orders of magnitude when the relative humidity of argon or air is increased to 100%. The wear track is smooth and wear debris are predominantly amorphous silicon oxide with dispersed fine crystallites of silicon nitride a few nanometers in size.5 Sliding in water produces ultrasmooth surfaces that allow hydrodynamic lubrication by a water film with very low friction at low velocities (6 cm/s) and high bearing pressures (100 MPa). Oxides formed on ceramics can act as lubricants under the right conditions. In silicon nitride, friction coefficients as low as 0.05 were measured when surfaces were reacted with water.17 Alumina sliding in water forms a lubricating hydroxide that reduces friction to 0.3. Cubic zirconia is very sensitive to chemisorption embrittlement, mostly by interaction of grain boundary phases with water. Sliding in humid air and in water13 causes an increase in wear rate of zirconia by about a factor of 10. Alumina exhibits a much weaker tendency to chemisorption embrittlement than zirconia.
Wear of Lubricated Ceramics Hydrocarbon lubricants are effective in reducing the friction coefficient of ceramics. A modest decrease of wear has been obtained with silicon nitride, but not with alumina and
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
106
toughened zirconia.13 Chemisorption embrittlement by the chemically active boundary lubricants (for instance, fatty acids) appears to increase wear and compensate for the reduction in contact stresses. It must be remembered, however, that unlubricated wear rate of alumina and toughened zirconia is very low, lower at low contact loads than for metals with boundary lubrication. A recent investigation on antiwear additives for ceramics18 has shown that phosphorouscontaining additives are effective for silicon nitride, especially when the ceramic contains iron impurities. No effective general guidelines for the development of lubricant additives for ceramics exist yet; their establishment depends on more thorough knowledge of the tribochemistry of these materials.
TRIBOLOGICAL BEHAVIOR OF SPECIFIC CERAMICS
Selection of materials for tribological applications is based not only on friction and wear behavior, but also on other application requirements such as strength, fatigue resistance, corrosion resistance, dimensional stability, thermal properties, reliability, ease of fabrication, and cost. This section reviews the processing techniques, properties, and tribological behavior of five important classes of structural ceramics: alumina, zirconia, silicon nitride, sialon, and silicon carbide. Additional information is available on structure, processing techniques, and properties.1,19–21
Alumina Ceramics Ceramics based on alumina have been used in commercial applications for many years because of their availability and low cost. Alumina ceramics are often classified either as high aluminas having more than 80% aluminum oxide, or as porcelains having less than 80%. High aluminas are used in many mechanical devices and in electronics. Pure aluminum oxide, Al2O3, has one thermodynamically stable phase at room temperature with hexagonal crystal structure, designated as alpha phase. Often the term corundum is used for alpha alumina. Commercial high purity (>95%) alumina ceramics usually contain MgO as sintering aid, and SiO2 and Na2O impurities. In less expensive lower grades, silicates are usually used as the sintering aid. Strength and other properties improve as the percentage of alumina is increased (Table 2). However, cost increases because of processing difficulty. Commercial aluminas are processed by pressureless sintering. Table 3 shows typical unlubricated friction and wear data for different types of alumina. The friction coefficient and the wear constants depend on the composition and microstructure of the material, as well as on the test conditions. These data should be used only as a guide, and the original references should be consulted for more details. Although the tribological behavior of different materials can be compared only if the tests are conducted under the same conditions, the data in the table show certain trends. In selfmated tests, the friction coefficients of high purity aluminas (>95%) at room temperature are smaller than those of lower purity grades (85%). The friction coefficient increases if the counterface is changed from alumina to steel. In addition, the friction coefficient decreases as sliding. speed increases. The friction coefficient in ceramic-ceramic sliding components is reduced if the alumina counterface is replaced with either zirconia, silicon nitride, or silicon carbide. The wear constant is used in Table 3 to represent the amount of wear. Because of the variability of test results, only a few conclusions are possible as to the effect of conditions Ion wear. However, data in Table 3 suggest that wear constant increases as either sliding speed or load is increased. The wear transition diagram in Figure 122 for a high purity alumina (99.8%) sliding in air displays the effects of normal load and temperature on tribological behavior. In this figure, the Copyright © 1994 CRC Press, LLC
Volume III
107
wear coefficient of Equation 1 is used to describe the wear data, which were obtained in selfmated unlubricated tests. At low temperatures in region I (T < 200°C), tribochemical reactions between alumina and water vapor in the environment control performance; friction coefficient and wear coefficient are low at 0.40 and 10-6, respectively. At intermediate temperatures (200°C < T < 800°C), wear behavior depends on contact load. At low loads in region II, wear occurs by plastic flow and plowing; friction and wear are low, similar to those observed at low temperatures. At loads higher than a threshold value, in region IV, severe wear by intergranular fracture results in a friction coefficient of 0.85 and a wear coefficient larger than 10-4. At temperatures above 800°C (in region III), both the friction coefficient and wear coefficient are low because of the formation of a silicon-rich surface layer from diffusion and viscous flow of the glassy grain boundary phase. The effect of sliding speed and temperature for a 99.7% alumina is shown in Figure 2 for a 10 N normal load.23 The wear constant in Figure 2 is larger at 400 and 800°C than at room Copyright © 1994 CRC Press, LLC
108
CRC Handbook of Lubrication and Tribology
FIGURE 1. Wear transition diagram for 99.8 alumina at 0.0014 m/s sliding speed. Friction coefficient, f, and wear coefficient, K, for each region are indicated on the figure. (From Dong, X. and Jahanmir, S., J. Am. Ceram. Soc., 74, 1036, 1991.)
temperature. At higher temperatures, the wear constant does not seem to depend on speed, but is very speed sensitive at room temperature, with an increase by more than two I orders of magnitude as the speed is increased. The friction coefficient at room temperature is also larger than at either 400 or 800°C. It increases from 0.4 to 0.7 as speed is increased from 0.003 m/s to 0.3 m/s. However, the friction coefficients at 400 and 800°C decrease as the speed increases.
Copyright © 1994 CRC Press, LLC
Volume III
109
FIGURE 2. Wear constants for 99.7 alumina at different temperatures and sliding speeds under a load of 10N. (From Woydt, M. and Habig, K. H., Tribol. Int., 89, 75, 1989.)
Zirconia Ceramics Zirconia ceramics are an important class of materials characterized by high strength and toughness at room temperature. Their major limitation in tribological service is a low thermal conductivity which causes wear by thermal shock at high sliding speeds. Pure zirconia exists in three crystal structures: monoclinic, tetragonal and cubic. The monoclinic phase is stable up to about 1170°C, where it transforms to the tetragonal phase. The tetragonal phase is stable up to 2370°C, where it transforms to the cubic phase. The microstructure of zirconia ceramics can be controlled by addition of various oxides such as MgO, CaO, Y2O3, and CeO2. Additive amount and thermal processing can be chosen such that the tetragonal and cubic phases become stable at room temperature. Zirconia ceramics used in technical applications are classified into three types: cubic, partially stabilized, and tetragonal zirconia. Cubic zirconia is obtained by fully stabilizing the high temperature cubic phase by addition of about 10 mol% oxides. Its relatively low fracture toughness and strength (Table 2) prevents its use in certain tribological applications. Partially stabilized zirconia, PSZ, has a two-phase structure consisting of cubic grains with tetragonal and/or monoclinic precipitates, depending on the thermal processing history. It exhibits increased fracture toughness and is therefore of importance in structural applications. The compressive stress associated with an increase in volume in the transformation of metastable tetragonal precipitates to the monoclinic phase reduces the stress at an advancing crack tip and results in a high strength and toughness. Typical commercial PSZ materials contain about 8 mol% MgO or CaO and have a composition of about 58% cubic, 37% tetragonal, and 5% monoclinic. Tetragonal zirconia polycrystal, TZP, is made by addition of about 2 to 3 mol% Y2O3 or CeO2 to stabilize the tetragonal phase. This material is nearly 100% tetragonal at room temperature and exhibits the highest toughness and strength among zirconia ceramics and other monolithic structural ceramic materials (Table 2). The toughening mechanism is similar to that for PSZ, namely, a tetragonal to monoclinic transformation under stress. TZP materials Copyright © 1994 CRC Press, LLC
110
CRC Handbook of Lubrication and Tribology
are only suitable for tribological applications at room temperature because of severe degradation and decrease in strength at temperatures above 200°C. Typical unlubricated data for zirconia in Table 4 indicate that at room temperature tribological performance of tetragonal zirconia is superior to that of either cubic or partially stabilized zirconia. PSZ materials are much better than cubic zirconia ceramics with respect to strength and wear resistance. The friction coefficient of TZP is lowest at room temperature when slid against alumina; the wear constant is smallest with either a silicon nitride or a silicon carbide counterface. The friction coefficient of TZP tends to increase with increasing load or speed. In contrast, the friction coefficient of PSZ tends to be lower at higher speeds. The friction coefficient of PSZ-steel sliding couple is lower than a self-mated PSZ combination. Also, the coefficient of friction in a PSZ-steel combination is reduced as the contact load is increased. Wear constants for a PSZ material are given in Figure 3 for a normal load of 10 N.23 As the sliding speed or temperature increases, the wear constant increases. However, the friction coefficient decreases as either speed or temperature is increased.
Silicon Nitride Ceramics Silicon nitride, one of the strongest structural ceramics, has emerged as an important tribological material, especially in rolling applications. It has excellent oxidation resistance due to a protective surface oxide layer, very good thermal shock resistance because of its low thermal expansion coefficient, low elastic modulus, high strength, and outstanding fatigue resistance. Silicon nitride does not melt, but decomposes in air at temperatures above 1900°C. In oxidizing environments, silicon nitride is stable only at very low partial pressures of oxygen; in air, it rapidly forms a silicon oxide surface layer. This layer is protective against further oxidation; if it is removed, for example by wear, oxidation occurs rapidly. Copyright © 1994 CRC Press, LLC
Volume III
111
FIGURE 3. Wear constants for a PSZ at different temperatures and sliding speeds under a load of 10 N. (From Woydt, M. and Habig, K. H., Tribol. Int., 89, 75, 1989.)
Pure silicon nitride, Si3N4, exists in two crystallographic forms: alpha and beta, both having a hexagonal crystal structure. Since the beta phase is thermodynamically more stable, silicon nitride materials are primarily in the beta phase; but starting powder is usually in the alpha phase. Commercial silicon nitride materials are processed with various oxide sintering aids. Silicon nitride materials are classified according to processing techniques: sintered, hotpressed, reaction-bonded (or reaction-sintered), sintered-reaction-bonded, and hot isostatically pressed. Variations in composition, microstructure, and properties depend on the processing route used in fabrication. Silicon nitride powder compacts can be sintered to full density using combinations of rare earth oxides and alumina sintering aids. However, mechanical properties of commercially available sintered silicon nitrides are inferior to those processed by hot-pressing (Table 2), usually containing MgO or Y2O3 sintering aids. Application of pressure during sintering is instrumental in achieving nearly full density and very good properties. Shapes that can be formed by hot-pressing and hot-isostatic-pressing (HIP), however, are limited and processing cost is relatively high. Reaction-bonded silicon nitride is made by pressing pure silicon powder and reacting the preform with nitrogen at high temperatures. While they are much less expensive than hotpressed or sintered materials, their porosity, which is greater than 10%, results in inferior mechanical properties (Table 2). Adding oxide sintering aids to the starting silicon powder and a subsequent sintering step of hot-pressing, or hot-isostatic-pressing, reduces this porosity and improves the properties. The principal advantage of reaction-bonded silicon nitride is its lower cost of starting powder. Tribological performance and mechanical properties of silicon nitride depend on composition and microstructure, as well as on the processing procedure and types of starting powders. Table 5 lists typical unlubricated friction and wear data for different types of silicon nitrides. The wide scatter, because of the dependence of performance on mechanical properties and test conditions, hampers a clear distinction between performance of different types of silicon nitrides. The data, however, show that replacing the silicon nitride counterface with steel or alumina slightly increases friction, and a zirconia counterface can decrease the friction coefficient. The wear constant increases rapidly as the temperature is increased, especially above 800°C. Also, the wear constant is increased as load is increased. The wear transition diagram for a hot-isostatically-pressed silicon nitride in self-mated unlubricated tests in air is shown in Figure 4.24 Transition boundaries (the cross-hatched area) Copyright © 1994 CRC Press, LLC
112
CRC Handbook of Lubrication and Tribology
FIGURE 4. Wear transition diagram for HIP’ed silicon nitride at 0.0014 m/s sliding speed. Friction coefficient, f, and wear coefficient, K, for each region are indicated on the figure. (From Dong, X. and Jahanmir, S., Wear, in press.)
Copyright © 1994 CRC Press, LLC
Volume III
113
FIGURE 5. Wear constants for three different types of silicon nitride tested at various temperatures and sliding speeds under a load of 10 N. (From Skopp, A., Woydt, M., and Habig, K. H., Tribol. Int., 23, 189, 1990.)
for the five regions are much wider than those observed for alumina. At low loads and relatively low temperatures in region I, the tribological behavior is controlled by formation of silicon hydroxide on the wear track, with a friction coefficient of 0.30 and a wear coefficient of approximately 10-4. In region II, selective oxidation of WC inclusions controls the wear behavior. The formation of crystalline precipitates from amorphous magnesium silicate grain boundary phase controls the wear process in region III, where both the friction coefficient and wear coefficient increase. The behavior in region IV is dominated by oxidation of silicon nitride; the friction coefficient is approximately 0.70, and the wear coefficient increases to 10-2. In region V, similar to alumina, microfracture is the primary wear mechanism. The friction coefficient in this region is approximately 0.80 and the wear coefficient is high at 10-2. The wear transition diagram can be used for determining the useable range of conditions for a given material, and for prediction of tribological behavior. Wear constants for three types of silicon nitrides are compared in Figure 5 for a test load of 10 N.25 These data show that the three different types of silicon nitrides, i.e., sintered, HIP’ed, and HIP’ed-reaction-bonded, give similar performance. The wear constant increases as temperature is increased, but decreases at higher speeds. The friction coefficient increases with increasing temperature at higher speeds.
Sialon Ceramics Sialons, solid solutions of Si, Al, O, and N with the beta silicon nitride crystal structure, are usually made by adding AlN, MgO, BeO, Y2O3, or other metal oxides to silicon nitride. Since added metal cations cause lattice distortion in the beta silicon nitride structure, these solid solutions are sometimes referred to as beta prime silicon nitride. Most mechanical and physical properties of sialons are intermediate to those of silicon nitride and alumina (Table 2). The primary advantage of sialons is lower processing cost than for silicon nitride, since densification can be achieved by pressureless sintering at lower temperatures. Table 6 lists some typical values of friction coefficients and wear constants for sialon ceramics in unlubricated sliding tests. Tribological performance of sialons, in general, is similar to those of silicon nitrides. The data in Table 6 suggest that the friction coefficient of steel sliding against sialon may be lower than that of silicon nitride sliding against sialon. The wear constant of sialon increases as temperature is increased.
Copyright © 1994 CRC Press, LLC
114
CRC Handbook of Lubrication and Tribology
Silicon Carbide Ceramics Silicon carbide ceramics are widely used in applications requiring wear resistance, high hardness, retention of mechanical properties at elevated temperatures, and resistance to corrosion and oxidation. The oxidation resistance is due to a protective SiO2 surface layer, as with silicon nitride. The thermal shock resistance is good, somewhat lower than that of silicon nitride. Silicon carbide, SiC, exists in hexagonal and cubic crystallographic forms, termed alpha and beta, respectively. Silicon carbide ceramics are grouped into four types depending on processing methods: reaction-bonded (or reaction-sintered), hot-pressed, sintered, and CVD (chemical vapor deposition). In the reaction bonding process, a mixture of silicon carbide powder, graphite, and a plasticizer is pressed in a mold to prepare a preform or “green” compact. After the plasticizer is burned off to produce a somewhat porous product, silicon metal is infiltrated into the pores as a liquid or vapor. The reaction between silicon and carbon to form SiC is not complete, leaving some residual Si and C; usually excess Si is used to fill the pores. The finished product has little porosity, and contains a mixture of Si, C, and reaction-formed SiC in between the original SiC particles. Mechanical properties depends on the amount of free Si and C. Since the densification process does not produce shrinkage, dimensional tolerances are more easily achieved than with other processes such as sintering. The primary advantage of this type of silicon carbide is a relatively low cost, since components can be made to near net-shape, with little machining required after densification. Hot-pressing is used to produce high strength silicon carbide of nearly full density. In this process, boron and carbon and sometimes alumina are used as sintering aids for processing of both alpha and beta silicon carbide components. Although this type of silicon carbide exhibits very good mechanical properties (Table 2), its use is limited by the high cost of finished components, due to difficulty in machining after densification. Silicon carbide components are also produced by sintering without the application of pressure, using carbon and boron sintering aids. The major advantage of this process is that most of the machining can be easily done on the green compact. The densified component is then finish machined by diamond grinding and polishing. CVD is used to produce a relatively pure and dense silicon carbide. This material is highly anisotropic, due to the columnar structure developed during the deposition process. In addition
Copyright © 1994 CRC Press, LLC
Volume III
115
to anisotropy, high cost and residual stresses are major drawbacks against widespread use. Nevertheless, CVD SiC is an excellent coating material where resistance to wear, erosion, and oxidation is required. Typical unlubricated friction coefficients and wear constants for several types of silicon carbide are listed in Table 7. Performance and properties are sensitive to the processing conditions and microstructure of the material. In general, the friction coefficient of reaction— bonded silicon carbide is lower than that of other types because the excess carbon can act as a solid lubricant. However, this effect depends on the specific microstructure and amount of free carbon, as well as test conditions. The friction coefficient for the sintered materials decreases with increasing temperature and speed. Both friction coefficient and wear constant for hot-pressed silicon carbides are lower in sliding against alumina and zirconia than sliding against silicon nitride and silicon carbide. The wear transition diagram for a self-mated sintered silicon carbide in Figure 6 shows four regions.26 At room temperature, high loads, and high relative humidities, friction coefficient is approximately 0.23 and wear coefficient is 10-3. In region I, tribochemical reaction between water vapor and silicon carbide controls the tribological behavior. As the humidity decreases from about 70 to 30%, friction coefficient increases to 0.70. In region II, wear occurs by plowing and the friction coefficient is 0.63, irrespective of humidity. Both friction and wear decrease in region III due to oxidation of silicon carbide and formation of cylindrical rolls on the wear track. At high loads in region IV, wear occurs by microfracture resulting in a high friction coefficient and wear coefficient. There is only a slight difference in the behavior of wear debris in the two parts of region IV. At low temperatures, the wear debris appear loosely attached to the surface; whereas at higher temperatures, some of the debris form compacted regions. Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
116
FIGURE 6. Wear transition diagram for sintered silicon carbide at 0.0014 m/s sliding speed. Friction coefficient, f, and wear coefficient, K, for each region are indicated on the figure. (From Dong, X. and Jahanmir, S., Tribol. I Int., in preparation.)
Results for two reaction-bonded silicon carbides containing different amounts of Si, under a normal load of 10 N,23 have shown that the friction coefficient is sensitive to test conditions. For example, one material exhibited a friction coefficient of 0.3 at room temperature and low speeds, whereas the other material exhibited a value of 0.6 under the same conditions. The wear constant for both materials increased as the temperature was increased; at room temperature, the wear constant increased with increasing sliding speed.
DESIGN CONSIDERATIONS
Implementation of ceramics in engineering design requires experience in designing with brittle materials, design data, field performance data, and performance and failure prediction methods. A key problem with ceramics is their brittleness or low resistance to fracture. The low fracture toughness of ceramics makes it imperative to avoid excessive tensile stresses, and especially stress concentrations due to sharp corners and reduced cross sections. Approaches being considered to improve fracture resistance include transformation toughening, development of ceramic-matrix composites, and control of glassy grain boundary phase.1,27 While these methods have improved fracture toughness, they have also introduced new fabrication problems and challenges. Nonuniform distribution of reinforcing fibers in the ceramic matrix is one such problem. Because of their brittleness, ceramics cannot tolerate imperfections in the form of inclusions, porosity, and fiber agglomerations. Lack of proper control of these microstructural imperfections leads to unreliable and unpredictable performance. This necessitates advances in nondestructive evaluation (NDE) techniques; most present techniques have been developed for inspection of metallic components which can tolerate larger flaws.
TRIBOLOGICAL APPLICATIONS
Ceramics are presently used in diverse applications28,29 such as precision instrument bearings and cutting tool inserts. Applications under development include prosthetic articulating joints and engine components. Each specific application capitalizes on a unique property or a set of specific properties. Tribological applications of ceramics can be divided Copyright © 1994 CRC Press, LLC
Volume III
117
into five categories based on their properties: (1) resistance to abrasion and erosion, (2) resistance to corrosive wear, (3) wear resistance at elevated temperatures, (4) low density, and (5) electrical, thermal, and magnetic properties.
Resistance to Abrasion and Erosion Resistance of materials to abrasive wear is generally related to their hardness. Therefore, a simple solution for abrasive wear is to increase hardness of the component, normally by hard surface coatings or by selection of hard materials. In solid particle erosion, experimental data and field experience show that erosion resistance at grazing angles is also related to hardness. Because of their high hardness, ceramics are well suited for applications requiring resistance to abrasive and erosive wear. Alumina and PSZ, for instance, are used for thread guides and process knives in textile fiber processing; and silicon carbide is used in rocket nozzles, and spray drying and sandblasting nozzles. Chromium carbide and tungsten carbide coatings are applied to turbine blades in jet engines and steam turbines. Silicon carbide and sialon are used as seals, bearings and bushings for slurry and particulate handling equipment.
Resistance to Corrosive Wear While ceramics are not totally inert, they are generally more resistant to chemical reactions and degradation than metals. This is of particular importance in the process industry. PSZ, silicon carbide, and silicon nitride are used as pump sleeves, seals, bushings, and valve components in chemical process industry to combat corrosive wear in harsh environments. In some applications, these components must operate at temperatures as high as 850°C. Perhaps the most common usage of ceramics is in flow control operations where soda lime silica glass, borosilicate glass, PSZ, alumina, or silicon carbide balls are used in check valves. PSZ tips of ball point pens resist abrasive wear by the paper, as well as corrosion by the ink. Other applications such as diesel injector needle valves and seals for coal particle slurry pumps require resistance to abrasive/erosive wear and corrosion resistance.
Wear Resistance at Elevated Temperatures Ceramics are currently used in metal forming and high speed metal cutting operations.30 In high speed machining, temperature at the cutting point can reach 1000°C. Therefore, high hardness, fracture strength, and wear resistance at these temperatures are required. Examples of high-speed cutting tool inserts include: sialon, TiB2, PSZ, silicon nitride, and composites such as SiC whisker-reinforced alumina, Al2O3/B4C, and Al2O3/TiC. Silicon nitride, PSZ, and sialon are also used in metal forming operations such as extrusion, drawing, bending, tube expanding, and others. These operations require high strength and wear resistance, sometimes at high temperatures. An important potential application for advanced ceramics is in internal combustion engines.31Tribological components presently under development are cylinder liners, piston rings, valves, valve seats, valve guides, tappet inserts, wrist pins, cam followers, and rocker arms. Key technical problems involved in the development of ceramics for engines are thermal shock during cooldown and warmup, control of friction and wear at high temperatures, reliability, adequate NDE techniques, and durability. Some of these problems are expected to be overcome through research and development activities currently in progress. Low Density Ceramic rolling element bearings are under development for use in abrasive, corrosive, and high temperature environments.32,33 One particular application is in the main bearing for gas turbines.32 An important advantage of ceramic materials, besides high strength and resistance to abrasion and corrosion, is their low density, which reduces the centrifugal force and skidding of balls and rollers in ultrahigh speed operations. This has allowed speeds of up
Copyright © 1994 CRC Press, LLC
118
CRC Handbook of Lubrication and Tribology
to 100,000 rpm under high thrust or radial loads at temperatures reaching 1000°C. All- ceramic bearings and hybrid bearings (ceramic balls with steel races) based on silicon nitride are in commercial use.33 The high strength and fatigue resistance of silicon nitride, together with the low coefficient of thermal expansion which minimizes distortions and thermal stresses, allow a longer life than conventional bearings. An example of a potential application for silicon nitride bearings is in machine tool spindles where a 30% increase in rotational speed is possible without a substantial design change Another example is in prosthetic articulating joints,34 where alumina is used because of its low weight, excellent wear resistance, and biocompatibility. The low density of ceramics is also exploited in automotive turbochargers to reduce turbolag. Other examples in which the low density of ceramics is advantageous include exhaust cones in rocket engines and tribological components (bearings, seals, and bushings) for space applications.
Thermal, Electrical, and Magnetic Properties The unique thermal, electrical, and magnetic properties of technical ceramics are of great importance in certain applications. For example, ceramic cylinder liners in low-heat rejection engines utilize the low thermal conductivity of ceramics. Ceramic bearings are indispensable in some instruments where the magnetic and electrical properties of other materials would interfere with the instrument operation.35 Recent advances in high temperature superconducting ceramics have renewed interest in the development of bearings based on magnetic levitation. Such concepts would allow continued evolution of new and unconventional tribological components.
REFERENCES
1. Richerson, D. W., Modern Ceramic Engineering, Marcel Dekker, New York, 1982. 2. Archard, J. F. and Hirst, W., The wear of metals under unlubricated conditions, Proc. R. Soc. London, Ser. A, 6, 397, 1956. 3. Evans, A. G. and Marshall, D. B., Wear mechanisms in ceramics, in Fundamentals of Friction and Wear of Materials. Rigney, D. A., Ed., American Society for Metals, Metals Park, OH, 1980, 439. 4. Fischer, T. E. and Tomizawa, H., Interaction of tribochemistry and microfracture in the friction and wear of silicon nitride, Wear, 105, 29, 1985. 5. Jahanmir, S. and Fischer, T. E., Friction and wear of silicon nitride lubricated by humid air, water, hexadecane, and hexadecane +0.5 percent stearic acid, ASLE Trans., 31, 32, 1988. 6. Klaus, E. E., Jeng, G. S., and Dudda, J. L., A study of tricresyl phosphate as vapor delivered lubricant, Lubr. Eng., 45, 717, 1989. 7. Lauer, J. L. and Bunting, B. G., High temperature solid lubrication by catalytically generated carbon, Tribol. Trans., 31, 338, 1988. 8. Archard, J. F., Contact of rubbing surfaces, J. Appl. Phys., 24, 981, 1953. 9. Fischer, T. E., Anderson, M. P., and Jahanmir, S., Influence of fracture toughness on the wear resistance of yttria-doped zirconium oxide, J. Am. Ceram. Soc, 72, 252, 1989. 10. Jahanmir, S. and Dong, X., Mechanism of mild to severe wear transition in alumina, J. Tribol., 114, 403, 1992. 11. Kim, H., Shin, D., and Fischer, T. E., Mechanical and chemical aspects in the wear of alumina, Proc. Jpn. Int. Tribol. Conf.. Nagoya, Japan, 1990, 1473. 12. Ajayi, O. and Ludema, K. C., Formation of transfer film during ceramics/ceramics repeated pass sliding, Wear of Materials, Ludema, K. C., Ed., American Society of Mechanical Engineers, New York, 349, 1989. 13. Fischer, T. E., Anderson, M. P., Salher, R., and Jahanmir, S., Friction and wear of tough and brittle zirconia in nitrogen, air, water, hexadecane and stearic acid, in Wear of Materials, Ludema, K. C., Ed., American Society of Mechanical Engineers, New York, 257, 1987. 14. Hsu, S. M., Lim, D. S., Wang, Y. S., and Munro, R. G., Ceramics wear maps: concept and method development, Lubr. Eng., 47, 49, 1991.
Copyright © 1994 CRC Press, LLC
Volume III
119
15. Kim, S. S., Kato, K., Hokkirigawa, K., and Abe, H., Wear mechanism of ceramic materials in dry rolling friction, J. Tribol, 108, 522, 1986. 16. Shimura, H. and Tsuia, Y., Effects of atmosphere on the wear rate of some ceramics and cermets, in Wear of Materials, Ludema, K- C, Ed., American Society of Mechanical Engineers, New York, 452, 1977. 17. Fischer, T. E., Liang, H., and Mullins, W. M., Tribochemical lubricious oxides on silicon nitride, Proc. Mat. Res. Soc. Symp., 140, 339, 1989. 18. Gates, R. S. and Hsu, S. M., Effect of selected chemical compounds on lubrication of silicon nitride, Tribol. Trans., 34, 417, 1991. 19. Wachtman, J. B. and Niesz, D. E., Commercial structural ceramics, in Handbook of Structural Ceramics. Schwartz, M., Ed., McGraw Hill, New York, 1992, chap. 3. 20. Leatherman, G. L. and Katz, R. N., Structural ceramics: processing and properties, in Superalloys, Supercomposites and Superceramics, Academic Press, New York, 1989, 671. 21. Stevens, R., Zirconia and Zirconia Ceramics, Magnesium Elektron Publ. No. 113, Magnesium Elektron Ltd., Manchester, U.K., 1986. 22. Dong, X. and Jahanmir, S., Tribological characteristics of alumina at elevated temperatures, J. Am. Ceram. Soc, 74, 1036, 1991. 23. Woydt, M. and Habig, K. H., High temperature tribology of ceramics, Tribol. Int. 89, 75, 1989. 24. Dong, X. and Jahanmir, S., Wear transition diagram for silicon nitride, Wear, 165, 169, 1993. 25. Skopp, A. Woydt, M., and Habig, K. H., Lubricated sliding friction and wear of various silicon nitride pairs between 22 and 1000°C, Tribol. Int. 23, 189, 1990. 26. Dong, X. and Jahanmir, S., Wear transition diagram for silicon carbide, Tribol. Int., to be published. 27. Clark, D. E., Ed., Ceramic Engineering and Science Proceedings, American Ceramic Society, Westerville, OH, 1988. 28. Jahanmir, S., Tribology of Ceramics. Vol. 1, Fundamentals. Special Publication S-23, and Vol. 2, Applications. Special Publication S-24, Society of Tribologists and Lubrication Engineers, Park Ridge, IL, 1987. 29. Jahanmir, S., Friction and Wear of Advanced Ceramics. Marcel Dekker, New York, 1993. 30. Machining Issue, Ceram. Bull., 67(6), 991, 1988. 31. Larsen, R. P. and Vyas, A. D., The outlook for ceramics in heat engines, 1900–2010, SAE Pap., No. 880514. 32. Zaretsky, E. V., Ceramic bearings for use in gas turbine engines, ASME Paper No. 88-GT-138. 33. Katz, R. N. and Hannoosh, J. G., Ceramics for high performance rolling element bearings: a review and assessment. Int. J. High Tech. Ceram. 1, 69, 1985. 34. Davidson, J. A. and Schwartz, G., Wear, creep and frictional heat of femoral implant articulating surfaces, J. Biomed. Mat. Res. 21(A3), 261, 1987. 35. Jahanmir, S., Ceramic Bearing Technology, NIST Spec. Publ. 824, National Institute of Standards and Technology, U.S. Department of Commerce, Gaithersburg, MD, 1991. 36. Gangopadhyay, A. and Jahanmir, S., Friction and wear characteristics of silicon nitride-graphite and aluminagraphite composites, Tribol. Trans., 34, 257, 1991. 37. Yust, C. S. and Carignan, F. J., Observation on the sliding wear of ceramics, ASLE Trans. 28, 245, 1984. 38. Usami, H., Funabashi, K., Nakamura, T., and Mabuchi, E., Friction test of ceramics, Jpn. J. Tribol., 35, 347, 1990. 39. Chen, Y. M., Rigaut, B., and Armanet, F., Friction and wear of alumina ceramics at high sliding speeds, Tribol. Trans., 41, 531, 1991. 40. Ajayi, O. and Ludema, K. C., Surface damage of structural ceramics. Wear of Materials, Ludema, K. C, Ed., American Society of Mechanical Engineers, New York, 1987, 349. 41. Denape, J. and Lamon, J., Sliding friction of ceramics, J. Mater. Sci., 25, 3592, 1990. 42. Gee, M. G., Matharu, C. S., Almond, E. A., and Eyre, T. S., The measurement of sliding friction and wear of ceramics at high temperature, Wear. 138, 169, 1990. 43. Libsch, T. A., Becker, P. C., and Rhee, S. K., Friction and wear of toughened ceramics against steel, Proc. JSLE Int. Tribol. Conf., Tokyo, 1985, 185. 44. Erdemir, A. Busch, D. E., Erck, R. A., Fenske, G. R., and Lee, R., Ion-beam-assisted deposition of silver films on zirconia ceramics for improved tribological behavior, Lubr. Eng., 47, 863, 1991. 45. Czichos, H., Becker, S., and Lexow, J. L., International multilaboratory sliding wear tests with ceramics and steel. Wear, 135, 171. 46. Cranmer, D. C., Friction and wear properties of monolithic silicon-based ceramics, J. Mater. Sci., 20, 2029, 1985. 47. Park, D. S., Danyluk, S., and McNallan, M., Friction and wear measurements of silicon nitride at elevated temperatures in air, Ar, and humid environments, in Proc. Int. Conf. Corrosion Degradation Ceramics, American Ceramic Society, Westerville, OH, 1989, 159. 48. Wang, H., Kimura, Y., and Okada, K., Sliding friction of ceramics at elevated temperatures up to 1000°C, Proc. Jpn. Int. Tribal. Conf., 1990, 1389.
Copyright © 1994 CRC Press, LLC
120
CRC Handbook of Lubrication and Tribology
49. Scott, H. G., Friction and wear of zirconia at very low sliding speeds. Wear of Materials, Ludema, K. C., Ed., American Society of Mechanical Engineers, New York, 1985, 8. 50. Mukerji, J., Bandopadhyay, S., Wani, M. F., and Parkash, B., Friction and wear behavior of hot pressed sialon sintered without externally added liquid, Proc. Jpn. Int. Tribol. Conf., 1990, 1401.
Copyright © 1994 CRC Press, LLC
PLASTICS AND PLASTIC MATRIX COMPOSITES Warren E. Jamison
INTRODUCTION
The unique properties of polymeric materials have caused a worldwide change in the way we make and use tribological components like gears and bearings, as well as the more mundane items such as furniture and milk bottles. Attributes which make plastics attractive as tribological materials can be associated either with their intrinsic physical properties or with their manufacturability as listed in Table 1. Under similar conditions, most plastics have lower friction than metals or other structural materials. They also resist galling and scuffing. These properties allow them to be used without additional lubrication for many applications. Some, but not all, plastics respond to added oil or grease lubrication for further reduction in friction and wear. While the low friction tends to provide cooler running of machine parts, the poor thermal conductivity of plastics makes it difficult to dissipate frictional heat generated in loaded contacts. Plastics have excellent low temperature properties and are frequently used at temperatures where greases and oils would solidify and fail to lubricate. The resilience and toughness of plastics give them excellent resistance to shock and fatigue and the ability to damp out machine noises. Their deficiencies for tribological machine parts, their lower maximum operating temperatures, their poor capability to maintain tolerances and their lower hardness can be offset somewhat by compounding them with various stiffeners and fillers. Metals are somewhat less costly than plastics on a weight basis, but their density of less than 1/4 than of metals commonly makes plastic parts much cheaper than metal parts of the same size. In addition, many plastic parts are efficiently made to finished size and shape by low cost injection or compression molding. If thermoplastic resin systems are used, scrap and reject parts can often be reprocessed, further increasing the efficiency of material usage. The friction and wear properties are improved for many plastics by compounding the base polymer with solid or liquid additives such as polytetrafluoroethylene (PTFE) and MoS2 powders in polyamide (nylon) and polyimide, and dispersion of oils through ultrahigh molecular weight polyethylene (UHMWPE), polyamide (nylon) and polyoxymethylene (acetal). Powders of PTFE are mixed with powders of other polymers, sometimes thermosetting resins, and compacted and sintered to form a self-lubricating solid. In other cases, PTFE fibers are incorporated either as a woven fabric or as random dispersions in compression-molded plastics. Composite bearings with unique properties are sometimes created by combining polymers with superior tribological properties and metals and ceramics. In this chapter the properties of the base polymers of tribological interest are described, along with representative examples of polymers incorporating performance-enhancing additives.
NATURE OF PLASTIC MATERIALS
It is convenient to divide plastic materials of tribological interest into three classes: thermoplastic polymers, thermosetting polymers, and elastomers. Thermoplastics can be repeatedly heated to a state of softness where they can be reshaped under low pressure without degrading the molecular structure. Because the temperatures at which they soften are usually below 150 to 200°C, thermoplastics such as polyethylene and nylon are usually restricted from applications involving hot machinery. Copyright © 1994 CRC Press, LLC
121
122
CRC Handbook of Lubrication and Tribology
Thermosetting polymers are made with resins that crosslink and form chemical bonds between the molecules comprising their structure. This provides greater hardness and strength, and also higher temperature capability in many cases. However, once reacted to form a solid structure, thermosetting plastics cannot be melted and reformed without major property degradation. Elastomers can be either thermoplastic or thermosetting, although most tribological applications involve the thermosets. Their great flexibility allows them to stretch, flex, and change dimensions with great resiliency. Although rubber bearings are employed in marine applications where water is the lubricant, elastomers are seldom used in bearing applications because their flexibility prohibits them from maintaining accurate positions under load. Their primary application is in sealing structural components against fluid leakage.
THERMOPLASTIC POLYMERS—STRUCTURE AND PROPERTIES
Thermoplastic polymers can be melt-processed to final shape and dimensions with great accuracy. This section will deal only with single component polymeric materials, that is, the base polymers whose properties have not yet been modified through the addition of powders or liquids, or through alloying with other polymers. Table 2 shows the molecular structure and lists a few basic properties of the base polymers of major tribological interest. Because plastics almost always comprise polymers with a range of molecular weights, they usually do not have a discrete melting point. Instead, they soften and start flowing with a viscosity which decreases with increasing temperature. Approximate melting temperatures, Tm, for the most common molecular weight members of the polymeric structures are shown in Table 2. In the case of polymers which can align their chains into crystalline domains, the glass transition temperature, Tg, defines the temperature above which crystallinity disappears.
Polyethylene (PE) Polyethylene is the prototype polymer with the simplest molecular structure. Low molecular weight polyethylenes are easily melt-processed and are widely used in lightly loaded sliding contacts such as furniture feet and drawer guides. Higher molecular weight polyethylenes have superior mechanical properties and are used in engineering applications. Table 2 shows that polyethylene is composed of a chain of carbon atoms filled by attachment to hydrogen atoms. Two variations in the structure account for the wide range of properties found in polyethylenes: chain length (molecular weight) and chain branching. Also, the manner of processing can alter some physical properties. For example, cooling a molten mass of polyethylene will produce a solid with random arrangement of entangled polymer molecules, while extrusion of the melt through a small die under high pressure will preferentially orient the molecules in the direction of flow.
Copyright © 1994 CRC Press, LLC
Volume III
Copyright © 1994 CRC Press, LLC
123
CRC Handbook of Lubrication and Tribology
124
FIGURE 1. Crystalline domains of PE formed by alignment of linear molecule segments.
A mass of polyethylene can be visualized as a tangled array of intertwined molecular chains. If the chains are linear, they are apt, in some cases, to align themselves into neat, dense packed crystal domains, as shown in Figure 1. Linear polyethylene has a high density due to the ease with which the chains can pack tightly together. Figure 2 indicates that the higher density increases both hardness and stiffness. If branching exists, the side chains make it more difficult to pack the molecules together, as shown in Figure 3. Branched molecules have lower density, but remain more flexible due to the extra space within the polymer for the molecules to slightly shift their positions under external forces. Three distinctly different states of ordering exist within polymers. The amorphous state contains purely random intertwined molecules. The crystalline state comprises a neatly arranged dense packing of molecules into crystal domains. The oriented state occurs when the polymer processing arranges the molecules into a partially ordered structure. Copyright © 1994 CRC Press, LLC
Volume III
125
FIGURE 2. Stiffness and hardness of PE as a function of density.
FIGURE 3. The intermolecular distance, d, in PE increases as the side chains grow longer, making packing difficult and decreasing the density.
The characteristics of polyethylenes of various chain lengths are shown in Figure 4. Low molecular weight molecules are gases or liquids. It takes a chain length of about 1500 CH2 units (mol wt = 20,000) for polyethylene to be considered a plastic. Polyethylenes up to about 500,000 mol wt can be melt-processed by injection molding and extrusion. Polyethylenes between 500,000 and about 1,000,000 mol wt are considered “very high molecular weight”. Materials up to 5,000,000 mol wt are called ultrahigh molecular weight polyethylenes and are processible only by compression molding and ram extrusion. These materials exhibit very low friction and wear and are extremely tough.
Polypropylene (PP) As Table 2 shows, polypropylene is essentially polyethylene in which a hydrogen atom on every second carbon atom has been replaced with a CH3 unit (a methyl group). This stiffens the polymer chain and gives it a higher melting point, but reduces the crystallinity for a given molecular weight. Polypropylene is not widely used for tribological applications. Copyright © 1994 CRC Press, LLC
126
CRC Handbook of Lubrication and Tribology
FIGURE 4. Effect of molecular weight and crystallinity of PE upon mechanical properties. (From Turner, A. I and Gurnee, E. F., Organic Polymers. Prentice-Hall, Englewood Cliffs, NJ, 1967, 245.)
Polyvinyl Chloride (PVC) Polyvinyl chloride, as shown in Table 2, comprises a substitution of a chlorine atom for a hydrogen atom on the polyethylene chain. PVC has little tribological significance, except as a prototype for polyfluorocarbons.
Polyfluorocarbons (PVF, PTFE, PCTFE, PVDF) Polyvinyl fluoride (PVF) is a direct analog of PVC, with substitution of a fluorine atom for a single carbon atom in the PE chain, as seen in Table 2. Polytetrafluoroethylene (PTFE) is perhaps the slipperiest of all polymers with fluorine atoms substituted for all four hydrogen atoms on polyethylene. The fluorine atom is quite large compared with the hydrogen atom and, like chlorine and hydrogen, has a single electron available for chemical bonding. When four fluorine atoms are substituted for the hydrogens on polyethylene, the fluorine atoms almost completely cover the carbon atom chain. Since the sole bonding electron on the fluorine atom is used to tie it to the carbon chain, the result is a sausage-like molecule with a completely inert surface. The slippery molecules of PTFE provide low friction against other surfaces and resist bonding to each other. Thus, PTFE cannot be melt-processed and it cold-flows under pressure. To take advantage of its outstanding frictional properties, PTFE is generally used as a powder additive to other polymers, or is mechanically constrained and supported by compressing it with fabric mesh or other structural materials. A lesser substitution of only two fluorine atoms for carbons produces polyvinylidine fluoride (PVDF). It can be melt-processed and has tribological properties better than PVC, but significantly inferior to PTFE. The substitution of three fluorine atoms and one chlorine atom for the four hydrogens on polyethylene produces polychlorotrifluoroethylene (PCTFE). This polymer and a copolymer of it with PVDF are manufactured under the trade name “Kel-F”. With only partial coverage of the carbon chain with fluorine atoms, the molecules are able to bond to each other and to allow melt-processing. The materials have good tribological properties which can be enhanced with additives.
Polyoxymethylene (POM—Acetal) The acetals have a highly regular structure, with oxygen atoms alternating with CH2 groups. The regularity of this structure allows dense packing and high crystallinity. Acetals are dimensionally stable, do not absorb water, and have low friction against themselves and against other materials. These properties, along with the ease of melt-processing, make the Copyright © 1994 CRC Press, LLC
Volume III
127
acetals widely used for gears, bushings, and other tribological components, both in the natural state and with fillers such as PTFE and MoS2.
Polyamide (PA—Nylon) There are many different types and grades of the nylons. The structures of two of the simpler types is given in Table 2. The nylons are described by the number of carbon atoms in the monomer chains. Amino acid polymers have a single number, as in nylon 6. Nylons from diamines and dibasic acids are designated by two numbers, the first representing the number of carbon atoms in the diamine segment and the second the atoms in the acid, as in nylon 6/10. The nylons are highly crystalline materials with stable properties and good processability. Although their friction coefficients are high, they respond well to low levels of lubrication and are extensively used for gears, bushings, etc. Nylons are resistant to chemical attack, but absorb moisture from their surroundings, which causes them to swell slightly.
Polycarbonate (PC) The regularity of the polycarbonate molecule makes it easily crystallizable. This factor and interaction between adjacent molecules through the phenyl groups gives the plastic high strength, impact resistance, and thermal stability, Although polycarbonates do not have good inherent friction and wear properties, they are frequently used in applications such as appliance handles, with secondary tribological contacts.
Polyetherether Ketone (PEEK) The search for tribological polymers with higher temperature capabilities has prompted the use of PEEK. This material derives its thermal and dimensional stability from its unique structure shown in Table 2. The C=O group adds stiffness, and the ether linkage, -O-, in the chain gives modest flexibility. PEEK can be melt-processed by conventional methods in spite of its high thermal stability. It is not used alone and requires fillers as a bearing material.
Polyphenylene Sulfide (PPS) The excellent dimensional stability and thermal resistance of PPS and its easy molding cause it to be used extensively for structural parts which may incorporate integral bearing surfaces. PPS parts usually incorporate fillers and additives to modify their structural and tribological properties.
THERMOSETTING POLYMERS—STRUCTURE AND PROPERTIES
Phenolic (PN) The workhorse of all thermosetting resin systems, phenolics are created by reacting phenol with formaldehyde, with the resulting structure shown in Table 2. Good chemical, electrical, and mechanical properties provide the basis for a wide variety of components, generally formed by compression molding to net shape and size from a granular feedstock. The higher thermal stability of phenolics over most thermosets suggest a wider application to hot machinery. However, this can be achieved only when the modest tribological properties of phenolics can be upgraded by the addition of lubricating fillers, etc.
Epoxy Resins Epoxy prepolymers are cured with a variety of polyamines, resulting in different molecular structures. A typical epoxy resin is shown in Table 2. With rather poor tribological properties themselves, epoxy resins are usually used as binders to form self-lubricating composites
Copyright © 1994 CRC Press, LLC
128
CRC Handbook of Lubrication and Tribology
with fillers such as graphite and molybdenum disulfide powders and fibers and powders of PTFE.
ELASTOMERS
Elastomers are polymeric materials which can be deformed extensively and then quickly recover their original dimensions, with little loss of energy as heat. Tribological interest in elastomers centers on the low wear and high wet friction of automobile tires, on rubber bearings, and on water and grease seals. Elastomers of major tribological interest are silicones, urethanes and copolymers of butadiene, chloroprene, isoprene and acrylonitrile. The polymer building blocks are listed in Table 3 and typical low temperature properties are given in Table 4. Butyl rubbers (copolymers of isobutylene and isoprene) and neoprene (polymers and copolymers of chloroprene and butadiene) are widely used because of their chemical inertness and resistance to swelling and weakening by hydrocarbons. Nitrile rubber (butadiene and Copyright © 1994 CRC Press, LLC
Volume III
129
FIGURE 5. Elastomeric friction. An elastomer asperity, A’ attaches to a rigid body at A (a). As the elastomer moves to the right, the bond at A-A’ is maintained while the bulk elastomer deforms to allow the motion, (b, c). At some amount of deformation, I’, the bond A-A’ is broken and the elastomer asperity springs forward to reestablish a bond to the rigid body at B (d).
acrylonitrile) is even more resistant to oils. SBR (polymerized from styrene and butadiene) is the principle ingredient of tire treads and shoe soles. Silicone elastomers, either methyl or phenyl polysiloxanes, retain their elasticity over a wide temperature range, but have poor tensile strength. Polyurethanes have excellent abrasion resistance. Although they have few tribological applications at present, the wide property variations available make them strong candidates for replacement of other elastomers for many applications.
FRICTION AND WEAR OF PLASTICS
Friction of Plastics The theory of friction of metals, both lubricated and dry, is developed in Volume II of this Handbook. Difference between the friction of metals and polymers lies in differences in their bulk properties. Metallic friction theory assumes a perfectly elastic process with negligible hysteresis losses. In plastics, not only are the hysteresis losses nonnegligible, but also the plastic deformation of asperities becomes significant in many cases. A simplified theory of polymer friction assumes that asperities adhere to each other, as in the case of metals. It also assumes that there is no “plowing” term: that the asperities are elastically deformed as they slide over each other without plastically displacing material. Figure 5 shows how this event occurs. Adhesion of the asperity at A to the substrate surface persists as the bulk polymer moves through a distance 1. The asperity stores up elastic energy Copyright © 1994 CRC Press, LLC
130
CRC Handbook of Lubrication and Tribology
until the adhesion is released after the stretch to 1’. The asperity then moves forward and reattaches itself to the substrate. The elastic energy is returned to the system except for a small portion which is the hysteresis loss. From the viscoelastic properties of the polymer, the frictional force becomes F = K(so/H)tanβ
where K is a proportionality constant, so is the maximum stress on the asperity, H is the hardness of the polymer, and tanβ is the polymer damping factor. Because of the wide variation in the values of these parameters as affected by polymer type, additive materials, and processing history, predictions of friction are not usually accurate enough for engineering purposes. Therefore, most friction data are obtained experimentally. Table 5 shows friction coefficients for several tribological polymers against steel in different environments.
Wear of Rigid Plastics The wear characteristics of plastics differ from those of metals in several respects. With their lower elastic moduli, plastic parts will undergo more elastic deformation. As in the case of friction, above, the asperities will deform elastically rather than plastically. Because plastics have generally lower mechanical strength and greater ductility, they will deform at lower loads. Thus, plastic tribological components must be designed to accommodate the differences in properties (with lower contact stresses, for example). Another physical property difference which affects wear is the lower thermal conductivity of plastics in comparison with metals. Because frictional heat is more difficult to conduct away from the contact in plastics, and because plastics tend to be operated closer to their softening temperatures, the wear characteristics of plastics are more temperature sensitive than those of metals. Metal powders are sometimes used as fillers to aid in heat dissipation.
Adhesive Wear Adhesive wear results from the attachment of asperities on one surface to a mating surface in relative motion. The asperities and perhaps part of the substrate are torn out and lost from the system as wear debris. Although much is known about the details of the processes, differences in wear rates between candidate plastics are often described in terms of an empirical wear coefficient, k. As with metals, adhesive wear is proportional to the product of the load (F) and the distance traveled (L): W = kFL
where W is the volume of material lost. Using L equal to the product of the velocity (V) and the time (T), the wear coefficient becomes: k = W/FVT
Copyright © 1994 CRC Press, LLC
Volume III
131
Table 6 gives adhesive wear coefficients for a variety of materials. More data are given later for particular plastics. However, because the wear rates are highly dependent on specimen configuration, temperature, environment and other factors, large errors may arise if the end use conditions differ significantly from those under which the data were taken. Wear data are usually measured with one of four types of instruments shown in Figure 6. In both the pin-on-disk and ring-on-block configurations, a portion of one member of the tribological couple passes through the contact zone and is then exposed to the environment. In the thrust washer configuration, continuous contact is made between both members, which dramatically reduces the ability to reject frictionally generated heat, and thus is a more severe test. If end use of the plastic is to be a bushing, friction and wear characteristics are measured with a radially loaded shaft rotating in a bushing. The test most closely approximating Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
132
FIGURE 6, Four types of friction and wear instruments.
the end use and speed, load, temperature, etc. should be used when comparing candidate plastics for any application.
Abrasive Wear Plastics, particularly elastomers, are frequently used for their greater capacity to operate satisfactorily in the presence of abrasive media than metals. Three different types of abrasive situations are recognized: (1) abrasion by loose abrasive particles, (2) abrasion by bonded abrasive particles, and (3) erosion. These types are differentiated as much by the test methods as by the end use application. Although test procedures may be precisely controlled, there is such variability in abrasive particle size and quantity and in other situational factors that a wear coefficient would be meaningless. Therefore, wear data are compared as volume of material lost under standard test conditions. Loose particle abrasion is typically measured with an apparatus in which a specimen is slid in a track on a disk onto which abrasive material is continually dropped. The specimen is raised and lowered periodically to allow abrasive particles to enter the sliding contact. In a standard test for abrasion with bonded abrasives (Taber abrasion test), a pair of abrasive wheels are rotated and slid around a specimen disk. Erosion characteristics are measured either by tumbling dry abrasive against a specimen or swirling them in a liquid slurry. Table 7 shows relative abrasion data from a sand slurry test for several plastics.
PV CHARACTERISTICS
The frictional heat generated during sliding as a tribological performance factor for plastics is defined as the product of the unit load, P, and sliding velocity, V. The unit load is the force, F, divided by the projected area, A, over which the load operates. The “limiting PV factor” is the value under which continuous normal operation can proceed. Two types of Copyright © 1994 CRC Press, LLC
Volume III
133
FIGURE 7. Typical PV curve for tribological plastics.
apparatus are used to measure limiting PV factors: a thrust washer and a radial bushing. For bushings, the area in centimeters squared used in calculating the unit load, P, is the projected area (axial length × bore diameter). The velocity in meters per minute (mpm) for this geometry is given by the formula: V = π n D/100
where D is the bushing diameter in centimeters and n is the shaft speed in revolutions per minute (rpm). The area for thrust washers is π/4 (D22 - D12) where D2 is the outer diameter of the contact and D1 is the inner. Velocity is mpm at the mean diameter D is calculated as V = π n D/100
While metric units for PV are commonly given as N/mm-s, values in the U.S. are frequently stated as psi-fpm. A typical PV chart is shown in Figure 7. The upper limit of P is determined by compressive strength of the plastic, and the maximum V is governed by the thermal stability, thermal Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
134
conductivity, heat generation, and materials configuration. The shape of the curve between these limits is dictated by the allowable wear rate, and will depend on the performance criteria established by the end user.
COMMERCIALLY AVAILABLE MATERIALS
Tribological plastics are available in such a wide variety of forms that documentation of all such materials is impossible. Examples are presented here of unmodified and tribologically enhanced polymers. Examples are also given of composite bearing materials comprising polymers and reinforcements such as woven fabrics and metal backings. The reader is referred to the references and citations in the bibliography for more information. Many materials are available as pellets for extrusion or injection or compression molding by the customer into the final form. Other materials are available only in preformed shapes or finished products. The nature of some material products, such as nylon-impregnated woven PTFE fabric, makes it impossible to supply them in a form which the customer can thermally process to his configurations. Still other materials, such as UHMWPE (which is not meltprocessible) and some thermosetting polymers, are not amenable to customer processing because of the intrinsic nature of the materials. Plastics, such as PE, POM, and PA have inherently good tribological properties and can be used for many applications without performance modifiers. However, the properties of these materials may be further improved for certain applications using various additives. Glass fibers, for example, improve mechanical and thermal properties, but do not improve tribological properties. Figure 8 shows that 30% glass fibers increase the heat deflection temperature of PA, POM and PEEK, but not that of polycarbonate, and provide a modest improvement in continuous use temperature for nylon, polycarbonate, and PEEK, but not for acetal. Other additives, notably PTFE, molybdenum disulfide, and graphite powders, and silicone and mineral oils decrease friction and wear. The combinations of strengtheners and tribological enhancers are countless, and form the basis of most of the commercial tribological plastics.
Polyethylene (PE) High density polyethylene, while not favored for primary tribological components such as bearing and gears because of limitations in hardness, elastic modulus, and molding tolerances, is often used as liners for push-pull cables or for components which must resist mild wear. Typical properties are shown in Table 8. UHMWPE, which provides excellent resistance to abrasion and impact, has an inherently low friction coefficient. However, UHMWPE, not being melt-processible, is only available in the form of sheets and bars made by ram extrusion or similar processes. Very high and ultrahigh molecular weight polyethylenes incorporating various amounts of oil to reduce friction and wear even further are available in pellet form.
Nylon (PA) Nylon is used extensively for gears and other tribological components because of its ease of casting and molding, its high melting point, and its good wear resistance. The property deficiencies of low elastic modulus and high friction coefficient are largely overcome will addition of various fillers and stiffeners. Properties of selected nylons are shown in Table 9. Nylon is resistant to attack by most solvents and organic acids, but absorbs water and swells in humid environments. The moisture absorption decreases with increasing chain length and the percent absorption at 100% humidity for nylon 6/6, nylon 6/9, and nylon 6/12 are 8, 4, and 3. Copyright © 1994 CRC Press, LLC
Volume III
135
FIGURE 8. Continuous use temperatures vs. heat deflection temperature for unreinforced (open hexagons) and 30% glass fiber reinforced (filled circles) plastics. (From Stanyl 46 Nylon, General Information, DSM Engineering Plastics, Reading, PA, April 1992.)
Polyimide and Polyamide-imide Polyimides and polyamide-imides do not have inherently good tribological properties, but they do retain their physical properties at temperatures well above 250°C. Incorporating graphite and PTFE in their structure gives them unique capabilities as high-temperature bearings. They have no discernible melting point and are therefore difficult to process by injection molding. Parts are usually compression-molded or machined from compressionmolded stock. Table 10 gives typical properties of commercially available materials.
Acetal (POM) Acetal has been termed the workhorse of engineering polymers because of its excellent mechanical properties and its ease of injection molding. Table 11 shows typical properties for both the homopolymer and copolymer grades, as well as properties of acetals with enhanced tribological properties. Figure 9 shows that additives can be equally effective in improving the performance of acetal (POM) and nylon (PA). The higher stiffness of acetal over nylon and the resistance to moisture absorption justify the higher material cost for many applications. PTFE PTFE, usually known by its trade name, Teflon, exhibits very low friction against most materials, but has inadequate compressive strength to be used as a bearing without
Copyright © 1994 CRC Press, LLC
136
CRC Handbook of Lubrication and Tribology
Copyright © 1994 CRC Press, LLC
Volume III 137
Copyright © 1994 CRC Press, LLC
138
CRC Handbook of Lubrication and Tribology
Copyright © 1994 CRC Press, LLC
Volume III
139
FIGURE 9. Wear coefficient vs. limiting PV for unfilled and tribologically enhanced polymers.
reinforcement. Therefore, PTFE is usually incorporated into other polymers as an additive, or is reinforced with woven fibers, or used with some other supporting member. Table 12 shows reduction in friction and wear effected by adding 15% PTFE to several polymers. PTFE may also be added to elastomers to improve their abrasion resistance, as Table 13 shows.
BEARING DESIGN
Many of the tribological plastics described in this chapter are commercially available in the form of sleeve and spherical bearings. The sleeve bearings shown in Figure 10a are usually constructed entirely of the tribological materials. Small, inexpensive bearings can be injectionmolded from enhanced nylon, acetal, or other self-lubricating materials. Other bearings are compression-molded from phenolic filled with PTFE or similar proprietary Copyright © 1994 CRC Press, LLC
140
CRC Handbook of Lubrication and Tribology
compositions. Still other sleeve bearings are filament-wound reinforced with woven fabric and impregnated with a tribological plastic. For better thermal conductance and to provide smaller envelope dimensions, steel-backed bearings of the type shown in Figure 10b are also commercially available. These have a relatively thin layer of a self-lubricating plastic material bonded to the bore. Inexpensive selfaligning bearings are made either from solid tribological plastic (Figure 11) or from steel which incorporates a self-lubricating composite liner (Figure 12). General guidelines for the design and selection of bearings follow.
Bearing Clearances Figure 13 shows that too loose a bearing-shaft clearance fit will cause as much excessive wear as too tight a fit. Typical diametral clearances for thick-walled bushings are 0.005 cm/cm of bearing diameter under steady state running conditions. Such bushings are press-fitted into metal housings. Since the metal housing expands less than the plastic bushing, the bushing I.D. will decrease due to thermal expansion at normal elevated operating temperatures. Approximate room temperature bearing-shaft clearances for several plastic bearing materials are shown in Figure 14 as a function of operating temperature. For critical applications, the material supplier should be consulted. Some plastics absorb water and swell when immersed in water or operated in humid environments. Additional clearances must be allowed for such conditions. Figure 15 shows that polyamides (nylons) are the worst offenders. Copyright © 1994 CRC Press, LLC
Volume III
FIGURE 10. Sleeve bearings. (A) Solid plastic bushing; (B) steel-backed sleeve bearing.
141
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
142
FIGURE 11. Simple self-aligning pillow block bearing (a) incorporating spherical inset (b)
FIGURE 12. Self-lubricating spherical bearing.
FIGURE 13. How bearing clearance affects wear.
Surfaces The surface of the plastic bearing must be as smooth as possible. Machining should avoided whenever possible, and the surface should be formed by molding against a high polished surface. The material which rubs against the plastic bearing surface should be as hard as possible to ensure that the bearing wears in preference to the counterface material. The counterface Copyright © 1994 CRC Press, LLC
Volume III
143
FIGURE 14. Recommended minimum clearance for journal bearings at operating temperatures. A, solid polymer bearings; B, metal-backed thin-layer bearings.
FIGURE 15. Swelling of plastics after 24-h immersion in water at room temperature.
surface finish can substantially affect the bearing wear rate. Ground surfaces are preferred and polishing will usually be beneficial. Average roughness values of 0.2 to 0.4 micrometers (8 to 16 microinches) are usually specified, but finer surface finishes will produce lower wear rates. Finishing marks should always be in the direction of motion.
Wall Thickness The walls of plastic bearings should be as thin as possible to help dissipate heat and to reduce distortion due to high loading. Bushings are generally press-fitted into metal sleeves or housings, and the wall thickness must be great enough to support the press fit, unless mechanical retention devices or chemical bonding is used. For thrust bearings, the thickness should be less than one fifth the bearing outer diameter. Copyright © 1994 CRC Press, LLC
144
CRC Handbook of Lubrication and Tribology
Installation Since many thermoplastic materials suffer from creep under load, the usual interference fit method of installing them in metal housings is suitable only for less severe applications. The amount of interference fit will depend on the plastic material; the more rigid materials such as polyacetals should use 0.005 to 0.010 millimeters of interference per millimeter of diameter. Less rigid plastics need 0.010 to 0.020 mm/mm interference. A mechanical restraint or chemical bond is always a good back-up.
Environment Plastics are more resistant to chemical attack than are metals and are primary candidates for use in corrosive or other harmful environments. However, some plastics are more susceptible than others to chemical attack or swelling in certain liquids. Table 14 indicates general suitability for different classes of bearing materials for various environments. More specific data should be obtained from the material supplier.
Heat Generation and Rubbing Speeds Frictional heat generated in a bearing is dissipated through the shaft and housing. The limiting speed for a bearing is dictated by the thermal stability of the plastic and the thermal conductivity of the counterface. Figure 16 shows typical limiting speeds for a variety of Copyright © 1994 CRC Press, LLC
Volume III
145
FIGURE 16. Limiting bearing speeds.
plastic bearings running on a carbon steel shaft. Stainless steel has a lower thermal conductivity and the limiting speeds with stainless shafts should be half those shown. Heat dissipation can be aided by consideration of the following during design: 1. 2. 3. 4. 5.
Large shaft and housing areas exposed to cooling environments Free access of cooling air to the bearing assembly; forced air or internal water cooling Minimum number of mechanical joints across the heat flow path in both shaft and housing Use of materials with high thermal conductivity Thermal insulation between the bearing assembly and hot structures
Bearing Wear Wear is usually the main factor in determining bearing life. The amount of acceptable bearing wear is determined by requirements of accuracy of shaft location or allowable displacement of other moving parts. For bearing design purposes, it is customary to specify the allowable wear after running-in. For most plastic bearings, a high wear rate occurs during running-in as shown in Figure 17. For plastic bearings that have worn-in and achieved a steady state wear condition, the depth of wear is expressed by h = k‘ PV T
where h is geometrically related to the volume wear discussed earlier, k‘ is the linear wear coefficient, P and V are the “PV factor”, and T is the running time. Figure 18 shows approximate values of k‘ for different plastic bearing materials.
Lubrication A major advantage of plastic bearings is that they can usually be operated without externally added lubrication. Many tribological plastics incorporate lubricants in their structure to make
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
146
FIGURE 17. Typical running-in wear.
FIGURE 18. Wear coefficients for bearing materials.
them “self-lubricating”. However, wear life of plastic bearings can frequently be extended significantly with an initial application of grease during wear-in. Some highly loaded bearings will further benefit from periodic application of grease or oil.
Copyright © 1994 CRC Press, LLC
Volume III
147
REFERENCES
1. Turner, A. and Gurnee, E. F., Organic Polymers, Prentice-Hall, Englewood Cliffs, NJ, 1967, 245. 2. Hostalen GUR Properties Data Sheet, Hoechst Cellanese Corp., Houston, TX, 1991. 3. The Material Advantage: Garland Gar-Dur, A UHMW Plastic, Garland Manufacturing Co., Saco, Maine, undated. 4. Modern Plastics Encyclopedia ‘93, McGraw Hill, Hightstown, NJ, 1993. 5. Glidestar 400 Data Sheet, E/M Corp., West Lafayette, IN, 1993. 6. 1989 Guide to Selecting Engineering Materials, ASM Metals Progress, ASM International, Materials Park, OH, 1989, 116. 7. Pennlon Catalog 970, Dixon Corp., Bristol, RI, 1970. 8. Solidur Catalog CATA0030, Portland, OR, undated. 9. Engineering Resins, Texapol Corp., Bethlehem, PA, 1990. 10. Plaslube Internally Lubricated and Reinforced Products, Akzo Engineering Plastics, Inc., Evansville, IN, undated. 11. Engineering Resins Catalog No. 1300, Dixon industries Corp., Bristol, RI, 1991. 12. Molded Aromatic Polyamide Resin Products, Oiles Aramide M, Oiles America Corp., Birmingham, MI, undated. 13. Nyloil Catalog, Copely Development Ltd., Leicester, England, 1993. 14. Teflon Fluoroadditives, DuPont Polymers, Wilmington, DE, undated. 15. Meldin (brochure), Furon Co., Bristol, RI, 1991. 16. Torlon (brochure), Amoco Performance Products, Inc., Atlanta, GA, 1991. 17. Oiles Plastic Based and Rolled Bearings, Catalog No. 011–1, Oiles America Corp., Birmingham, MI, undated. 18. Celcon Acetal Copolymer, Short Term Properties (CE-4), Hoechst Celanese Corp., Chatham, NJ, 1990. 19. Stanyl 46 Nylon, General Information, DSM Engineering Plastics, Reading, PA, April 1992.
BIBLIOGRAPHY
Seymour, R. B., Engineering Polymer Sourcebook, McGraw-Hill Publishing Co., New York, 1990. Sperling, L. H., Introduction to Physical Polymer Science, John Wiley & Sons, New York, 1986. Modern Plastics Encyclopedia, Vol. 69, No. 13, McGraw-Hill, New York, 1993. Deanin, R. D., Polymer Structure, Properties and Applications, Cahners Books, Boston, 1972. Plastics Technology Manufacturing Handbook and Buyer’s Guide, Bill Communications Inc., New York, 1992. Yamaguchi, Y., Tribology of Plastic Materials, Elsevier, Amsterdam, 1990. ASM Handbook, Vol. 18, Friction, Lubrication, and Wear Technology, ASM International, American Society for Metals, Metals Park, OH, 1992.
Copyright © 1994 CRC Press, LLC
METAL MATRIX—SOLID LUBRICANT COMPOSITES Pradeep K. Rohatgi Subrat Ray, and Yongbing Liu
INTRODUCTION
Solid lubrication is observed in solids with lower shear strength between certain planes resulting in easy movement along these planes and reduced friction and wear. Most solid lubricants are therefore the layer-lattice or lamellar solids such as graphite and molybdenum disulfide, containing weakly bonded layers which facilitate relative movement under shear.1 In composites, solid lubricants are embedded in the matrix as a constituent. The friction and wear of metal matrix solid lubricant composites depend on smearing of solid lubricants on the mating surfaces to form a lubricating film. The lubricating film forms by transfer of lubricating constituent on the mating surface. The adhesion of solid lubricant to the underlying surface is an important factor for smearing. The tribological behavior normally displays two distinct stages, (a) an initial transient state while the film is forming, and (b) steady state when a stable film has formed.
SYNTHESIS OF METAL MATRIX—SOLID LUBRICANT COMPOSITES
Table 1 lists selected solid lubricants incorporated in metal matrix composites. Coefficient of friction of these lubricants is around 0.1 to 0.25. The lubricants commonly employed at elevated temperatures, such as BN and CaF, have relatively higher coefficients of friction. Generally, fabrication methods for metal matrix composites containing lubricating particles fall into three main categories: (a) powder metallurgy, (b) casting metallurgy, and (c) spray deposition. Table 2 lists selected composites prepared by different techniques.
Powder Metallurgy Basic manufacturing processes in powder metallurgy (P/M) include mixing, compacting, and sintering of particulate raw materials. The mixing process is the important first step and controls the particle distribution in composites. Since the present state of the art of mixing by blending does not allow close control, segregation or clustering of particles is a common problem at this stage. The primary reason for segregation is the different flow characteristics of different powders during mixing.2 The larger the particle size, generally the better will be the degree of distribution. Spherical particles mix better than irregular particles. Density difference also affects the results of mixing two or more powders: light particles stay on top, while heavy particles tend to sink to the bottom. After mixing, powders are compacted in a die at pressures that make the particle adhere at contacting points. Sintering is the last manufacturing step, and control parameters in this stage are the temperature and atmosphere. The primary problems in fabrication of metal graphite composites by P/M are sweating during liquid phase sintering and poor strength in solid phase sintering. Sweating is commonly corrected by adding a small amount of calcium in the form of calcium-silicon alloy to the powder mixtures in iron-graphite systems, and the volume fraction of graphite can then be increased even up to 90%.3 Techniques developed to increase strength include mechanical alloying and sintering under pressure.2
Casting Casting offers a relatively low cost alternative to powder metallurgy techniques. Presently, two casting methods are employed:4 (a) impregnation of a bed of dispersoids by liquid metal 0-8493-3903-0/94/$0.00 + $0.50 © 1994 by CRC Press, Inc.
Copyright © 1994 CRC Press, LLC
149
CRC Handbook of Lubrication and Tribology
150
or alloy under pressure in squeeze casting and pressure infiltration; and (b) dispersion of I particles or fibers in liquid or semisolid alloy by stirring, and the resulting slurry is cast by I gravity or pressure die casting. In squeeze casting, liquid metal is forced into a bed or a preform of particles or fibers under high pressure (70 to 100 MPa). In pressure infiltration, I molten alloy is usually forced at low gas pressures of ≤15 MPa to flow into a compacted preform or a bed in a tube and allowed to solidify. With this method, composites can be produced with a high volume fraction of dispersoids.5 Infiltration pressure can also be applied by a hydraulic ram in a die-casting machine.
Spray Deposition In this method, liquid metal and dispersoid powders are co-sprayed through an atomizer onto a substrate to form billet, disk, tube, strip, or laminated structures. Particles of 5 to 500 µm size have been used with metal flow rates of 0.25 to 2.5 kg/s to produce composites with 5 to 35 vol% particles.6 Aluminum, iron, nickel, titanium, copper, and cobalt base alloys have been used to produce metal matrix composites by this method.
THEORETICAL BASIS FOR UNDERSTANDING FRICTION AND WEAR BEHAVIOR IN COMPOSITES
Even the most carefully prepared real surface is gently undulating and consists of many microscopic and macroscopic asperities. Friction between two solid surfaces arise from interaction at discrete asperity sites where actual contact occurs. The basic processes involved are (a) adhesion at the contacting points, and (b) deformation of asperities due to load. To overcome friction, forces are required both to shear the adhesion bond, Fa, and also Fd to elastically or plastically deform obstructing asperities of the relatively softer material in the path of asperities of the harder material. If N is the applied normal load on the contacting surface, coefficient of friction f is given by7
Copyright © 1994 CRC Press, LLC
Volume III
151
where fa and fd are, respectively, the coefficients of friction due to adhesion and deformation. Adhesion strength and the resulting friction, f” are strongly influenced by the presence of surface oxides, absorbed films, and contaminants which prevent intimate contact between mating surfaces and inhibit strong adhesional bonds. When a surface consisting of a solid lubricant like graphite dispersed in a metallic matrix slides on another surface, a thin graphite film forms on the mating surfaces, reducing adhesional friction. However, the deformation contribution to friction is still primarily determined by mechanical properties of the matrix alloy.
Friction and Thin Film Lubrication Bowden and Tabor7 developed a theory for thin film lubrication, expressing coefficient of friction, f, as
Copyright © 1994 CRC Press, LLC
152
CRC Handbook of Lubrication and Tribology
Low shear strength of the film on sliding surface, Sf, and a relatively high substrate hardness, Hs, then result in a low coefficient of friction, f. However, this simple picture fails to explain many details in the complex frictional behavior of film-substrate combinations. Shear strength of the film under pressure as proposed by Bridgeman8 is where α is a material constant and P is the normal pressure. Friction coefficient becomes
At low pressure, indentation of asperities of the mating surface may be confined within the film layer if the film is thick and the load will be supported by the film resulting in P = Hf, the hardness of the film. If the film is soft, coefficient of friction, f, will be higher due to higher contribution of the first term in Equation 4 because of low film hardness. But for a thin film, the indentation load will be supported by relatively hard substrate material and P = Hs resulting in a lower coefficient of friction. As the normal load increases, real area of contact increases and becomes equal to the apparent area of contact. There will be no further increase in this variable. As pressure increases further, the first term in Equation 4 drops and, in the limiting case of very high pressure, coefficient of friction approaches the value of α, a characteristic of the film material. In the case of graphite bearing composites, the coefficient of friction both during run-in period and in steady state may be correlated approximately with the extent of film formation by the rule of mixture. where fm and fg are, respectively, the friction coefficients in the exposed matrix area and in the graphite film area. Ag is the fraction of composite sliding surface covered by graphite film. Following Equations 1 and 4: and,
Thus, the coefficient of friction may vary from fm to fg, depending on composition of the composite and ability of the solid lubricant phase to spread over the matrix and counterface. During initial sliding of composites, the solid lubricant comes out from its embedded state and spreads over the sliding surface of the composite and/or transfers to the counterface. These processes continue, and rate of wear eventually balances the rate of fresh supply of solid lubricant to the film from its embedded state in a dynamic steady state characterized by a steady value of friction. Evolution of the lubricating film to its dynamic steady state can be characterized by the changing coefficient of friction.
Wear in Composites Containing Solid Lubricants Wear particles in metal matrix composites are generated by the following mechanisms:9 (a) adhesion, deformation and fracture of asperities due to repeated interaction during sliding, Copyright © 1994 CRC Press, LLC
Volume III
153
(b) plowing by hard entrapped particles or hard asperities at the sliding surface, and (c) delamination due to subsurface crack nucleation and propagation. Wear debris generated by these mechanisms mostly form loose particles or sometimes transfer to the countersurface by mechanical interlocking or by adhesion. In the dynamic steady state of wear, asperities are continuously generated and removed by deformation and fracture. Plowing may also contribute to wear debris. When a sample undergoes wear, the extent of material removal depends on the size of the asperities. The higher the initial roughness, the more material is removed in the initial transient period before a steady state wear rate sets in, which is independent of initial roughness.10 The often-omitted data on the initial surface roughness of the samples are very important when reporting results on bulk wear averaged over a period including that of the transient state. Wear arising out of the three basic mechanisms711 is generally proportional to the applied normal load, N, sliding distance, S, and inversely proportional to the hardness of the wearing body, H. Thus, where W is the wear volume, K a wear constant, and C a geometrical factor equal to 1 for abrasion and 3 for adhesion. Since both hardness and the wear constant for the material depend on its microstructure, Equation 8 can be simplified as
where wear factor depends on the material and microstructure. Surfaces of pure metals and alloys are often contaminated with oxides and absorbed gases. During sliding at very small loads, wear behavior will correspond to that of the undisturbed oxide or contaminated absorbed layer. At a still larger load, the oxide or absorbed layer may wear away, exposing fresh metallic surface; but the surface may become contaminated again before its next contact with the counterface at the same location, depending on test configuration. Reforming of the oxide or absorbed layer is promoted by higher local temperatures at contact spots. When the conditions of load and sliding velocity are such that the oxide layer wears out during contact and fails to reform before the next contact, the wear behavior changes from mild oxidative wear to severe and metallic wear. This transition also depends on test configuration and the rime between successive contacts at a given location. In mild wear, the wear particles are very fine (≈1 to 10 µm), the subsurface is not heavily deformed, and coefficient of friction fluctuates. In severe wear, wear particles are large (10 to 100 µm) and metallic, the subsurface is heavily distorted, and coefficient of friction is relatively smooth. For a given load, transition from mild to severe wear takes place at a specific sliding velocity. For a given sliding velocity, the transition takes place at a specific load. This transition is quire; distinct from time-dependent transition at constant load. If the composition of the composite is such that the tribosurface is partly covered by a lubricating film, the change in the nature of wear on the metallic surface will be reflected in the overall wear of the composite. For composites containing smaller amounts of solid lubricant, one may observe this transition in the mechanism of wear of the exposed matrix on the sliding surface; but this effect will be obliterated when a large part of the tribosurface becomes covered by a lubricant film in composites containing a high amount of solid lubricant. Wear in the film of solid lubricant may take place primarily by delamination when film thickness exceeds the critical value. Accumulation of dislocations below the surface may then lead to subsurface cracking resulting in delamination.9 Delamination involves two types of
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
154
stress: (a) triaxial compressive stress, and (b) shear stress. The former is maximum at the surface opposing nucleation of voids. Below the surface, the compressive stress reduces, and after a certain depth, the shear stress can nucleate voids, preferentially at the interface between a second phase particle and the matrix. The voids so nucleated extend and coalesce to form an unstable crack which propagates to the surface, generating particles of debris. This wear process continues to erode the soft layer of film unless its thickness is so small that it is free from dislocations, and triaxial compressive stress prevents void nucleation and resulting delamination.
FRICTION AND WEAR BEHAVIOR OF METAL MATRIX-SOLID LUBRICANT COMPOSITES
Friction Characteristics Figure 1 shows the coefficient of friction observed in a number of composites containing graphite as the solid lubricant.12–24 When the lubricant content in the composites exceeds a critical level of about 28 to 30 vol%, both the mating surfaces of the composites and the counterface, like that of steel, become completely smeared with graphite, resulting in a friction coefficient independent of the matrix alloy. The contribution of the first term in Equation 4 or 7 to coefficient of friction, f, is dependent on the hardness of the matrix alloys and is not significant in composites containing more than the critical amount of lubricating particles. Microstructure and the hardness associated with the microstructure, influences friction as evident (Figure 2) in gray cast iron.25 For similar values of coefficient of friction reported by others for gray cast iron,26 there is increase in friction with hardness of the matrix contrary to what is expected from Equation 4. Barry and Binkelman27 observed a sharp increase in the coefficient of friction with lowering of hardness of the substrate with a thin surface film of MoS2 on substrates softer than the film material; however, for harder substrates, the friction becomes independent of substrate hardness. Thus, the trend of variation of coefficient of friction in Figure 2 results from the process of film formation on the sliding surface. Microscopic examination of gray cast iron samples showed that a softer matrix leads to larger area of the sliding surface being covered by graphite film. Plastic flow of the surface layer of the matrix appears to help in spreading of surface graphite into a film. The surface layer is capable of deforming continuously without much work hardening or fracture. Since the graphite film may not cover the entire sliding surface, the overall coefficient of friction will reflect the friction of the matrix as well. The pearlite matrix in cast iron reduces friction over that for ferrite matrix because poor adhesion of carbide in the exposed matrix area and higher strength of pearlite more than balance the effect due to increase in flow stress.26 Also, pearlite matrix contains part of the carbon as carbide, influencing the amount of graphite available for film formation. A mixed ferrite-pearlite matrix represents an optimum balance and shows a lower friction than either ferrite or pearlite matrix alone. A relatively higher friction in martensite or troostite matrix indicates that its higher strength has been more than offset both by a lower amount of graphite due to carbon in solution in the matrix and also by a difficult spreading of graphite due to higher flow stress of the surface layer. If steady state has been achieved, contribution of the exposed matrix on the sliding surface to the overall friction is governed by Equation 5, and fm should be reasonably constant. However, the contribution from the regions of tribosurface covered by graphite film may vary with pressure. For lower contact pressure, P in Equation 7 is equal to the film hardness, and fg should be constant. As pressure increases beyond the point of total contact, P increases till the asperities indent the matrix below, and P is equal to the matrix hardness. The friction coefficient should thus be constant at lower load and then ultimately decrease with a further increase in pressure. This decrease at high load has been observed by Muran and Srnanek28 in Copyright © 1994 CRC Press, LLC
Volume III
FIGURE 1. Variation of coefficient of friction with graphic content in metal matrix graphite particle composites sliding against steel.
155
Copyright © 1994 CRC Press, LLC
156
CRC Handbook of Lubrication and Tribology
FIGURE 2. Effect of matrix microstructure on the coefficient of friction in gray cast iron sliding against steel.
an Al-1.5 vol% graphite composite at a sliding velocity of 1.5 ms-1, as shown in Figure 3.
Wear Characteristics Figure 4 shows that wear rate generally decreases with an increase in the amount of solid lubricant in different metal matrix composites.23.29–30 In composites containing graphite, the wear rate is stabilized at a low value as the graphite content increases beyond some critical percentage which suggests that a lubrication film has covered the tribosurface completely For copper-WS2 composite, the wear rate initially reduces, but it increases when WS2 content is high. A similar phenomenon has been observed in Al-Pb composites by Mohan et al.31 This increase in wear has been attributed to a drastic reduction in the strength of the composite and a faster build of the film and its wear. This critical lubricant level above which wear increases may vary from system to system. Wear behavior in composites depends on the inherent nature of lubricating particles and their response to the smearing process. While wear rate generally increases as the load increases for aluminum silicon alloy-graphite composites, transition from mild to severe wear is not distinctly reflected in results under increasing load. However, the results on I copperbased lubricating particle composites show the transition in wear mechanism.35 Figure 5 shows the variation of wear with sliding velocity in Al-Si alloy base composites containing I graphite as compared with the matrix alloys.33–36As sliding speed increases, interface temperature also increases, resulting in (1) an enhanced rate of formation of oxides on the sliding surface, and (2) a decrease in flow stress. In addition, there may be thermally activated microstructural changes like dissolution of precipitates, etc. In Figure 5 the composite with only 5% graphite retains more or less the same trend of wear rate with sliding speed as that of the matrix. However, the composite with 15% graphite shows a different trend, which indicates that the sliding surface is largely covered by graphite film, and the wear rate becomes insensitive to changes in sliding speed. Copper alloy base composites containing graphite and MoS2 show a similar trend for variation of wear rate with sliding speed.37 Copyright © 1994 CRC Press, LLC
Volume III
157
FIGURE 3. Variation of coefficient of friction with normal pressure in AI-1.5 vol.% graphite composite sliding with a velocity of 1.5 m/s.
FIGURE 4. Variation of wear rate with the volume percent solid lubricant in copper- and silver-base composites containing MoS2, WS2, or graphite.
Effect of size of graphite and tungsten disulfide particles on the wear behavior of bronze composites is shown in Figure 6.38-39 The wear rates decrease as particle diameter increases. This effect has been attributed to plastic flow of matrix alloy during sliding to cover smaller embedded graphite particles before they are squeezed out and transfered onto the sliding
Copyright © 1994 CRC Press, LLC
158
CRC Handbook of Lubrication and Tribology
FIGURE 5. Variation of wear rate with sliding speed in Al-Si alloy base composite containing graphite and the base alloys sliding against steel.
FIGURE 6. Variation of wear rate with particle size in bronze-based composites containing WS2 or graphic particles.
surface. The smaller the particle and the more ductile the matrix, the greater the extent to which the particles are covered. Sugishita and Fujiyoshi25 have observed the same effect h nodular cast iron where larger nodule size results in lower wear. Also, the thin layer of matrix flowing over the graphite during sliding undergoes larger deformation and results in increased metal removal by fatigue. Kawamoto and Okabayashi22 investigated the effect of matrix microstructure on wear in spheroidal gray cast iron in dry sliding, as shown in Figure 7. Fully pearlitic matrix shows the Copyright © 1994 CRC Press, LLC
Volume III
159
FIGURE 7. Effect of matrix microstructures on wear rate in gray cast iron sliding against steel under normal load of 50 N.
lowest wear when compared with fully ferritic or a bull’s eye structure with free ferrite and pearlite in ratio of 1:1. Okumoto et al.40 observed that wear also depends on graphite shape. Gray cast iron with flake graphites has inferior wear resistance as compared to spheroidal gray cast iron. This may be due to lower matrix strength in gray cast iron containing flake graphite, and also the small transverse dimension of the flake makes it more easily covered by the matrix due to plastic flow at the surface during sliding. Wear rate also becomes anisotropic and dependent on relative orientation of the flakes and sliding direction.
Seizure Characteristics Seizure resistance of a material can be defined as its ability to withstand cold welding under pressure during sliding contact. Seizure of aluminum on aluminum, particularly severe under boundary lubrication and troublesome even under full film lubrication, can be improved significantly by addition of only 2 vol% graphite particles.41 Das and Prasad34 concluded that 3 vol% of graphite in Al-Si alloys increases seizure pressure by about 2 MPa over that for the base alloy under boundary lubrication. Rohatgi et al.42 summarized seizure behaviors in Algraphite by using normalized velocity and pressure. Liu et al.43 reported that Al-50 vol% graphite particle composites under dry sliding show almost the same seizure behavior as that of the base alloy when speed is below 3 m/s, but superior seizure resistance is observed in Al50 vol% graphite alloy above 3 m/s. It is evident that seizure resistance of aluminum alloys can be improved by adding graphite particles and that solid lubricants in composites are generally effective in interfering with asperity interactions and cold welding.
EFFECT OF ENVIRONMENTAL FACTORS ON FRICTION AND WEAR
Environmental Conditions and Lubrication Environmental factors significantly affect the lubricity of solid lubricants such as graphite, BN, and graphite fluoride. Graphite has a layered structure with weak interlayer bonding which allows smearing on the surface rubbed against it by interlayer slippage, but easy slippage of one layer over another occurs only in presence of water vapor or some volatile organic solvents.44 Similar results have been reported in metal matrix composites containing graphite. Effect of environmental gases on friction and wear of Ag-25% graphite composites
Copyright © 1994 CRC Press, LLC
160
CRC Handbook of Lubrication and Tribology
is shown in Table 3.51 Energy loss and wear are significantly reduced in each of five moist nonoxidizing gas environments compared to moist ail. Even very low moisture partial pressure (600 Pa, compared to total pressure of 105 Pa) affects the sliding results. The effect of atmosphere on friction and wear of various solid lubricants is described qualitatively in Table 4.23 The dependence of friction and wear of graphite on moisture can be reduced if WS2 is introduced into metal-matrix composites containing graphite.29 Friction of graphite in vacuum is reduced significantly by adding only 5% of WS2, and it becomes independent of the atmospheric pressure. Friction and wear of copper base-intercalated graphite (43 vol% of NiCl2) composites were reported by Ruff and Peterson53 to be affected significantly by argon gas only in composites which contained less than 10% graphite; that was almost no difference in the friction coefficient in air and argon if volume fraction of graphite exceeded 30%. Effect of operating temperature on friction and wear has been reported for different metal matrix-solid lubricant composites.21,23,52 Increase in temperature from 100°F (≈37°C) to 1000°F (≈537°C) increases friction coefficient of iron-graphite composites only by about 0.05.21 Copyright © 1994 CRC Press, LLC
Volume III
161
FIGURE 8, Variation of wear volume with sliding distance in Al-Si alloy base composites containing graphite sliding against steel under lubrication of turbine oil.
Friction coefficient increases by about 0.1 for Cu-20% WS2 composites.29 Lubricity of solid lubricants is generally retained up to the temperature ranges shown in Table 1. Easy flow of matrix material at elevated temperature may be primarily responsible for a higher coefficient of friction because more coverage of solid lubricant particles by matrix reduces the supply of solid lubricants onto the tribosurface for film formation. Matrix flow also explains the large increase in coefficient of friction in composites containing smaller particles and lower volume fractions of lubricant. Galling resistance of aluminum alloys in the presence of oil lubrication can be improved by dispersion of graphite particles54–55 in the matrix of aluminum alloys. Three causes may contribute: (a) lubrication by graphite film between sliding surfaces, (b) improved lubrication due to dispersion of debris of fine graphite particles in oil, and (c) voids left in the matrix after transfer of graphite acting as oil reservoir. Minimum graphite content required to inhibit galling is about 2 wt% in Al-Si-Ni alloys.54–55 Effect of oil lubrication on wear behavior of composites containing graphite particles is illustrated in Figure 8, where wear volume is compared for Al-Si alloy base composites with different graphite contents both during dry and during turbine oil lubrication.56 Only a small amount of graphite (=4.2%) reduces wear volume drastically in the presence of oil, while increase in graphite content beyond this level results in enhanced wear. This may be due to dispersion of higher amounts of graphite debris resulting in lower oil mobility.
Wear in Electrical Contacts Several composites designed for use in electrical contacts combine constituents to impart excellent wear resistance with high electrical conductivity.30–37–45–51 The most widely used include the less expensive copper-graphite composites and silver-graphite composites with very low bulk and contact resistance.45 The amount of material removed in sliding electrical contact is the sum of contributions from purely mechanical wear and an increment of mechanical wear resulting from matrix softening by local heating due to arc.52 The wear mechanism for composites in electrical contacts is complex, and depends on composition of
Copyright © 1994 CRC Press, LLC
162
CRC Handbook of Lubrication and Tribology
composites, the contact pairs, current and voltage level, sliding speed, environment, and contact pressure.45,50,57,58 Tsuya and co-workers29 showed that the lubricating film affects the contact resistance in sliding between a copper pin and the composites containing different amounts of WS2 particles. Contact resistance did not increase significantly with an increase of WS2 particles up to about 40% in a pure copper matrix, but contact resistance increased for WS2 content above 20% for copper-tin alloy base composites. The results suggest that contact resistance will not increase sharply until the rubbing surfaces of the composites are covered completely by lubricating film. Lee and Johnson51 reported that in a silver-graphite system, wear rate of composites increased with an increase of current density at both low and high temperatures. Current density is considerably higher for the same wear rate at higher temperatures as compared to that at ambient temperature. Coefficient of friction decreased with increasing current density, both in air and in CO2 atmospheres. The effect of sliding velocity on wear is complex. Teraoka37 has reported that in pantographs with contact strips made of copper-graphite composites, wear rate decreases as sliding velocity increases from 6.9 to 27 m/s. The opposite results have been reported by Casstevens et al. for copper-lead alloy containing graphite at high sliding speeds of 750 m/s and by Johnson and Kuhlman-Wilsdorf20 for silver-graphite composite at speeds of 13 and 26 m/s. Arc erosion tests in copper-base composites show a steady increase in erosion with are current.59 Erosion rates in copper-graphite and copper-matte composites are higher than that of base metal. Marshall60 suggested that wear of these composite brushes resulted mainly from mechanical factors as compared to that from electrical current. Teraoka37 reported that the different rail car-base affected the wear results of pantograph contact strips. Similar results have been reported by Lee and Johnson.51
FILM FORMATION
It has been observed that films of solid lubricant form on the sliding surfaces of various composites containing solid lubricant particles. These films reduce the extent of direct metalmetal contact, as can be observed in typical SEM micrographs.43 When the surface of a composite containing solid lubricant particles is polished, plastic flow of the matrix occurring at the surface layer may cover particles, if small, to restrict their transfer to the tribosurface. Friction and wear of such surfaces are high until the layer over the particles wears away partially, and the normal load can then squeeze lubricating material onto the sliding surface.25 The lubricant particles are then sheared by asperities on the sliding surface and eventually spread into a film. Rohatgi et al.”17,61 analyzed the sliding surface of Al-10 vol% graphite by Auger spectroscopy and established that the major elements on the tribosurface were oxygen, carbon, and aluminum. Over 30% of the surface was covered by graphite, a clear indication of smearing. However, thickness was not uniform; the film was generally 100 to 200 Å thick, but there are places where the thickness was relatively large.” Film formation has also been reported in other composites containing quite different solid lubricants like Pb.62 Ruff and Peterson53 observed similar film formation on copper-intercalated (NiCl2 graphite composite test pins and on the steel counterface. The films were patchy in distribution; while surface profilometry indicated an average thickness of about 1.1 µm, some patches were I as thick as 10 µm. They also observed that wear debris is preferentially collected at the I entrance edge of the recessed graphite region and the graphite film is formed at the exit edge. Baranov and Pademo63 observed that the graphite smeared on the tribosurface in coppergraphite composites is preferentially oriented with basal planes (0001) parallel to tot sliding surface.
Copyright © 1994 CRC Press, LLC
163
Volume III
FIGURE 9. The variation of MoS2 content on the sliding surface of Cu-MoS2 composite and its wear rate with bulk composition under loads of 5 and 50 kg.
Tsuya23 reported that the extent of the MoS2 film formed on a tribosurface is a function of volume fraction of MoS2 in copper-MoS2 systems, as shown in Figure 9. Friction coefficient decreases as MoS2 concentration on the tribosurface increases. Similar results have been reported in aluminum-graphite and aluminum-lead composites. Wilsdorf and co-workers45,50,57–58 observed a change in the wear mechanism with temperature in silver/copper-graphite particle composites. At temperatures below 100°C, a thick layer of water in graphite permits easy plastic shear on the basal plane. At higher temperatures, the water film desorbed, resulting in the rise of the critical resolved shear stress and limiting the ability of graphite to spread into films. The wear mechanism then becomes predominantly that for metal-metal sliding.
INDUSTRIAL APPLICATIONS
A variety of applications reported for metal matrix composites containing solid lubricant particles is shown in Table 5. Typical components produced from these composites include general and dry bearings, bushing, sliders, electrical contacts, pistons and liners, gears and shafts, washers and seals, valve seats, and bearing retainers. Copper base composites containing graphite can be used up to a temperature of 700°F in oxidizing atmosphere and 1700°F in submerged conditions or under nonoxidizing atmospheres. These composites are suitable for high load and slow speed applications, such as stoker bushings, drying oven conveyor bushings, high temperature stirring shafts and agitators, and electrical brushes for general use in electrical machinery. Other applications have been conceived by Hitachi Ltd., particularly for copper-graphite composites called
Copyright © 1994 CRC Press, LLC
164
CRC Handbook of Lubrication and Tribology
GRADIA, cast under pressure. When dispersed in copper base alloys, graphite imparts excellent machinability by breaking chips and providing lubrication at chip-tool interface. Graphite has, therefore, potential as a substitute for lead, which is a health hazard in copper alloys particularly when applied in plumbing.64 Stir-cast copper alloy-graphite composites have been developed at University of Wisconsin-Milwaukee for various applications. Silver based composite brushes are marked by very low noise, and stable contact resistance, low friction, and high conductivity. Silver-graphite brushes suppress radio interference noise and are useful for slipping, segmented rings, applications at high current densities, and other applications where special requirements justify high cost. Development of cast aluminum-graphite composite alloys started in 1966 for antifriction applications. Much of the early work used powder metallurgy which is relatively expensive and limits the size of components. Cast aluminum-graphite composites have unique microstructure in which graphite particles are located in interdendrite regions; they are reported to have superior tribological properties as compared to the base alloys. These composites have been produced by sand casting, permanent mold casting, centrifugal casting, and pressure die casting. Cast aluminum-graphite particle composites with over 2 vol% graphite have improved bearing parameters, improved galling resistance, and reduced friction coefficients as compared to the base alloy when evaluated under boundary lubrication either self mated or running against other graphite-free aluminum alloys.54 Examination showed that a graphite film had formed on a sample containing 2 vol% graphite particles; extensive subsurface shear contributed to formation of this graphite film which apparently imparted antiseizing properties. Pistons of aluminum-graphite particle composites, when tested in a HP diesel engine, led to reduced wear of the piston and rings, reduced frictional horsepower, freedom from seizing under adverse lubrication, and decreased specific fuel consumption. Pistons and liners of AlSi eutectic alloy-graphite particle composite led to similar results in half horsepower petroengines.65,66 Copyright © 1994 CRC Press, LLC
165
Volume III
Associated Engineering Company in Italy dispersed 4 vol% graphite particles in aluminum-18% silicon alloy. Tests of this composite mated with Al-11.48% silicon alloy in heated oil showed that scuffing resistance improved by a factor of two in comparison with the base alloy.67.68 They further evaluated liners of these alloys in two-stroke and four-stroke engines in collaboration with Ferrari, Hiro, and Alpha Romeo for passenger and racing car applications. The power generated was improved by 10%, there was no significant linear wear, and the pistons showed no signs of scuffing. Aluminum-graphite liners were fitted in Alpha Romeo racing cars which were victorious in the Formula 1975 World Championship. No seizure was experienced in 1975, 1976, and 1977 racing sessions, and power ratings were found to be high.68
REFERENCES
1. Lancaster, J. K., Solid lubricants, in CRC Handbook of Lubrication, Vol. n, Booser, E. R., Ed., CRC Press. Boca Raton, FL, p. 269. 2. Rack, H. J., in Proc. Conf. Powder Metallurgy Composites, MSI, 1987, 155. 3. Clauss, F. J., Solid Lubricants and Self-Lubricating Solids, Academic Press, New York, 1972. 4. Ray, S., Indian J. Tech., 28, 368, 1990. 5. Rohatgi, P. K., Asthana, R., and Das, S., Int. Met. Rev., 31, 115, 1986. 6. Lavernia, E. J., Int. J. Rapid Solidification. 5, 47, 1989. 7. Bowden, T. P. and Tabor, D., Friction and Lubrication of Solids I, Oxford Clarendon Press, Great Britain, 1950, p. 19. 8. Bridgeman, P. W., Proc. Am. Acad. Arts Sci., 387, 1936. 9. Suh, N. P., Tribophysics. Prentice-Hall, NJ, 1986. 10. Abrahamson, E. P., Jahanmir, S., and Suh, N. P., CIRP Ann. Inst. Inst. Prod. Eng. Res., 24, 513, 1975. 11. Raboniwicz, E., Friction and Wear of Materials, 1966, John Wiley & Sons, New York, 1966. 12. Gibson, P. R., Clegg, A. J., and Das, A. A., Wear, 95, 193, 1984. 13. Lancaster, J. K., in New Directions in Lubrication, Materials, Wear and Surface Interaction-Tribology in the 80’s, Loomis, W. R., Ed., Noyes Publications, Park Ridge, NJ, 1983, 320. 14. Pardee, R. P., IEEE Trans., PAS-86, 616, 1967. 15. Dillich, S. and Kuhlmann-Wilsdorf, D., Mater. Sci. Eng., 57. 213, 1983. 16. Rybakova, L. M. and Kuksenova, L. I., Soviet Eng. Res., 5, 9, 1985. 17. Rohatgi, P. K., Liu, Y., and Bar, T. L., Mater. Sci. Eng., A123, 213, 1990. 18. Rohatgi, P. K., Liu, Y., and Barr, T. L., Metall. Trans. 1991. 19. Yuasa, E., Morooka, T., and Hayama, F., J. Jpn. Inst. Met., 50, 1032, 1986. 20. Johnson, L. B., Jr. and Kuhlmann-Wilsdorf, D., Mater. Sci. Eng. 58, 4, 1983. 21. Bowen, P. H., Much. Des., 7, 195, 1963. 22. Kawamoto, M. and Okabayashi, K., Wear. 58, 59, 1980. 23. Tsuya, J. Jpn. Inst. Composites, 11, 127, 1985. 24. Owen, K. C., Wang, M. J., Prasad, C., and Eliezer, Z., Wear, 120, 117, 1987 25. Sugishita, J. and Fujiyoshi, S., Wear, 68, 7, 1981. 26. Kawamoto, M., Adach, M., Ando, A., and Okabayashi, K., J. Jpn. Foundrymen’s Soc. 50,32, 1978. 27. Barry, H. F. and Binkelman, J. P., Lubr. Eng., 22, 139, 1962. 28. Muran, M. and Srnanek, M., Kovove Mater., 23, 107, 1985. 29. Tsuya, Y., Shimura, H., and Umeda, K., Wear, 22, 143, 1972. 30. Johnson, J. L. and Morberly, L. E., IEEE Trans: Compon. Hybr. Manuf. Tech., 1978 (CHMT-1), p. 36. 31. Mohan, S., Agarwala, V., and Ray, S., Z. Metallkunde. 80, 904, 1989. 32. Suwa, M., Komuro, K., and Soeno, K., J. Jpn. Inst. Met., 40, 1074, 1976. 33. Choo, W. K. and Hong, C. H., J. Korean Inst. Met.. 17, 474, 1979. 34. Das, S. and Prasad, V., Wear. 133, 136, 1989. 35. Suwa, M., Hitachi Graphite-Dispersed East Alloy-Gradia. Hitachi Report, 1986. 36. Yuasa, E., Morooka, T., and Hayama, F. J., J. Jpn. Inst. Met., 50, 1032, 1986. 37. Teraoka, T., Wear of arc resistant sintered copper alloy for pantograph. Technical Report, Railway Technical Research Institute, Japan, 1983, 5. 38. Al’tman, V. A., Malakhov, G. V., Memelov, V. L., and Osipova, E. G., Sov. J. Friction and Wear, 10, 873, 1989. 39. Suwa, M., Komuro, K., and Yamada, T., J. Jpn. Inst. Mel., 42, 1034, 1978.
Copyright © 1994 CRC Press, LLC
166
CRC Handbook of Lubrication and Tribology
40. Okumoto, T., Sasaki, T., and Yamada, T., J. Jpn. Foundrymen’s Soc, 46, 913, 1974. 41. Pai, B. C., and Rohatgi, P. K., Trans. Indian Inst. Met., 27, 97, 1974. 42. Rohatgi, P. K., Lin, Y., and Asthana, R., in Proc. Conf. Tribol. Composite Mater., Rohatgi, P.K., Blau, P. J., and Yust, C. S., Eds., ASM Int., 1990, 69. 43. Liu, Y., Rohatgi, P. K., Ray, S., and Barr, T. L., in Proc. Int. Conf. Composite Mater., (ICCM/8), Tsai, S. W. and Springer, O. S., Eds., Honolulu, 1991, 204. 44. Savage, R. H. and Schaefer, D. L., J. Appl. Phys., 27, 136, 1956. 45. Kuhimann-Wilsdorf, D., Makel, D. D., Sondergaard, N. A., and Maribo, D. W., in Proc. Cost Reinforced Metal Composites, Fishman, S. G. and Dhingra, A. K., Eds., ASM Int., 1988, 347. 46. Johnson, J. L. and Scheurs, J., Wear, 78, 219, 1982 47. Casstevens, J. M., Rylander, H. G., and Eliezer, Z., Wear, 48, 121, 1978. 48. Casstevens, J. M., Rylander, H. G., and Eliezer, Z., Wear, 48, 409, 1978. 49. Baker, R. M. and Hewitt, G. W., J. Bear. (London). 33, 287, 1936. 50. Johnson, L. B. and Kuhlmann-Wilsdorf, Mater. Sci. Eng., 58, 21, 1983. 51. Lee, P. K. and Johnson, J. L., IEEE Trans., Vol. CHMT-1, 1978, 40. 52. Tsuya, Y., Umeda, K., and Saito, K., in Proc. 2nd Int. Conf. Solid Lubr.. Denver, 1978, 212. 53. Ruff, A. W. and Peterson, M. B., in Proc. Tribology of Composite Mater.. Rohatgi, P. K., Blau, P. J.. and Yust, C. S., Eds., ASM Int., Oak Ridge, TN, 1990, 43. 54. Badia, F. A. and Rohatgi, P. K., Trans. Am. Foundrymen’s Soc, 77, 402, 1969. 55. Badia, F. A., SAE Pap.. No. GT89–073, 1989, 1. 56. Suwa, M., Komuro, K., and Soeno, K” J. Jpn. Inst. Met., 40, 1074, 1976. 57. Kuhrmarm-Wilsdori, D., ASME, J. Tribal, 109, 321,1987. 58. Kuhlmann-Wilsdorf, D., Makel, D. D., Sondergaard, N. A., and Marino, D. M., in 14th Int. Conf. Electr. Contacts, Paris, IEEE, 1988, 1. 59. Jones, L., The Physics of Electrical Contacts, Clarendon Press, Oxford, 1957. 60. Marshall, R. A., Report No. EP-RR-3, Canberra, Australia, 1964. 61. Rohatgi, P. K., Liu, Y., and Barr, T. L., in Proc. Tribology of Composite Mater., Rohatgi, P. K., Blau, P. J., and Yust, C. S., Eds., ASM Int., 1990, 113. 62. Mohan, S., Agarwala, V., and Ray, S., Wear, 140, 83, 1990. 63. Baranov, N. G. and Paderno, V. N., Sov. J. Friction and Wear, 10, 662, 1989. 64. Rohatgi, P. K., Ray, S., and Liu, Y., Int. Metall. Revs., 37, 3, 129, 1992. 65. Krishnan, B. P., Raman, N., Narayauaswamy, K., and Rohatgi, P. K., Tribol. Int., 16,239,1983. 66. Krishnan, B. P., Raman, N., Narayauaswamy, K., and Rohatgi, P. K., Wear. 60, 205, 1981. 67. Bruni, L. and Iguera, P., Automob. Eng., 3, 29, 1978. 68. Bruni, L., AE Symposium, Part m, Italy 1987, 207.
Copyright © 1994 CRC Press, LLC
BONDED SOLID FILM LUBRICANTS Robert M. Gresham
INTRODUCTION
Development of bonded solid film lubricant products began in the late 1940s in the aircraft industry. Their use accelerated in the 1950s with the birth of the national space program and its need for lubricants in outer space subject to wide temperature extremes, radiation, and vacuum and other extreme environmental conditions. In the intervening years, bonded solid film lubricant technology has grown considerably and is now applied to a wide variety of industrial, automotive, military, and of course, aerospace applications. The subject of solid lubricants is covered in Volume II of this series.’ Bonded solid film lubricants contain materials with inherent lubricating properties (solid lubricants as covered in Volume II of this series) which are firmly bonded to the surface of a substrate. Major methods of bonding are resin bonding, burnishing, mechanical impingement, and sputtering, with resin bonding having the most commercial significance. Often the environment in which the component is to operate and the required tribological properties affect the type of bonded solid film lubricant to be used. There are three major areas which have to be addressed: first, solid lubricant pigment selection; second, resin or binder selection; and third, ratio of pigment to binder. Once these three areas have been defined, the formulations are augmented with flow agents, corrosion inhibitors, surfactants, and various solvents to ease application and to provide a variety of ancillary properties. This chapter will explore the different types of bonded solid film lubricants, the different mechanisms for bonding solid lubricants to the substrate, and corresponding performance properties which can be expected.
Design Considerations In order to select the proper solid lubricant or blend, a number of design parameters must be addressed to properly define the necessary performance properties for a specific application. Examples would be coefficient of friction, load carrying capacity, corrosion resistance, electrical conductivity, temperature, vacuum, humidity, and presence of liquid oxygen or radiation. Once these have been defined, solid lubricant materials can be selected. The most commercially significant solid lubricants are molybdenum disulfide, graphite, and polytetrafluoroethylene (PTFE). However, as described in Volume II, there are many other solid lubricant materials which are used, often as blends with molybdenum disulfide or graphite. In selecting the appropriate binder, similar design parameters which must be addressed include cure temperature limits, wear life (short-term, long-term), solvent resistance, low VOC (volatile organic component), and substrate material. Ratio of lubricant pigment to binder also has a significant effect on the overall performance properties. For example:
0-8493-3903-0/94/$0.00+$.50 ©1994 by CRC Press, inc
Copyright © 1994 CRC Press, LLC
167
CRC Handbook of Lubrication and Tribology
168
Most products involve a compromise between the extremes to establish optimum results for the conditions imposed. Solvent selection is governed primarily by the resin binder system. Other factors may include flash point, evaporation rate, toxicity, EPA regulations, OSHA regulations, volatile organic components, and shipping and storage. A variety of additives are also used to improve manufacturing efficiency, ease of application of bonded solid film lubricants, and overall performance. Typical additives include dispersants, anti-settling aids, wetting agents, flow agents, corrosion inhibitors, and colored pigments of dyes. Application of bonded solid film lubricants to the substrate material is of critical importance. As much as 80% of field failures are due to poor pretreatment and misapplication of the solid film lubricant, as opposed to improper selection of the product.2 Most bonded solid film lubricants are applied by techniques similar to those in the painting industry. One critical factor that must be controlled is film thickness: the normal recommended is between 0.0002” and 0.0005”. Several factors which govern selection of the application methods include: • Number of parts • Available equipment • Size of parts • Labor • Type of parts • Cost • Film thickness tolerance • Masking • Blind holes
The coating may be applied by conventional spray equipment, electrostatic spray equipment, dipping, roll coating, brushing, etc. While all of these methods are frequently used, spraying is the most effective in terms of wear performance and lubricity.
COATING CLASSES As we have seen in Volume II of this Handbook, a wide variety of materials can be used as solid film lubricants. The key to their use, however, involves getting them to adhere to the substrate in a uniform, thin film. The variety of bonded solid film lubricants developed over the years includes the following coating classes.
Impingement In the beginning, solid lubricant coatings were obtained by simply rubbing or burnishing the solid lubricant onto the substrate surface. These thin films adhere purely due to van der Waals and similar forces of attraction. The burnished films generally exhibited extremely short endurance life and were primarily used for assembly or mild forming operations. In order to enhance performance, impingement techniques were developed which in effect blast the substrate surface with a solid lubricant. Impingement films usually incorporate a low concentration of a proprietary inorganic binder system to enhance adhesion of the substrate. Surface morphology of the substrate is of prime importance and is usually achieved by abrasive blasting under controlled conditions. The resulting thin films find their major applications on fine-thread machine screws, high vacuum applications such as air bearings on satellite telescopes, and precision “clockwork” mechanisms such as timing and fusing devices in ordinance applications. These thin films can also be used as mold release agents. Copyright © 1994 CRC Press, LLC
Volume III
169
The impingement processes are generally described by the following specifications:
Resin-Bonded Coatings Resin-bonded solid film lubricants represent the largest and most commercially significant class of solid film lubricant products.3 As such, these products vary widely in their performance properties. For example, some products containing an air-cured acrylic or vinyl binder in small concentrations relative to the lubricating pigments provide minimal adhesion to the substrate. However, such products give inexpensive short-term lubrication, are suitable for forming applications, and are useful where excessive build-up of coating thickness is of concern. Other products contain minimal lubricating solids and are utilized more for their paint-like properties.
Organic Air Dry Coatings These generally provide improved performance vs. impingement coatings because, in addition to lubricity, they provide additional properties such as corrosion protection. Since they can be packaged in aerosol form, they are suitable for many field applications. Of all solid film lubricant types, these are probably least expensive and most easily applied, but with overall lower performance properties. The organic bonding agents are typically acrylics, alkyds, one and two-part epoxies, vinyls, and acetates. While most are not covered by military specifications, a few typically describe these products. For example, currently canceled MIL-L-46009 is typically an aerosol molybdenum disulfide/graphite mix with a minimum amount of resin binder to hold the solid lubricant to the surface for a wide variety of short-term applications. MIL-L-23398 and MIL-L-46147 are quite similar, but with much improved properties over 46009. MIL-L-23398D, used widely by all branches of the military, provides products with outstanding lubrication and reasonably good corrosion protection. The products are available in both aerosol and bulk form, which air cure in 6 hours. They are resistant to a wide variety of fluids such as aircraft turbine oils, solvents, and jet fuels. As a general rule, air-dry resins used in bonded solid film lubricants lack sufficient crosslinking and molecular weight when fully cured to compete with organic thermoset products in solvent resistance, wear life, and durability. A possible exception would be twopart epoxy and catalyst-cured silicone systems which substitute chemical energy for thermal energy to effect the resin change.
Organic Thermoset Coating Organic thermoset solid film lubricants are the largest single class of resin-bonded solid film lubricants. In addition to providing lubricity, a wide variety of products are used in thermal applications from cryogenic to about 750°F; coating applications involving extreme solvent or chemical resistance; and decorative applications from colored to the typical gun metal gray color characteristic of molybdenum disulfide and graphite. In addition to the most common thermoset phenolic and epoxy-phenolic resins, silicones, epoxies, urethanes, polyimides, polyamides-imides, and phenoxies are used. Representative specifications which describe products in this large class of solid film lubricants include the following: Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
170 䊉
Automotive • GM-6046 (GM) • M21-P8A (Ford) • PS-7001 (Chrysler)
A black PTFE-containing solid film lubricant used primarily on fasteners for corrosion protection and torque tension control. • DDC 95350 (Detroit Diesel)
A PTFE solid film lubricant with a relatively soft resin binder used to seal threads on freeze plugs on diesel engines. 䊉
Air frame • FPS-3006 (General Dynamics) • BMS 3–8 (Boeing) • LAC 34362 (Lockheed) • PS18021–3.1A (McDonnell Douglas) • RL-5A (Northrop) • LSM 146003 (Grumman)
Typically molybdenum disulfide and/or graphite with a phenolic binder system capable of extreme wear resistance, high load-carrying capability with low coefficient of friction, and resistance to all aviation lubricants and fluids. 䊉
Jet engines • A50TF147 (General Electric) • PWA 474 (Pratt & Whitney) • EMS 52402 (Garrett) • 11700A (Allison) • PWA 550 (Pratt & Whitney)
Typically molybdenum and/or graphite in an epoxy or phenolic binder system. 䊉
Military • MTL-L-46010 TY I/II • MTL-L-8937D • WS 20290
Molybdenum disulfide with an epoxy or phenolic binder system.
Organic Thermoplastic Solid Film Lubricants4 Typically these coatings are self-lubricating polymers which are applied in powder or dispersion form at coating thicknesses from 1 to 50 mils or more. The polymer is then fused to the surface of the part to provide a thick barrier coating which provides lubricity, abrasion resistance, chemical resistance, or release properties. Typical of these polymers are:
Polytetrafluoroethylene (PTFE)—A completely fluorinated polymer which melts about 620°F and is useful up to temperatures of 500°F. It has outstanding antistick characteristics, a low coefficient of friction, good resistance to most chemicals, and high dielectric constant. Typical applications include chemical processing equipment, high temperature cable insulation
Copyright © 1994 CRC Press, LLC
Volume III
171
and molded electrical components. This material, in lubricating grade powder form, is also used as a solid lubricant in resin-bonded products as described earlier. Fluorinatedethylenepropylene copolymer (FEP)—FEP is a copolymerization product of tetrafluoroethylene and hexafluoropropylene. It typically has a melting point of 550°F and a useful working temperature up to about 400°F. The material has outstanding weatherability, low friction, and is typically used for chemical process equipment, roll covers, and wire and cable applications. This material is also used in powder form in resin-bonded products, primarily as a release agent. Perfluoroalkoxy resin (PFA)—PFA is generally similar to FIFE and FEP, although with somewhat better mechanical properties. It is useful to temperatures as high as 500°F. Ethylenechlorotrifluoroethylene copolymer (ECTFE)—ECTFE is predominantly a 1:1 alternating copolymer of ethylene and chlorotrifluoroethylene forming linear chains. With a melting point of approximately 470°F, it is useful from cryogenic temperatures up to about 330°F. Its strength, wear resistance, and creep resistance is significantly greater than those of PTFE, FEP, and PFA. ECTFE is resistant to most corrosive chemicals and organic solvents over a wide temperature range. While fairly expensive, it is probably the most effective product in its most common use as a corrosion resistant coating. Lubricity is of secondary importance. Polyvinyladine fluoride (PVDF)—PVDF is a high molecular weight polymer of polyvinyl fluoride with a melting point of about 340°F. PVDF has substantially greater strength, wear resistance, and creep resistance than PTFE, FEP, and PFA. It resists most chemicals and solvents including liquid bromine and bromine salt solutions. PVDF is more commonly used for lining chemical piping systems and reaction vessels than as a lubricant. Low Volatile Organic Component (VOC) In the early 1980s, the Southcoast Air Quality Management District (SAQMD) in Southern California promulgated rules 1124, 442, 443, and 1145, which served notice to the paint and coatings industry that fundamental changes would have to be made. Compliance with the earliest of these regulations was often accomplished by reformulation of solvent systems, including the use of so-called “compliant solvents”. However, in the area of bonded solid film lubricants, the development of technology has been more difficult and regulatory agencies have extended compliance deadlines. Exemptions have been issued and the specialty coating5 subdivided into even more narrow groupings with differing regulatory limits. These changes result in a dynamic regulatory program which maintains a realistic balance with available technology. Technology has been developed to meet the requirements for less volatile organic component emissions while maintaining the performance of many solid film lubricant products. Typical of this new generation of products are solid film lubricants which meet the performance requirements of MIL-L-46010B Type II, but with a VOC of 250 g/l, well below most regulatory goals. Likewise products have been developed under MTL-L-85614, an aluminized fastener coating commonly used in the aircraft industry. New low VOC solid film lubricants will undoubtedly be an area of intense future R & D involvement. Inorganic Bonded Solid Film Lubricants These generally provide resistance to vacuum outgassing or resistance to liquid oxygen and are useful at high temperatures and in high radiation environments.3 The most common binder systems are silicates, phosphates, aluminates, and some organometallic materials such as titanates and some silicon based materials. The organometallics, when used in high temperature applications, become inorganic on curing or exposure to extreme temperature. These materials are commonly used in jet engines as antiseize coatings for threaded fasteners Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
172
and in a wide variety of fuel control valves and related moving parts. Typical specifications for inorganic solid film lubricants are 䊉
Jet engines • A50TF9 (General Electric) • PWA 298 (Pratt & Whitney) • PWA 36545
Graphite or molybdenum disulfide in a silicate binder. 䊉
Aerospace • P20013 (McDonnell Douglas) • MS-FC106 (NASA) • LB0140–007 (Rockwell)
Typically graphite or molybdenum disulfide lubricants with a phosphate binder for use in liquid oxygen service. 䊉
Military • MIL-L-81329 • MIL-L-47081
Graphite or molybdenum disulfide lubricants with silicate or phosphate binder systems. Ceramic-Bonded Solid Film Lubricants This is an “emerging class of solid film lubricants for high temperature application. These products contain high temperature solid lubricating materials such as graphite, a calcium fluoride/barium fluoride eutectic (as developed by NASA) and a variety of proprietary systems still under development.”5 The binder is typically a glass frit which is fused to form a continuous film. In some cases, these materials are applied from powder plasma guns which fuse the binder as it is applied. Other systems involve a liquid dispersion of glass and lubricant which is spray applied and oven cured. These ceramic solid film lubricants are capable of extreme wear resistance, and some can pass a 1/4” Mandrel test without cracking and flaking. Most formulations are proprietary and expensive and are used primarily in developmental aerospace applications. These coatings represent an area of major involvement I by solid film lubricant researchers. Sputtered Films With development of sputtering deposition in the early 1970s, it became possible to apply very thin solid lubricant films. Inherent films of easily controlled coating thickness from 1/10th to >10.µ with reasonable life and low friction are well suited for precision bearing elements and for extreme vacuum applications in spacecraft. The fundamental problem with sputtered deposited films is their morphology. Sputtered films generally develop a low I density, two dimensional columnar-void structure.6 In the case of molybdenum disulfide, the platelets of molybdenum disulfide grow nearly normal to the substrate with the edge sites exposed outward. These edge sites, in turn, readily react with oxygen or water vapor, causing increased friction and wear of the film. In addition to molybdenum disulfide, other materials such as gold, silver, lead, lead-tin alloys, and cubic boron nitride have been applied by this technique to lower coefficients of friction. Research is currently directed toward modifying the film behavior by alloying (co-deposition), by forming multilayered films, or I by bombarding with high energy ions. More recently, enhanced sputter deposition techniques have deposited dense Copyright © 1994 CRC Press, LLC
Volume III
173
molybdenum disulfide films with the platelets parallel to the substrate surface and exhibiting increased resistance to oxygen and water vapor and decreased friction and wear. Composite Coatings In an effort to reduce cost and weight, and to improve performance, increasing use is being made of composite coatings using more than one coating technology. Some examples are described below:
Anodization, impregnation and impingement—Wear resistance of aluminum is substantially improved by anodizing. Since an anodized surface is somewhat abrasive and is not lubricious, commercial processes have been developed for impregnation of lubricating material into the pores of the anodize. The most common lubricant has been PTFE under tradenames such as Nedox and Sintef. These processes can also provide a more attractive appearance than conventional anodize with better scratch and mar resistance. Additionally, molybdenum disulfide has been impinged into the surface of anodized parts with satisfactory results. Aluminum alloy/solid lubricant—Properties of aluminum metal have been enhanced by powder metallurgy technology. In one case, aluminum alloy (6061) was sintered into a metal matrix composite with as much as 14 vol% of graphite. These composite materials showed improved wear rates vs. aluminum metal, and improvement increased with increased graphite volume fraction.8 Aluminum matrix composites with up to 5 wt% molybdenum disulfide show similar reductions in wear rates.9 Physical vapor deposition/impingement—Physical vapor deposition (PVD) is a new technology used primarily to provide very thin, very hard, wear-resistant coatings to metal substrates. The most common example is titanium nitride. These coatings are advantageous in some applications as substitutes for hard chrome plating and similar hard surfacings used in the past. PVD coatings are nonpolluting to the environment, eliminate the possibility for hydrogen embrittlement, and, because of thinness, eliminate the need for subsequent machining and polishing. These coatings generally exhibit a high coefficient of friction on the order of 0.4 to 0.5 However, taking advantage of the fact that these films typically have a low density, two-dimensional columnar void structure, it is possible to combine physical vapor deposition with the impingement processes. These thin films have a relatively limited lubricating life owing to the thinness of the lubricating layer. Since the lubricant actually enhances interfacial shear rather than intrafilm shear between mating layers,6 the surface roughness of the parts to be coated is of critical importance. Coatings are most commonly used on precision parts such as guidance bearings, high temperature fuel control valves, and related precision hardware. Composite sputtered films—As alluded to earlier, lubricating solids deposited by various sputtering or PVD processes have certain limitations caused by the crystal morphology of the film (crystal orientation and density). Co-sputtered films of molybdenum disulfide with chromium, cobalt, nickel, tantalum, and gold have all improved the performance of these films.10 Electrolytic platings and lubricants—In many cases, solid film lubricants are formulated to provide very precise lubrication, for example with threaded fasteners where a specific application torque is desired to achieve a precise clamp load. Since it may be difficult or impossible to formulate the solid film lubricant with other performance characteristics such as corrosion resistance, in these cases solid film lubricants are commonly applied over platings. The plating is often coated with a chromate to further enhance corrosion protection before application of a solid film lubricant. This must be done with care in order to achieve adequate adhesion of the solid lubricant to the plated chromated substrate. In some cases, for example, Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
174
a preferred process would employ a zinc phosphate designed specifically for zinc plating, followed by clear chrome seal rinse and subsequent topcoating with a solid film lubricant. Electroless plating/lubricants — Corrosion and wear properties of steel substrates have been improved for a number of years through composite technology, where materials such as PTFE are codeposited along with an electroless nickel coating. In these processes, the electroless nickel coating contains as much as 25% by volume PTFE. This provides a wearresistant coating with a lower coefficient of friction, but at the expense of corrosion protection. Electroless nickel also can provide a good substrate for conventional resin-bonded solid film lubricants to enhance corrosion protection. Electroless nickel has also been combined with impingement processes to further reduce the coefficient of friction.
SURFACE PREPARATION FOR SOLID FILM LUBRICANTS
General Pretreatment of the metal substrate surface prior to application of bonded solid film lubricants is the single most critical item affecting performance. Pretreatments are performed on a metal substrate to modify surface roughness, hardness, and/or chemical reactivity to promote adhesion and enhance lubrication performance. It is extremely important to properly perform the optimum pretreatment for the specific metal in order to achieve maximum performance. Since solid film lubricants by themselves do not exhibit uniquely outstanding wear life, improper or inadequate pretreatment results in approximately 80% of most solid film lubricant failures.
Resin-Bonded Lubricants Ideal pretreatment processes for resin-bonded solid film lubricants fall into three basic operations: degreasing, grit blasting, chemical treatment. Each of these will be covered in some detail, with recommendations of the preferred approaches. Since each of these operations add cost, however, it is often necessary to compromise performance. The design engineer must be careful that these compromises do not ultimately result in a poorly designed, non-costeffective component.
Degrease Degreasing is necessary to remove machining oils, corrosion inhibitors, and related solvent soluble contamination. Failure to remove these contaminants usually results in poor adhesion of the solid film lubricant. There are three methods commonly used to degrease metal parts. The preferred vapor degreasing is typically done in accordance with MIL-T-7003 in specially designed equipment using common solvents such as 1,1,1-trichloroethane, trichloroethylene, or perchloroethylene. If vapor degreasing is not practicable, an alternative is cold degreasing immersed in a solvent such as 1,1,1-trichloroethane. Care must be taken that the solvent does not become excessively contaminated because it will leave a residual thin film of oil on parts. Finally, a wide variety of commercial caustic cleaners are used with outstanding results. With care that the caustic cleaner bath does not become contaminated or neutralized, this method is particularly useful for large volume production in applications such as automotive fasteners. Grit Blast After vapor degreasing, abrasive blasting or grit blasting is recommended. For most metals, aluminum oxide is the preferred medium. A wide variety of mesh sizes is available; for most work 220 mesh is preferred (see Figure 1). In addition to aluminum oxide, sand, starulite, walnut shells, peanut shells, and glass beads are commonly used, particularly on
Copyright © 1994 CRC Press, LLC
Volume III
175
FIGURE 1. Wear life vs. surface finish.
Some of the softer metals. The goal of grit blasting is to provide a uniform surface profile with a surface roughness about 16 to 32 rms for most applications. Table 1 shows the effect of various grit sizes of aluminum oxide on aluminum, titanium, and stainless steel.
Chemical Treatment The final operation necessary for optimum performance of bonded solid film lubricants is a chemical treatment. Usually this represents some kind of conversion coating such as a phosphate for ferrous metals and zinc; anodization for aluminum, magnesium, or zinc; passivation for corrosion resistant steels; black oxide for copper and iron; chromate for copper, Copyright © 1994 CRC Press, LLC
176
CRC Handbook of Lubrication and Tribology
aluminum, cadmium, anodized metals, phosphated ferrous metals, and zinc; and various etchants for corrosion resistant steels, ferrous metals, copper, zinc, and titanium.
Phosphate The preferred phosphate process is described in DOD-P-16232 (see Table 2). Under this military specification there are two primary phosphate types, zinc phosphate and manganese phosphate. Manganese phosphate provides a better wear-resistant base for solid film lubricants. Zinc phosphate provides additional corrosion protection, a desirable for many applications. In addition to these phosphates, calcium/zinc phosphate, iron phosphate, and nickel/manganese phosphate are used, although these are more applicable to paint systems. Generally phosphates are solutions of secondary and tertiary metal phosphates along with I other anions, which function as accelerators, and phosphoric acid. Reaction of the solution results in a chemical conversion. For example, when an iron surface is phosphated, an iron zinc phosphate crystal is created. This crystal becomes an integral part of the chemistry of the metal surface. When operating conditions are correct, the deposited phosphate is adherent with very little residual porosity. After applying a phosphate, parts are dried under controlled conditions to eliminate trapping of water in the crystalline structure which could cause flash corrosion and poor adhesion. Additionally, it is important to avoid hydrogen embrittlement. In the case of high strength steels, including spring steels, trapped hydrogen in the metal can potentially cause stress corrosion cracking and related failures. Therefore, it is important that the phosphated part be baked at a temperature that will eliminate any entrapped hydrogen.
Anodization Anodizing is the preferred conversion coating for such materials as aluminum and I magnesium. In this process a metal oxide coating is formed electrolytically, as described in NDL-C-8625C. This military specification recognizes three fundamental types: chromic, sulfuric, or hard anodize. Hard anodize is a type of sulfuric acid anodize which leaves a much harder, more wear-resistant coating and is preferable for solid film lubricants. In addition to
Copyright © 1994 CRC Press, LLC
Volume III
177
these three, phosphoric and oxalic anodize are less commonly used. For magnesium metals, the anodize process is covered under MIL-M-45202. Finally, zinc anodize is performed in accordance with MIL-A-81801 and provides an outstanding substrate for zinc as well. A proprietary process, Ticote, is also effective for anodizing titanium. Passivation Passivation generally involves treatment of the substrate metal with nitric acid. The purpose is to dissolve iron from the surface of corrosion resistant steels making the surfaces nickel rich relative to the normal composition of the material. This eliminates any micro corrosion/ oxidation which might occur prior to the application of the solid film lubricant. Passivation is typically done in accordance with MIL-STD QQ-P-35.
Miscellaneous Chemical Treatments Black oxide is used on copper and to a lesser extent on iron, as covered by military specification MIL-F-495 for copper and MIL-C-13924 for iron. This involves sulfide treatment for copper and caustic nitrate treatment for iron. This chemical conversion, while very inexpensive, does not provide the performance of a good phosphate on iron. Chromate conversion is also commonly used on copper, zinc, and phosphated ferrous metals. The purpose of this process is to treat the substrate with hexavalent chromium along with various activators such as acetates, sulfates, and fluorides under a controlled pH. This process provides extra corrosion protection. Finally, various etchants also represent a cost-effective method for pretreating metal parts. However, use of etchants alone can represent a compromise in performance properties. For corrosion-resistant steels, ferric chloride solution at approximately 40% provides a good base for adhesion. On ferrous metals, a hydrochloric acid etch works well in place of grit blasting. However, care must be taken if hydrogen embrittlement is a consideration. Etching other metals can be accomplished as follows: aluminum with nitric acid and hydrofluoric acid, copper with sulfuric acid and nitric acid, zinc with sulfuric acid and chromic acid, and titanium with nitric acid and hydrofluoric acid. Each of these provides surface “tooth” to promote adhesion. The critical importance of pretreatment for resin-bonded solid film lubricants is demonstrated in the following test. Three identical Timken T54148 test races were coated with a commercially available, phenolic bonded, MoS2-graphite product. Coating thickness was 0.00035 inch. All specimens were identically baked, and the only difference was the surface pretreatment each received prior to the application of the lubricant. Each specimen was then tested on the LFW-1 at 72 rpm and 630-lb load (ASTM D-2714). The results are summarized below.
Clearly it can be seen that the effect of pretreatment is dramatic. Table 3 reviews the recommended pretreatment process for the appropriate metals.
Inorganic Bonded Pretreatment for inorganic bonded solid film lubricants is equally as important as with the resin-bonded products and generally follows the same guidelines of cleaning, degreasing, grit blasting, etc. Since inorganic solid film lubricants are generally used for high temperature applications, phosphating and chromating are usually avoided, since they decompose at temperatures in excess of 450°F and 150°F, respectively. Generally the preferred pretreatment
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
178
is degreasing and grit blasting with aluminum oxide, and, where applicable, passivation. Inorganic solid film lubricants are generally not used on low melting materials such as aluminum unless the application is for high vacuum where the vacuum outgassing properties of an inorganic solid film lubricant are a primary consideration, or where specific chemical resistance, such as liquid oxygen service, is of importance. Even then, the lubricant cure temperature may be excessive for the grade of aluminum.
Ceramic-Bonded Solid Film Lubricants Substrates for ceramic-bonded solid film lubricants are pretreated in much the same fashion as for inorganic bonding. Chemical treatment such as phosphates are not used since ceramic solid film lubricants are generally cured at temperatures around 1000°F or higher. Therefore, the preferred pretreatment is vapor degreasing, followed by grit blasting. Since ceramic solid film lubricants are usually applied to corrosion-resistant materials such as Hastalloy and Waspalloy, passivation is usually not required. The main purpose of the pretreatment is to remove any oils, dirt, and loose debris, and to obtain the necessary surface roughness for optimal adhesion.
Sputtered and PVD-Applied Films Pretreatment for sputtered and physical vapor deposition-applied films generally involves a degreasing/cleaning operation often incorporating ultrasound. In some special cases, vapor honing is also done. However, since vacuum-applied films are usually applied to very smooth surfaces, the abrasive honing is usually accomplished with a slurry of very fine aluminum oxide, followed by subsequent cleaning in an ultrasonic bath. Once cleaned in this manner, the substrate is etched in the sputtering chamber via ion bombardment with an inert gas such as argon. The argon plasma cleans the surface on an atomic level to provide the required adhesion.
Composite Films Pretreatments for composite films vary considerably with the substrate. For example, the pretreatment for anodization/impingement involves the typical steps called out in MIL-C8625 for anodization as described earlier. After the anodization process, parts are withdrawn from the bath, dried, and immediately coated by impingement, impregnation, or topcoating. Other composite films have appropriate pretreatments for their specific processes and follow the same pattern as anodized impregnation.
COMPARATIVE PERFORMANCE CHARACTERISTICS
Many resin-bonded solid film lubricant formulations are commercially available to meet the wide variety of engineering applications. Tables 4 and 511 attempt to codify performance of a number of common systems. Always, however, testing and prototyping is necessary to fully qualify the solid film lubricant system for each application.
Copyright © 1994 CRC Press, LLC
Volume III 179
Copyright © 1994 CRC Press, LLC
180
CRC Handbook of Lubrication and Tribology
Use of bonded solid film lubricants on threaded fastener products can prevent galling, provide low prevailing torque, or, most important, provide a narrow consistent torque/tension relationship. This is particularly important in critical applications where the clamp force delivered by the threaded fastener is critical. The action of torquing a threaded fastener assembly stretches the bolt to introduce the critical working clamp load in the system. The torque/tension relationship in the following formula takes into account a number of different factors: type of bolt, material, strength level, type of finish, torquing mode, surface condition of the joint, etc.
where T is installation torque (lb-in), K is torque coefficient, D is nominal bolt diameter (in), and L is clamp load objective (lb). The prime variable in this formula is torque coefficient (K) which can be controlled by use of bonded solid film lubricant coatings. The torque coefficient itself may vary widely, due to surface conditions of the threads (usually due to manufacturing inconsistencies). A bonded solid film lubricant can reduce torque/tension I variability and clamp load variability.12 As an example, a typical six-sigma variation with zincplated fasteners would be ±30% of clamp force. For M10 plated bolts this variation range might be as much as 7,000 lb. Use of a solid film lubricant could reduce this variability by as much as 50% or 3,750 lb. Copyright © 1994 CRC Press, LLC
Volume III
REFERENCES 1. 2. 3.
4. 5. 6.
7.
8.
9.
10. 11. 12.
181
Booser, E. R., Ed., Handbook of Lubrication, Vol. II., CRC Press, Boca Raton, FL, 1984, 269. Gresham, R. M., Solid film lubricants: unique products for unique lubrication, Lubr. Eng., 143, 1988. McMurtrey, E. L., High Performance Solid and Liquid Lubricants, Noyes Data Corp., Park Ridge, NJ, 1987, 5. Modern Plastics Encyclopedia, Vol. 57, No. 10A, Oct. 1980, 31. Sliney, H. L., Status and new directions for solid lubricant coatings and composite materials, Tribology in the 80’s. Proc. Int. Conf. NASA, Cleveland, 1983, 665. Singer, I. L., Solid lubricating films for extreme environments, in Proc. Symp. Mat. Res. Soc., 148, 217, 1989. Hilton, M. and Fleischauer, P. D., in New Materials Approaches to Tribology: Theory and Applications, Pope, L., Fehrenbacher, L., Winer, W., Ed., (MRS Proc. 140, 1989). Jha, A. K., Prasad, S. V., and Upadnyay, G. S., Sintered 6061-aluminum alloy-solid lubricant particle composites: sliding wear and mechanisms of lubrication, Conf. Proc. Wear of Materials, Vol. 1, Denver, 1989, 9. Zanzam, M. A., Wear resistance of agglomerated and dispersed solid lubricants in aluminum, Mater. Trans.. JIM, 7, 516, 1989. Stupp, B. C, Thin Solid Films, 84, 257, 1981. Gresham, R. M., Bonded Solid Film Lubricants for Fastener Coatings, Fastener Technology International, April/May 1987. Frederick, W. R., Solid Film Lubricants—The fastener finish to minimize clampload variability, Fastener Technology International, December, 1990
Copyright © 1994 CRC Press, LLC
AEROSPACE APPLICATIONS OF SYNTHETIC FLUIDS AND LUBRICANTS Carl E. Snyder Jr. and Lois J. Gschwender
INTRODUCTION: UNIQUE AEROSPACE DEMANDS Synthetic fluids and lubricants have found wide use in aerospace equipment, primarily because they are better suited to the more demanding requirements of aerospace applications than mineral oil-based materials. The main condition that makes aerospace applications so demanding is the requirement to operate over an extremely wide temperature range. Aerospace operational fluids contain various additives critical to successful performance. These in general may include antioxidants, antiwear additives, extreme pressure additives, foam inhibitors, metal deactivators, viscosity index improvers, and dyes. Because of the complexity, and often company-proprietary nature, of many of the formulations, no detailed discussion will be provided on additives for each type of fluid. The upper operational temperature of functional fluids is typically extended above that for non-aerospace applications by two factors. One is the environment. Although aircraft typically operate at altitudes where the ambient temperature is very low, significant heating can occur by aerodynamic heating of the fluids and components during flight, by friction in high performance machinery, and by proximity to jet engines and to other hot surfaces. Another factor is the higher temperature caused by the need to use minimum weight mechanical systems in aerospace applications, utilizing minimum fluid and lubricant volumes. In addition, minimum size and volume of components, including the heat exchangers, require a very small quantity of functional fluid or lubricant to circulate very quickly through systems operating at relatively high temperatures. Aerospace fluids and lubricants also must operate at the extremely low temperatures encountered by aerospace equipment at high altitudes, in cold climates, and in space. While wider temperature range operational requirements have also been a concern for aerospace hydraulic fluids, the primary reason for development and subsequent use of synthetic based fluids for aerospace hydraulic system applications has been the need for more fire resistance. The MIL-H-5606 mineral oil-based hydraulic fluid, once the standard for the aerospace industry, was recognized from its introduction as a very flammable fluid, and hydraulic fluid fires caused significant losses among both military and commercial aircraft. This flammability hazard was worsened by the inability, in many cases, for adequate fire fighting resources to respond in a timely manner. Often, by the time fire fighters reached the scene, the hydraulic fire had spread to the fuel system, destroying the aircraft. The high cost associated with aircraft hydraulic fires has led to the development of new, fire-resistant synthetic hydraulic fluids and, more recently, to nonflammable hydraulic fluids. As non-aerospace applications become more costly and sophisticated and synthetic fluids and lubricants become more readily available at lower costs, synthetics will be used more widely in them as well. To better define the scope of aerospace applications, it must be recognized that the largest volume applications occur in aircraft equipment. While the excellent characteristics of synthetic fluids and lubricants have also led to their use in spacecraft, missiles, and satellites, these volumes are significantly smaller and they are not as well known or defined. Most of the classes of synthetic lubricants covered in this chapter will be covered in detail in subsequent chapters of this part of the Handbook. Consequently, references will be cited only when a specific class of synthetic lubricants is not covered elsewhere in this book or because the information has important significance. 185
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
186
FIGURE 1. Temperature ranges for aerospace turbine lubricants.
The two main applications of synthetic fluids and lubricants are as liquid lubricants, primarily in gas turbine engines, and as hydraulic fluids. Applications involving synthetics that are of lower volume axe greases, coolants, and inertial guidance damping fluids. In many of these critical low volume applications, a synthetic lubricant is the only suitable material.
GAS TURBINE ENGINE OILS
Gas turbine engine oils constitute the highest volume application of synthetic liquid lubricants in the aerospace industry. The operational temperature ranges of the existing synthetic gas turbine engine oils and their chemical classes are shown in Figure 1. Anticipated requirements for future advanced turbine engines are also shown. The open area represents the potential maximum upper temperature requirement for the advanced turbine engines. Esters are the most widely used class of synthetic lubricants employed as gas turbine engine oils. This class was chosen because of its wide usable temperature range and its excellent thermooxidative stability in the presence of metals. Esters are required to operate at extremely low temperatures (down to -54°C), at which their viscosity must be low enough to permit the engines to start, as well as at high bulk fluid temperatures (reaching 204°C), at which they must provide lubrication for the main shaft bearings in the engine. The ester-based engine lubricants are described in military specifications MIL-L-78081 and MIL-L-23699.2 The upper temperature limits in Table 1 are rough estimates to be used cautiously because many other variables contribute to usable upper temperature. Two temperature ranges are shown for MTL-L-7808. The upper bar, denoting the -54 to 175°C temperature range, is for the J revision of the specification, which is in effect at the time of the writing of this chapter. The lower bar, denoting the -51 to 204°C temperature range is for the K revision, which is anticipated to be in effect soon. Commercial air carriers prefer the MIL-L-23699, since a -40°C low temperature performance is adequate and since the higher viscosity at upper temperatures is desirable for longer engine and oil life, if not for the actual higher temperature use. Materials conforming to the current MIL-L-7808 and MIL-L-23699 specifications are adequate to meet the lubrication requirements for most current aerospace gas turbine engines. However, in an attempt to improve fuel efficiency, higher operational temperatures will be Copyright © 1994 CRC Press, LLC
Volume III 187
Copyright © 1994 CRC Press, LLC
188
CRC Handbook of Lubricants and Tribology
required. A portion of those requirements can be met by the improvements detailed in the MIL-L-7808 K revision which extends the upper temperature operational capability to 204°C, This requires a careful balance of ester-based stocks and improved additives to achieve the balance of viscosity-temperature properties and excellent thermooxidative stability as well as other requirements for a gas turbine engine lubricant.3 The lower operational temperature of the MIL-L-7808K revision was relaxed from -54 to -51°C. The most stable ester-based gas turbine engine oil developed to date is a material conforming to the requirements of the inactive specification MIL-L-27502, which had a -40 to 240°C temperature range. In addition to turbine engine lubrication, the esters are used in aerospace applications as low temperature greases, e.g., MIL-G-23827,4 gear oils, as in DOD-L-85734,5 and to a lesser degree as instrument lubricants, one example being MIL-L-6085.6 Because some aircraft engine operational temperatures exceeded the limits of ester-based lubricants, another class of synthetic lubricants, polyphenylethers, was developed. They possess significantly higher high temperature stability.7 One liquid lubricant in this class is described in military specification MIL-L-871008 with an upper operational temperature of 300°C. MIL-L-87100 also has excellent fire resistance as demonstrated by a flash point in excess of 275oC and an autogenous ignition temperature of 610°C. Their major deficiency is extremely poor low temperature operational capability. Their pour points of +5°C and higher limit their low temperature use to no lower than + 15°C. Another drawback to this class of synthetics is that the formulation currently described by the specification has relatively poor lubricity characteristics. These limitations, along with their high cost ($1000+ per gallon), have limited their use to applications where no other liquid lubricants will function. Nevertheless, as more efficient gas turbine engines operating at higher temperatures are I developed, the polyphenylethers, either as MIL-L-87100 or as an advanced version of the specification, will find increased applications. Along with advances in high temperature lubricants, advances in the entire lubrication system, i.e., seals and other materials, must I also happen concurrently. When capabilities of the polyphenylethers are exceeded or when liquid lubricants capable of operating not only at their elevated temperatures, but also at the more typically required low temperatures of -40°C and below, liquid lubricant of choice will be likely based on a perfluoropolyalkylether (PFPAE).9 Commercial versions of this class of synthetic lubricants are currently available that have the potential for operating over a -54 to 300°C temperature range. Research and development programs are currently underway to increase the upper temperature to at least 345°C. The major deficiency of this class of synthetic lubricants is the lack of suitable additive technology. The chemical behavior of the PFPAE fluids is so different that the additives used to enhance the properties of other lubricants are not even soluble in PFPAE fluids. There are very limited examples of soluble additives and those were all specifically synthesized to be soluble in PFPAE fluids.10-13 While this class of fluids has very attractive and impressive I properties as unformulated fluids, their true potential cannot be realized until a supporting technology base of performance improving additives has been developed. The types of I additives required for PFPAE fluids in aerospace applications are (1) metal deactivator I stability additive, (2) rust inhibitor, and (3) lubricity additive. The PFPAE synthetics are I used in oxidatively stable greases as described in military specification MIL-G-27617.14 I Other potential aerospace applications for formulated PFPAE fluids include long-life lubricants I for space, instrument lubricants, and high temperature nonflammable hydraulic fluids.
HYDRAULIC FLUIDS
Synthetic-based hydraulic fluids are widely used in aerospace. The operational temperature I capabilities of current mineral oil and synthetic-based hydraulic fluids are shown Copyright © 1994 CRC Press, LLC
Volume III
189
FIGURE 2. Temperature ranges for aerospace hydraulic fluids.
in Figure 2. The solid areas of the bars represent temperatures at which satisfactory operating experience has been demonstrated. The open areas represent temperatures for which very limited data exists for the specific fluids, but at which satisfactory performance is expected. The mineral oil hydraulic fluid which the synthetics replaced in both commercial and military aircraft is described in specification MIL-H-5606.15 The reason synthetic fluids were developed to replace MIL-H-5606 was to provide increased fire safety. The flammability characteristics of MIL-H-5606 and selected synthetic aerospace hydraulic fluids are shown in Table 1.16 While MIL-H-5606 naphthenic mineral oil-based hydraulic fluid has proven to be adequate from an operational aspect, the high flammability hazard associated with its use is well known.16 The commercial aircraft industry recognized this hazard first and, in conjunction with the fluid industry, developed a fire-resistant hydraulic system around the phosphate ester class of synthetics. It was necessary to develop an entire hydraulic system because the phosphate esters are not compatible with the same seals, paints, wiring insulation, etc. that are used in aircraft with the hydrocarbon-based hydraulic fluids. In addition, hydraulic system components had to be modified to provide optimum performance with the new phosphate ester fluids described in AS1241b.17 The military community did not follow the commercial industry in the switch from MILH-5606 to phosphate esters. This decision was driven primarily by the noncompatibility of the phosphate esters, not only with the aircraft systems designed to use the hydrocarbon-based MIL-H-5606, but also with the ground service equipment. In fact, mixtures of MIL-H-5606 and AS 1241 hydraulic fluids resulted in gel formation causing excessive maintenance to correct the problem. In addition, the aggressive solvency of the phosphate esters toward seals, paints, and wiring insulation used in aircraft with hydrocarbon oil-based hydraulic systems prevented their consideration as a retrofit option. The military conversion from MIL-H-5606 to a fire-resistant synthetic required development of a new class of fluids, i.e., synthetic hydrocarbon fluids based on polyalphaolefins (PAOs) as described in military specification MIL-H-83282.18 MIL-H-83282 was developed as a no-retrofit, drain-and-fill replacement for MIL-H-5606. This required total compatibility with the materials used in MIL-H-5606 systems and with the MIL-H-5606 system designs. Most military aircraft were converted to MIL-H-83282 by 1985. The only aircraft for which the conversion was not approved were those for which acceptable operation at—54°C could not be compromised. With its higher viscosity at low temperatures, MIL-H83282 is described as a -40 to 204°C hydraulic fluid compared to -54 to 135°C for MIL-H-5606. A recently completed development program has provided a PAO-based fireresistant hydraulic fluid, MIL-H-87257, with equivalent -54° viscosity to MTL-H-5606.19–21 The improved fire-resistant properties of MIL-H-83282 over MIL-H-5606 which have resulted in significant reductions in hydraulic fluid fire damage as shown in Table 1 include: (1) higher
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
190
flash and fire points; (2) higher autogenous ignition temperature; (3) lower flame propagation rate; and (4) improved resistance to gunfire ignition. The conversion of aircraft from MIL-H-5606 to MIL-H-83282 was accomplished by both drain-and-fill and attrition, which were equally successful and without problems. Other quite important, but smaller volume applications, of PAOs are greases such as MTL-G-81322,22 instrument lubricants such as MIL-L-85812,23 and liquid coolants.24 The PAO-based greases provide excellent usable temperature range and good reliability with low maintainability requirements. Instrument lubricants based on PAO have successfully replaced the difficult-toobtain paraffinic-based mineral oil instrument lubricants previously used. PAO-based coolants meeting the properties defined in MIL-C-8725225 are in the process of replacing another class of synthetic fluids, the orthosilicate esters, as dielectric and liquid coolants in military electronic systems.24 The polyalphaolefins have excellent properties as lubricants and hydraulic fluids. Their compatibility with mineral oils and systems designed to use mineral oilbased lubricants and fluids makes them excellent candidates for use both in newly emerging aerospace systems as well as in replacement of mineral oil-based products when they become either difficult to obtain or no longer provide adequate performance. Both phosphate ester and PAO hydraulic fluids have given excellent performance and significantly reduced, hydraulic fluid fire hazards in both commercial and military aircraft. However, they are not nonflammable, but are capable of ignition if sufficient energy (temperature, flame, etc.) is available. On current and future aircraft, high fire hazard areas involving hydraulic fluids are brake systems where brake temperatures can approach 1600°C on an aborted take-off and engine nacelles where temperatures are in excess of 800°C. Both of these conditions exceed the autogenous ignition temperatures and flash and fire points of both phosphate ester- and PAO-based hydraulic fluids. As costs of aircraft and other aerospace systems continue to increase, it becomes even more important to minimize the possibility of hydraulic fluid fires. The development and validation of a completely nonflammable hydraulic fluid and compatible seals has recently been completed.26–28 The synthetic hydraulic fluid is based on chlorotrifluoroethylene oligomers (CTFE) and is described in military specification MIL-H-53119 (ME).29 The CTFEbased hydraulic fluid is not compatible with hydraulic systems designed for use with other hydraulic fluids and therefore requires systems designed around its unique properties. MILH-53119 is specified for use from —54 to 175°C and is compatible with a number of elastomeric seals. One of the major disadvantages of MIL-H-53119 is significantly higher density, which results in a serious penalty in aerospace applications. In order to overcome this penalty, higher pressure hydraulic components were developed and systems were designed and validated. At higher pressures (8000 psi) the penalties associated with the higher density are minimized due to the extremely small volumes of hydraulic fluid required. If the weight penalty were not important, MIL-H-53119 could be used at lower pressures for a variety of nonflammable application areas. Another important application for higher molecular weight versions of CTFE as well as polymers of bromotrifluoroethylene (BTFE) is as high density flotation/damping fluids for inertial guidance systems.
OTHER FLUIDS AND LUBRICANTS
The only class of synthetic fluids which has been developed for quite some time that has not been discussed in this chapter is the silicones. This class has some very interesting properties which would make it seem to be a serious candidate for a number of aerospace applications. Most important of those is the extremely good viscosity-temperature properties the silicone fluids possess, especially the polydimethylsiloxanes. Copyright © 1994 CRC Press, LLC
Volume III
191
However, the silicones also possess properties that make them less desirable for the two major volume applications in aerospace, i.e., gas turbine engine lubricants and hydraulic fluids. The more significant deficiency is their inability to provide lubrication for steel on steel rubbing surfaces. Lubricity additives are generally not effective in silicones. This deficiency has limited their use as both liquid lubricants and hydraulic fluids. In addition, another deficiency that limits their use as hydraulic fluids is their low bulk modulus, or high compressibility. This would require larger actuators than for less compressible fluids to compensate for the “sponginess” of the fluid; but the weight of the hydraulic system would be significantly increased as a result of the larger components and the extra volume of hydraulic fluid required, which is unacceptable for aerospace applications. However, silicones have been used in a variety of greases30 that are widely used in aerospace applications. Another member of the silicon containing class of synthetic fluids is the silicate ester.7 This class has had two areas of application: wide temperature-range hydraulic fluids and coolants. The original application of the silicate esters as a hydraulic fluid was described in military specification MTL-H-8446.31 This specification, which has been cancelled due to lack of a need for the fluid currently, described a hydraulic fluid for use over the temperature range of -54 to 204°C. The silicate esters were the most acceptable class of hydraulic fluids for that requirement. Their major deficiency was their propensity to hydrolyze with any moisture getting into the hydraulic system. The resulting hydrolysis products were an alcohol, which degraded the fire resistance of the fluid, and a gelatinous precipitate that clogged system filters and the small orifices that exist in hydraulic systems. (Although the silicate ester is no longer used as a military hydraulic fluid, a silicate ester fluid is currently used on the Concorde commercial aircraft.) Similar hydrolysis problems were experienced with the silicate ester-based coolants described in military specification MIL-C-47220.32 This problem with hydrolysis which resulted in a high level of maintenance has led to the recent substitution of the PAO-based coolant MIL-C-87252 for MIL-C-47220 in many military aerospace applications.
DEVELOPMENT FLUIDS AND LUBRICANTS
The synthetic fluids and lubricants discussed previously in this chapter have either found significant application in the aerospace industry, or there is a significant production capability and potential applications have been identified. In this section, classes of newly emerging synthetic lubricants and fluids will be briefly discussed as well as the properties which make them promising. The first class of newly emerging synthetics is the silahydrocarbon or tetraalkylsilane. While this class has been known for quite some time, their potential application in the aerospace industry has been significantly advanced only recently.33 The largest volume application for the silahydrocarbons is as wide temperature range, high temperature, fireresistant hydraulic fluids. Their outstanding viscosity-temperature properties make them excellent candidates for use down to—54°C while still maintaining adequate viscosity at elevated temperatures to provide adequate film thickness for lubrication. Their excellent stability at temperatures up to 370°C permits extended use at elevated temperatures. Since these fluids contain aliphatic carbon-hydrogen bonds, oxygen must be excluded at elevated temperatures. These fluids have extremely low volatility which makes them excellent for long life, noncontaminating liquid lubricants for space.34,35 Their excellent. viscosity-temperature characteristics permit the selection of extremely high molecular weight (1000 to 1500 amu) silahydrocarbon fluids. Another class of synthetic fluids and lubricants which are still in the development stage are n-alkyl benzenes.36 These fluids have very good thermal stability and viscosity-temperature
Copyright © 1994 CRC Press, LLC
192
CRC Handbook of Lubricants and Tribology
properties. One advantage these fluids have over the PAO and silahydrocarbon classes for use at high temperature is their improved solubility for performance-improving additives. The benzene ring appears to provide significant solubility enhancement for the typically polar performance-improving additives needed for a wide temperature range, high temperature hydraulic fluid. In this chapter, we have attempted to provide an accurate picture of the current applications of synthetic fluids and lubricants in aerospace. As requirements change, new synthetic fluids and lubricants will be developed to meet them. The excellent properties demonstrated by the various classes of synthetic fluids and lubricants and the ability of the fluid and lubricant technologist to develop tailored materials with optimized properties will ensure their continued utilization in the future.
REFERENCES 1. MIL-L-7808J Military Specification, Lubricating OU, Aircraft Turbine Engine, Synthetic Base, NATO Code Number 0–148 (11 May 1982). 2. MIL-L-23699D, Military Specification, Lubricating Oil, Aircraft Turbine Engine, Synthetic Base, NATO Code Number 0–156 (9 Oct. 1990). 3. Gschwender, L. J., Snyder, C. E., Jr., and Beane, G. A., TV, Military aircraft 4-cSt gas turbine engine I oil development, Lubr. Eng., 43(8), 654, 1987. 4. MIL-G-23827B Military Specification, Grease, Aircraft and Instrument, Gear and Actuator Screw, NATO Code G-354 (20 June 1983). 5. DOD-L-85734, Lubricating Oil, Helicopter Transmission System, Synthetic Base (21 Feb. 1985). 6. MIL-L-6085C Military Specification, Lubricating Oil, Instrument, Aircraft, Low Volatility (5 Feb. 1991). 7. Gunderson, R. C. and Hart, A. W., Eds., Synthetic Lubricants, Reinhold Publishing, New York, 1962, 402. 8. MIL-L-87100, Military Specification, Lubricating Oil, Aircraft Turbine Engine, Polyphenyl Ether Base (12 Nov. 1976). 9. Snyder, C. E., Jr. and Gschwender, L. J., Fluoropolymers in fluids and lubricant applications, Ind. Eng. Chem. Prod. Res. Dev., 22, 383, 1983. 10. Tamborski, C. and Snyder, C. E., Jr., Perfluoroalkylether Substituted Aryl Phosphines and Their Synthesis, U.S. Patent 4,011,267 (Mar. 8 1977). 11. Tamborski, C. and Snyder, C. E., Jr., Perfluoroalkylether Substituted Phenyl Phosphines, U.S. Patent 4,454,349 (June 12 1984). 12. Sharma, S. K., Gschwender, L. J., and Snyder, C. E., Jr., Development of a soluble lubricity additive for perfluoropolyalkylether fluids, J. Syn. Lubr., 7, 15, 1990. 13. Gschwender, L. J., Snyder, C. E., Jr., and Fultz, G. W., Soluble additives for perfluoropolyalkylether liquid lubricants, Lubr. Eng., 49, (1993). 14. MIL-G-27617D, Military Specification, Grease, Aircraft and Instrument, Fuel and Oxidizer Resistant (14 Nov. 1984). 15. MIL-H-5606E, Military Specification Hydraulic Fluid, Petroleum Base, Aircraft Missile and Ordnance, NATO Code Number H-515 (26 Jan. 1978). 16. Snyder, C. E., Krawetz, A. A., and Tovrog, T., Determination of the flammability characteristics of aerospace hydraulic fluids, Lubr. Eng., 37, 705, 1981. 17. AS 1241b, Fire resistant phosphate ester hydraulic fluid for aircraft, Society of Automotive Engineers, Warrendale, PA, March 1992. 18. MIL-H-83282 Military Specification, Hydraulic Fluid, Fire Resistant, Synthetic Hydrocarbon Base, Aircraft, Metric, NATO Code Number H-537 (25 March 1986). 19. MIL-H-87257, Military Specification, Hydraulic Fluid, Fire Resistant, Low Temperature, Synthetic Hydrocarbon Base, Aircraft and Missile, NATO Code Number H-538 (2 March 1992). 20. Gschwender, L. J., Snyder, C. E., Jr., and Fultz, G. W., Development of a -54°C to 135°C synthetic I hydrocarbon based, fire-resistant hydraulic fluid, Lubr. Eng., 42, 485, 1986.
Copyright © 1994 CRC Press, LLC
Volume III
193
21. Gschwender, L. J., Snyder, C. E., Jr., and Sharma, S. K., Pump evaluation of hydrogenated polyalphaolefin candidates for a -54°C to 135°C fire-resistant Air Force aircraft hydraulic fluid, Lubr. Eng., 44, 324, 1988. 22. MIL-G-81322D, Military Specification, Grease, Aircraft, General Purpose, Wide Temperature Range (2 Aug. 1982). 23. MIL-L-85812, Military Specification, Lubricating Oil, Instrument, Ball Bearing, Synthetic Hydrocarbon (issue pending). 24. Gschwender, L. J., Snyder, C. E., Jr., and Coote, A. A., Jr., Polyalphaolefins as Candidate Coolants in Military Applications, Lubr. Eng., 41, 221, 1985. 25. MIL-H-87252B, Military Specification, Coolant Fluid, Hydrolytically Stable, Dielectric (25 Aug. 1993). 26. Snyder, C. E., Jr. and Gschwender, L. J., Development of a nonflammable hydraulic fluid for aerospace applications over a -54° to 135°C temperature range, Lubr. Eng., 36, 458, 1980. 27. Snyder, C. E., Jr., Gschwender, L. J., and Campbell, W. B., Development and mechanical evaluation of nonflammable aerospace -54° to 135°C hydraulic fluids, Lubr. Eng., 38, 41, 1982. 28. Gschwender, L. J., Snyder, C. E., Jr., and Sharma, S. K., Development of a -54° to 175°C high temperature nonflammable hydraulic fluid MIL-H-53119 for Air Force systems, Lubr. Eng., 49 (1993). 29. MIL-H-53119, Military Specification, U.S. Army, Hydraulic Fluid, Nonflammable, Chlorotrifluoroethylene Base (1 March 1991). 30. MIL-G-25013E, Military Specification, Grease, Aircraft, Ball and Roller Bearing, NATO Code Number G-372 (20 June 1983). 31. MIL-H-8446B, Military Specification, Hydraulic Fluid, Nonpetroleum Base, Aircraft (12 March 1959). 32. MIL-C-47220B, Military Specification, USAF, Coolant Fluid, Dielectric (29 Dec. 1982). 33. Snyder, C. E., Gschwender, L. J., Tamborski, T., Chen, G., and Anderson, D. R., Synthesis and characterization of silahydrocarbons—A Class of Thermally Stable Wide Liquid Range Functional Fluids, ASLE Trans., 25, 299, 1982. 34. Paciorek, K. J. L., Shih, J. G., Kratzer, R. H., Randolph, B. B., and Snyder, C. E., Jr., Polysilahydrocarbon synthetic fluids. I. Synthesis and characterization of trisilahydrocarbons, Ind. Eng. Chem. Prod. Res. Dev., 29, 1855, 1990. 35. Snyder, C. E., Jr., Gschwender, L. J., Randolph, B. B., Paciorek, K. J. L., Shih, J. G., and Chen, G. J., Research and development of low volatility long life silahydrocarbon based liquid lubricants for space, Lubr. Eng., 49, 1993. 36. Gschwender, L. J., Snyder, C. E., Jr., and Driscoll, G., Alkyl benzenes—candidate high-temperature hydraulic fluids, Lubr. Eng., 46, 377, 1990.
Copyright © 1994 CRC Press, LLC
INDUSTRIAL APPLICATIONS Dennis A. Lauer
INTRODUCTION
During the 1930s, it became recognized that the effectiveness of mineral oils could be enhanced with additives. Since additives could only take the performance of mineral oil to a limit, other lubricating base fluids had to be developed. This was the birth of synthetic lubricants. As industries matured and economies thrived, the push for increased production involving more severe service conditions opened the door for synthetic lubricants into many applications. Depending on their chemistry, synthetic fluids may have some outstanding inherent properties such as: • • • • • • •
Low pour point High viscosity index Low vapor pressure Low flammability High flash point High temperature stability and oxidation resistance Extreme pressure (EP) properties • Compatibility with seal materials, plastics, and paint and lacquer finishes
The different synthetics are discussed in detail in other sections of the-Handbook, but are summarized here so that the selection logic for industrial applications is better understood. There are many different synthetic lubricating oils and all have specific advantages and disadvantages. Table 1 compares the properties of the most commonly used base oils. Characteristics of a blended lubricant or grease will be modified by its additive package and/or thickener characteristics. Figure 1 shows the typical temperature limitations of these synthetic base oils as compared to mineral oil. Again, finished lubricant limitations could be significantly different, based on additive and thickener systems. Synthetic lubricants have enabled industry to go beyond the limits of mineral oil lubricants. They have increased productivity, efficiency, and energy conservation. All synthetic lubricants have specific benefits and, when used within their optimal performance range, will provide the most efficient and effective lubricant for the application. The selection of a synthetic lubricant for a specific application must start with the tribological system defining that application, and then matching the performance range of a specific synthetic lubricant to that system. Consequently, synthetic lubricants have been developed to meet the following extreme conditions encountered: temperature, speed, load, sealed-for-life, and environment.
TEMPERATURE EXTREMES
Many manufactured products today require extreme heat, whether in manufacturing, finishing, curing, or drying. These temperatures are often above 150°C, and sometimes can be as high as 1000°C. At these temperatures, mineral oil-based lubricants have a relatively short life or can cause carbon residue build-up.
Textile Industry To transport the fabric through tenter ovens for drying and stretching, both ball bearing roller chains and sliding chains are used at temperatures between 80 to 250°C. Between 120 0-8493-3903-0/94/$0.00 + $.50 © 1994 by CRC Press, Inc.
Copyright © 1994 CRC Press, LLC
195
196
CRC Handbook of Lubrication and Tribology
Copyright © 1994 CRC Press, LLC
Volume III
197
FIGURE 1. Base oil temperature limits.
and 200°C, diesters are used to lubricate the sliding chains with their benefit of low evaporation rate and minimal residue build-up. Above 200°C, polyol esters provide this same type of performance. With the synthetic oils, the tenter requires about one third of the lubricant necessary with mineral oil. Ball bearing roller chains have sealed bail bearings to carry the tenter’s stretching load. These bearings are typically packed with a perfluorinated aliphatic ether (PFAE or PFPE) oil-based grease thickened with PTFE particles. Even at 240°C, only a yearly addition of grease is required. In finishing fabrics, as well as in bonding nonwovens, a calender machine using two or more heavy rolls passes the fabrics under heavy pressure to produce special effects such as high luster, blazing, moiré and embossing. The rolls may be heated up to 200°C by hot thermal oil entering and exiting in the calender roll through the center shaft which therefore heats the inner race of the bearing to the thermal oil temperature. If mineral oil is used to lubricate these bearings, the oil has relatively short life and tends to carbonize, producing excessive maintenance requirements and increased failures. Polyglycol oils are primarily used in a circulating oil system for this application to provide the necessary lubrication and survive the high temperatures for a reasonable period of time. Hot oil entering and exiting the rotating roll in the calender must pass through a rotating union with its ball bearings supporting the rotor. Because the hot oil passes through the center of the bearing, the bearing temperature approaches that of the calender. Depending on the temperature of the hot oil, these bearings can be lubricated with an ester oil based grease up to 150°C. Up to 200°C, greases based on silicone oil can be used. Above 200°C, PFAE greases perform best. In synthetic textile fiber production, the synthetic polymer must first be converted into fiber and further into yam and fabric. There are four basic processes involved: wet spinning, dry spinning, melt spinning, and emulsion spinning. All these processes require extrusion of the melted polymer or polymer solution through a metal disk containing numerous minute holes, called the spinnerette, to form continuous fiber filaments. In the melt spinning process, the spinnerette must be coated with a thin film of lubricant/separating agent to avoid building polymer residues on the spinnerette face and clogging the fine holes. Here a silicone oil provides the best performance since the spinnerette head is extremely hot (250°C).
Copyright © 1994 CRC Press, LLC
198
CRC Handbook of Lubrication and Tribology
Wood Products Industry In wood processing, many applications benefit by the use of synthetic lubricants. One example is the machines that manufacture continuous board, such as medium density fiberboard (MDF), oriented strandboard (OSB), and particleboard (PB). A loosely layered mat of wood particles and fibers mixed with resins is supported between two steel belts and enters this machine at one end and, through the use of high temperatures (up to 240°C) and high pressure, a rigid sheet of wood product exits the other end. The major area of lubrication is the rollers between the hot platens and the steel belts. Because of the heat and load in this contact area, an ester oil is used to minimize residue build-up due to oxidized oil. The chains that drive these rollers, as well as guide chains at the in-feed end of the machine, also experience high temperatures and are lubricated with an ester oil. The chain sprocket bearings, which also experience similar heating, are lubricated with a PFAE grease to provide the longest possible life. Pulp and Paper Industry With many processes that are very similar to the textile industry, high temperature calenders and rotating unions are lubricated in the same way as in the textile industry. Polyglycol is used as a high temperature circulation oil on roll bearings, and ester grease, silicone grease, or PFAE grease is used on the rotating union bearings, depending on temperature. Paper corrugators also benefit from use of synthetic lubricants. The comigator roll bearing, depending on actual operating temperature, can be lubricated periodically with an ester grease thickened with polyurea to minimize residue build-up or with a PFAE base oil thickened with PTFE particles less frequently. Plastic Film Industry Plastic film stretching has several high temperature applications. The process begins with an extruder which extrudes the plastic into sheet or tubular film. The tubular film, or blow film, is forced through a rotating heated die at approximately 240°C. The die bearings are lubricated with a PFAE grease. The sheet film is further processed by a film stretching machine, which operates like a fabric tenter identified earlier. The lubrication points are identical, and the same synthetic ester oils are used on the chains of this machine as on the textile machine. Adjustment spindles which set the width between the two chain sides, thereby establishing the width of the film as it is stretched, are in the heat zone and can be lubricated with silicone grease. Automotive Manufacturing Industry An important part of the automobile manufacturing process is curing of paint and coatings after the metal parts have been finished and assembled. The trolley wheel bearings of the conveyor in this paint-curing oven are exposed to the same high temperatures to which the automotive parts are exposed, up to 250°C. Because it is important not to interrupt production, these high temperature bearings are lubricated with a PFAE grease. With the ability of this grease to withstand high temperature, bearings commonly are regreased annually or even at several year intervals, depending on actual operating temperature. If oil-lubricated overhead conveyor systems are used in similar applications, typically an ester-based oil is used.
Brick and Ceramics Industry One of the most critical machines in manufacturing brick or ceramics is the kiln. Bricks are moved into the kiln on a large, heavy-duty cart running on steel tracks. Since the wheel bearings in these carts approach 1000°C, the fluid portion of any lubricant does not survive. Therefore, a lubricant is selected that will leave little or no carbon residue to prohibit the
Copyright © 1994 CRC Press, LLC
Volume III
199
FIGURE 2. Running time in SKF grease testing machine. (1) Perfluorinated aliphatic ether grease; (2) silicone grease; (3) mineral oil lithium grease.
wheel bearings from rotating. Good performance is obtained with a poly glycol fluid enhanced with solid lubricants in a grease-like paste. The polyglycol fluid evaporates away cleanly at approximately 200°C, leaving the dry lubricant on the bearings for removing the cart from the kiln. Exhaust fans are also necessary to remove hot air and fumes from the process. In order to lubricate the fan motor bearings at the class F motor temperature in a sealed-for-life type application, the best product is a PFAE oil thickened with PTFE particles. Figure 2 shows the comparative analysis of the PFAE grease vs. silicone and mineral oil greases in an SKF grease tester at various temperatures.
Food Processing Industry In industrial baking, me food product is frequently conveyed through the oven in a continuous fashion. For this process, grease-filled sealed bearings used to support the chains driving the oven conveyor must withstand oven temperatures of 200°C and above. At these temperatures, a PFAE grease frequently provides the best performance. In another hot application in food processing, beverage cans are painted or decorated prior to filling and then heat-cured in an oven. To withstand temperatures typically around 200°C during this paint curing, the high speed chains are successfully lubricated with ester type oils for less residue formation than encountered by the oxidation of mineral oil at the high temperature. At the opposite temperature extreme are bearings that must survive in blast freezers. These bearings are subject to continuous temperatures of—40°C and must provide smooth motion and low torque to the mechanism. They are lubricated with silicone or ester greases using thickener systems that provide low apparent dynamic viscosities. Metal Industries Synthetic hydrocarbon and silicone oils are used as die lubes and separating agents for the pressure die casting industry. In this same industry, synthetic base oil pastes are used as ladle dressing. Here, synthetic hydrocarbon oils are mixed with metallic solids and spread on the ladle. The ladle is exposed to the molten metal and the synthetic hydrocarbon oil oxidizes and carbonizes rapidly to form part of the bonding matrix of the protective insulating solids on the
Copyright © 1994 CRC Press, LLC
200
CRC Handbook of Lubrication and Tribology
FIGURE 3. Life expectancy of inhibited lubricants in air.
ladle. Phosphate esters are used as fire-resistant hydraulic fluids in steel mills, foundries, and underground mines, and water-miscible synthetic waxes are used as separating agents in die casting. Synthetic fluids are being used more and more for metal working in water-soluble cutting fluids, grinding coolants, and rolling fluids. Many synthetic fluids are also used to replace the older petroleum-based products with consumption reductions up to 50%, part cleaning costs reductions up to 75%, and increases in tool life in excess of 300% in many metal-working applications. Typically, these synthetic metal-working fluids include amines and glycols as well as special synthetic combinations.
Air Compressors Reciprocating, rotary vane and rotary screw compressors operate at elevated temperatures and provide a broad range of severe conditions for the lubricant. Synthetic oils provide improved efficiency, extended oil life and extended equipment life. The two primary synthetics used in compressor lubrication are organic esters and polyalphaolefin (PAO) synthetic hydrocarbons. Table 2 shows a life comparison in hours of a mineral oil versus a PAO-based rotary screw compressor oil. Ester oils can provide the same benefits and even longer life at higher temperatures. Figure 3 shows the life expectancy of inhibited synthetic lubricants vs. mineral oil in an air environment.
Copyright © 1994 CRC Press, LLC
Volume III
SPEED EXTREMES
201
While synthetic lubricants do not necessarily handle speed better than mineral oils, several characteristics of a synthetic lubricant may improve its performance in high speed applications. Ester oils have a higher viscosity pressure coefficient for example, and therefore provide a thicker film in the contact zone in ball bearings and gears. Also, synthetic oils commonly have higher oxidation resistance in this high speed environment, where there is significant agitation and aeration. The following are a few examples of high speed industrial applications.
Machine Tool Industry Machine tool manufacturers continue to increase spindle speed for more efficient cutting. Bearings for these spindles are typically lubricated with a mineral oil mist which creates a potential health hazard for the operators. As an alternative, many spindle manufacturers are using an ester or polyalphaolefin grease to lubricate these same bearings. When the grease/ bearing system is run-in properly, it can reach similar speeds to the oil-misted bearings and therefore provide better service. Grease-lubricated bearings have also been advantageous for high speed spindles in machining composite materials, such as those used in the aircraft industry, which are very sensitive to mineral oil contamination. With grease-lubricated bearings this oil mist contamination problem is alleviated while enabling spindle speeds of over 50,000 rpm. These same ester and polyalphaolefin greases can be used to lubricate high speed ball screws on these machine tools. High-speed gears in the gear head of multispindle drives are typically lubricated with mineral oil. Because of sealing difficulties, mineral oil can escape the gearbox and leak into the high-speed spindle bearings. This can contaminate and flood the bearings, thus reducing the bearing service life. Via the use of sandwich lubrication with an ester or PAO grease, this problem can be eliminated. The gear teeth are first coated with a bonded dry film lubricant and then lightly lubricated with the synthetic grease. With the proper run-in procedure, the gear drive then can operate at full speed for 6 months to 1 year without relubrication. Textile Industry High speed is frequently experienced in the textile industry. Separator rolls in fiber manufacturing can reach speeds up to 30,000 rpm, and false twist tubes in textile machines have operated at speeds as high as 80,000 rpm. These high-speed bearing applications, as well as many other textile applications, can be successfully lubricated with ester and PAO greases instead of oil mist.
Other Unfortunately, synthetics are not the panacea for all extremes. Oscillation or vibration also require a special lubricant. When surfaces in contact are exposed to vibration or oscillation, there is relative movement between the surfaces that can cause fretting in bearings, couplings, and joints. Fretting and fretting corrosion are considered to be special wear problems resulting in the removal of material. Dr. Fritz Wunsch’s research shows that many synthetic lubricants accelerate fretting and fretting corrosion over mineral oils.
LOAD EXTREMES
There has been specific research to determine if synthetic lubricants will provide less loss of energy than mineral oil lubricants in heavily loaded gears. One specific test was performed using the FZG gear-testing rig to determine the relative friction loss with mineral oil at various viscosities as compared to the frictional loss of a synthetic hydrocarbon fluid and poly glycol.
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
202
FIGURE 4. Gear transmission loss as a function of peripheral velocity.
For the same viscosity grade, a synthetic hydrocarbon fluid and poly glycol gave lower frictional losses than straight mineral oil. With less friction, the synthetics provide less heat, less energy consumption, and a higher efficiency rating for the gear drive. Friction modifiers help mineral oils, but at high speeds the synthetic lubricants significantly outperform the mineral oil lubricants. At the higher speed, the polyglycol and synthetic hydrocarbon have very similar frictional loss. In worm gearboxes with high reduction ratios, polyglycol oil provides a significant advantage over mineral oils in the following performance factors:
• • •
Improved energy efficiency Reduced maintenance, improved reliability, and longer life Increased design ratings
At 60% of rated power, a polyglycol was found to operate approximately 10°C cooler in a worm drive gear than a mineral oil. At 100% rated power, the polyglycol operates at the same temperature as mineral oil operating at 75% rate power. Less heat means less friction, which consumes less energy and produces less wear. Polyglycols effectively reduce friction in high sliding gears, especially for steel on bronze, as indicated in Figure 4.
SEALED-FOR-LIFE
In some applications where it is impractical or impossible to relubricate, the initial lubricant becomes a sealed-for-life lubricant. Also, some applications can be converted from relubricated systems to sealed-for-life systems to increase productivity and minimize contamination. The increased oxidation stability, lower evaporation rate, and higher viscosity index of synthetic lubricants make them more desirable in lubricating a sealed-for-life application than mineral oil. The following are two examples of sealed-for-life applications.
Sintered Bearings Sintered metal plain bearings are used in large quantities in appliances and hand tools where small, economical, ready-to-install bearings are needed. They are also used in the
Copyright © 1994 CRC Press, LLC
Volume III
203
FIGURE 5. Friction behavior of lubricants for sintered plain bearings with increasing speed. (1) Silicone; (2) mineral oil; (3) ester oil; (4) ester fluid grease.
automotive and other industries for small fractional horsepower motors and servo drives. The sintered metal can be iron or bronze, with the lubricant impregnated into the porosity of the metal structure. Since the initial impregnation is generally the only lubrication received over the life of the application, the lubricant should not significantly:
• • •
Increase in viscosity or neutralization number Experience evaporation loss Change in color or form deposits
The best-performing synthetic lubricants in sintered plain bearings are PAOs, ester oils, and PFAE. Even though all these lubricants show excellent performance in sintered bearings, they must be matched with the individual sintered material for optimum performance. For instance, ester oils show excellent friction reduction over other oils, but they are not compatible with many paints and plastics. Depending on the environmental conditions and temperatures, they may hydrolyze or have a catalytic reaction with the sintered metal. Figure 5 shows the change in the coefficient of friction of silicone oil, mineral oil, and ester oil at increasing speed in sintered metal plain bearings. With its high surface area and at high operating temperatures (between 100 and 150°C), frequently the sintered metal can act as a catalyst to increase the rate of oxidation of the lubricant.
High-Speed Spindle Bearings As mentioned earlier for speed extremes, there are advantages to lubricating high-speed spindle bearings with ester greases and polyalphaolefin greases for sealed-for-life applications. Success is contingent upon the proper run-in procedure, which expels excess grease from the bearing and orients the grease structure. There are records of machine tool spindles operating up to the rated life of the bearing on the initial lubrication.
ENVIRONMENT EXTREMES
Because of individual characteristics of specific synthetic lubricants, they can withstand different extreme environments much better than mineral oils. Certain synthetics can survive
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
204
in high vacuums, pure oxygen, acids, caustics, and solvents. The goal of environmental protection also puts extreme requirements on lubricants, such as biodegradability in case of a lubricant spill or disposal. Vacuum Pumps/Diffusion Pumps Depending on the strength of the vacuum, different lubricants must be used. Diffusion pumps create extremely strong vacuums, and the primary synthetic lubricants used in diffusion pumps are PFAE oils. High grade diffusion pump fluids must meet the following requirements: • • • • • • • • • •
Low vapor pressure (less than 109 mbar at room temperature) Good pumping performance Thermal stability in vacuum with no deposit formation up to 350°C (650°F) Chemical resistance to aggressive gases, oxygen, very strong acids and bases Resistance to electron and ion bombardment Corrosion resistance to metallic materials, excellent rust protection No tendency to spread on metal surfaces Compatibility with polymer and metal sealing materials No autoignition with unexpected inrush of air at high working temperature No toxic effect at service temperatures up to 350°C (650°F)
For vacuum pumps that do not require diffusion oils, PFAE oil thickened with PTFE provides excellent performance. Aggressive Media Frequently bearings must operate in very aggressive media, such as pure oxygen, chemical gases, acids, caustics, and solvents. Mineral oils have a very short life in most of these media, and in certain instances can even produce hazardous conditions. Synthetic lubricants are used to lubricate compressors which compress chemically hostile gases. Many chemical plant applications do not allow petroleum-based lubricants due to the possibility of contamination of catalysts. Typical process gases involved are methyl chloride, sulfur dioxide, hydrochloric acid, ammonia, chlorosiloxanes, chlorinated hydrocarbons, and miscellaneous gases containing traces of aggressive contaminants. Polyalphaolefins, silicones, and polyglycols show resistance to these gases to varying degrees. Because of their resistance to dilution with hydrocarbons and less viscosity loss compared to conventional mineral oil of the same viscosity, polypropylene glycols are used in propane production and refrigeration applications. PFAE is essentially an inert lubricant and can also be used in almost all aggressive environments. Biodegradability Since the mid-1980s, interest in biodegradable lubricants has grown in Europe and North America because of increasing environmental concerns and government regulations. Europe is leading the way in biodegradable lubricant development and regulation establishing the 21day biodegradability test CEC-L 33-T 82. A lubricant is considered highly biodegradable if 90% or more of a sample degrades during the test. Typically, mineral oils will degrade 20 to 40%, PAOs 20 to 40%, diesters more than 90%, polyolesters more than 90%, and polyethylene glycols more than 90%. For greases, all currently used thickeners are suitable. Table 3 lists some current additives that have been developed for biodegradable lubricants.
Copyright © 1994 CRC Press, LLC
Volume III
205
REFERENCES
1. Barnes, J. E. and Wright, J. H., Silicone greases and compounds: their components, properties and applications, presented at National Lubricating Grease Institute (NLGI), October 1988. 2. Douglas, P. J., An environmental case for synthetic lubricants, presented at STLE, May 1991. 3. Edwards, D. J., Synthetic lubricants get tougher, Plant Eng., August 18, 1983. 4. Facchiano, D. L. and Johnson, R. L., An examination of synthetic and mineral based gear lubricants and their effect on energy efficiency, presented at NLGI, October 1983. 5. Holzhauer, R., Cross reference product guide for synthetic lubricants, Plant Eng., September 22, 1988. 6. Hunz, R. P., Industrial gear lubricants: their evolution and applications, presented at NLGI, October 1983. 7. Korff, J. and Fessenbecker, A., Additives for biodegradable lubricants, presented at NLGI, October 1992. 8. Kussi, S., Polyethers as base fluids to formulate high performance lubricants, Lubr. Eng., November 1991. 9. Lakes, S., Synthetic gear and transmission lubricants, presented at NLGI, October 1991. 10. Mang, T., Environmentally harmless lubricants, current status and relevant German environmental legislation, presented at NLGI, October 1992. 11. Roberts, W., Lubrication—looking forward to the nineties, S. Afr. Mech. Eng., Vol. 40, June 1990. 12. Synthetic gear lubricant and oil analysis keep critical gear drive in operation, P/PM Tech., September/October, 1989. 13. Tolfa, J., Synthetic lubricants suitable for use in process and hydrocarbon gas compressors, Lubr. Eng., April 1991. 14. Umhoeffer, E., Tribology saving energy and material with special purpose lubricants, Triboschluessel #6, Kluber Lubrication, Munich, Germany, August 1985. 15. van der Waal, G., Properties of ester base fluids and P. A. O. s, presented at NLGI, October 1988. 16. Wunsch, F., Grease starvation lubrication in high-speed spindle bearings, presented at NLGI, October 1990. 17. Wunsch, F., Relationship between the chemical structure of a lubricant and fretting corrosion, presented at NLGI, October 1987. 18. Wunsch, F., Synthetic fluid based lubricating greases, presented at NLGI, October 1990. 19. Yoshizaki, M., Nazuse, C, Nemento, R., and Haizuka, S., Study on frictional loss of spur gears, Tribol. Trans., 34(1), 138, 1990.
Copyright © 1994 CRC Press, LLC
AUTOMOTIVE APPLICATIONS Bruce J. Beimesch
INTRODUCTION AND BACKGROUND
Automotive engineers in recent years have begun to appreciate the important role of the lubricant as an integral part of the total mechanical system. Therefore, more importance and attention are being placed on both the lubricant physical properties as well as its performance. A prominent advantage that synthetic lubricant basestocks enjoy over conventional petroleum basestocks is that they can be tailor-made to do the job. It is possible to build into the molecule such features as low temperature fluidity, low volatility, oxidation stability, and thermal stability.1 For automotive applications, the prominent synthetic basestocks are synthetic hydrocarbons and esters. Synthetic hydrocarbons are generally synonymous with polyalphaolefins (PAO), and their structure most closely resembles isoparaffinic hydrocarbons. They are derived from oligomerizing alpha olefins such as decene-1, which is derived from the petrochemical feedstock ethylene. Figure 1 depicts idealized structures for petroleum base stock and PAO. Also shown are generic structures for diesters and polyol esters. Esters generally fall into two categories: diesters and polyol esters. The prominent esters used for automotive applications are the diesters of adipic and azelaic acids and the polyol esters of trimethylolpropane and pentaerythritol. Table 1 shows a few typical examples from the numerous possibilities that esters offer. Combinations of esters and PAO synthetic hydrocarbons are becoming the choice for many automotive applications because they offer an optimum balance of properties such as: additive solubility, sludge control, and elastomer compatibility. Structural similarity of PAO and petroleum oil make them similar in response to performance additives and packages. However, occasionally additive solubility problems are encountered with PAO at room temperature storage or at moderately low temperatures. Esters blended with PAO at moderate levels (10 to 30%) usually overcome this problem. Additionally, esters, because of their polar nature, behave as solvents for sludge and varnish. For example, esters blended into petroleum hydrocarbon or PAO at 10 to 30% commonly provide enhanced component cleanliness. Straight ester systems will also perform very well in high temperature applications. This is in part attributed to their excellent response to ashless antioxidants. Ashless antiwear additives such as phosphate esters show excellent response in both PAO and esters for many applications. For automotive engine oils, the zinc dialkyldithiophosphate additives are used. Due to their polarity, esters tend to adhere more closely to the metal surface and provide a higher film strength as well as a more tenacious barrier. Some rebalancing of the zinc additive system may be needed in order to provide a greater tendency for surface absorption in an ester environment.2 The 100% ester synthetic systems may require special elastomer seal consideration. High molecular weight esters or nonpolar esters are similar to mineral oil in seal swell. However, lower molecular weight esters tend to be more polar and can plasticize many of the common elastomers such as nitrile and polyacrylate.2 For these systems, medium to high acrylonitrile (>32%)-containing nitrile seals or fluoroelastomer seals are recommended. PAO tends to shrink some elastomers such as nitrile or polyacrylate or not swell as much as their mineral oil counterpart. Blends of PAO with ester (10 to 30%) often prevent seal shrinkage. In summary, achieving a balance of lubricant properties is now possible by tailoring and blending synthetic base stocks. The following application areas highlight the utility and importance of synthetic ester and PAO to the automotive area. 0-8493-3903-0/94/$0.00 + $0.50 © 1994 by CRC Press, Inc.
Copyright © 1994 CRC Press, LLC
207
CRC Handbook of Lubrication and Tribology
208
FIGURE 1. Idealized structures for petroleum base stock, PAO and esters.
AUTOMOTIVE ENGINE OILS
Passenger Car Motor Oils The trend since the 1970s has focused on fuel efficiency improvements for both the gasoline and diesel engine. The CAFE (corporate average fuel economy) standard, instituted in 1974, has required significant improvements in fuel efficiency for passenger cars as given in Table 2. The 1990 CAFE requirement of 27.5 mpg is truly a remarkable increase when compared to just a decade ago. This increased fuel economy mandated by federal law has not only the Copyright © 1994 CRC Press, LLC
Volume III
209
effect of conserving crude oil reserves but also of reducing engine pollutants emitted to the atmosphere. Concern for fuel economy and the high price of fuel brought about the greater use of 4cylinder overhead cam engines and lower viscosity oils like SAE 5W-30. Additionally, the concerns with the low temperature engine cranking ability and oil pumping ability of the overhead cam engines helped to promote the usage of the SAE 5W-30 oils. In 1990, SAE 5W30 oils accounted for approximately 8% of the 750 million gallons of passenger car oil sold in the U.S. All of the U.S. original equipment manufacturers (OEMs) recommend the use of SAE 5W-30 oils. There are, however, major OEM concerns with SAE 5W-30 petroleum-based motor oils in the areas of volatility and loss of oil viscosity caused by the shearing of viscosity index improvers. Table 3 shows data for volatility and high temperature high shear (HTHS) viscosity for U.S. commercial SAE 5W-30 oils. The tapered bearing simulator measures dynamic viscosity at 150°C and 106 s-1 shear rate (ASTM 4683) which is expected in the bearings of today’s engines under severe operating conditions. A minimum HTHS viscosity of 2.9 centipoise was established by the OEMs as necessary to ensure the oil film thickness essential for bearing protection. A revision of SAE J300 engine oil classification is under consideration and may be adopted. Some commercial engine oils available in 1990 did not meet the more stringent requirements. The International Lubricant Standardization and Approval Committee (ILSAC) established by the OEMs in 1989 expressed concerns with oil quality in general, and in particular ILSAC continued the drive toward fuel efficiency and catalytic converter longevity. ILSAC proposed a new oil standard, GF-1, to effect improvements in these areas:
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
210
• • • • •
API SG engine performance requirements Shear stability/high temperature, high shear viscosity NOACK volatility or GLC method simulated distillation Fuel efficiency/energy conserving II Phosphorus at 0.12% maximum
To meet the requirements of SAE 5W-30 and GF-1 requires tight control of the petroleum refining operation and generally narrow cuts of 100 neutral oil stocks. The desired balance of high temperature and low temperature properties will be increasingly more difficult to meet with conventional petroleum-base oils alone. By blending 100 neutral oils with PAO or esters, improvements in both low temperature and high temperature properties are possible (see Table 4). Replacement of the 90 to 95 viscosity index 100 neutral oil stocks with the higher viscosity index (120 to 160) synthetic stocks allows use of more shear-stable viscosity index improvers which will pose less viscosity loss during service. An additional possible benefit which needs further research is the response of low level (10 to 30%) replacement of mineral oil with PAO or ester to achieve fuel economy benefits. It has been found, however, that low level addition (0.5 to 2.0%) of glycerol monooleate or pentaerythritol monooleate will yield fuel economy improvements of 0.5 to 2.0% by acting as friction modifiers.3,4
Full Synthetic Engine Oils Full synthetic gasoline engine oils offer even greater advances in lower volatility, potential for longer drain intervals, better low temperature properties, and improved fuel economy. Full synthetic arctic grade 20 oils were developed in the mid-1960s to service the construction of the Alaskan pipeline for use down to -55°C. The U.S. Army adopted military specification MIL-L-46 167 in 1974 (see Table 5) for servicing vehicles in cold climate or in special applications for hydraulics systems and transmissions.5,6 The original arctic fluids were based on di-2-ethylhexylazelate or di-2-ethythexylazelate blended with dialkylbenzene. Today these fluids are a balanced blend of basestocks consisting of 70% PAO and 30% azelate diester.
Copyright © 1994 CRC Press, LLC
Volume III
211
Commercial fully synthetic SAE 5W-30 engine oils were developed in the mid-1970s in the U.S. Current formulations meet the requirements of API SG or SG/CE. Emerging in Europe are other viscosity grades which utilize combinations of PAO plus ester. The strong European trend towards high quality synthetic oils is actively supported by the prime European OEMs. The ACEA (Association of European Automotive Manufacturers), formerly the CCMC (Committee of Common Market Automobile Constructors), has issued their service fill classifications of G-4 and G-5 for gasoline engines.7 The high speed driving, small engines, and turbocharging prevalent in Europe all combine to put a high degree of performance stress on the oil. The G-5 requirements covering the viscosities of SAE 5W-X and SAE 10W-X are such that only partial or full synthetic oils will be suitable in many cases. Fully synthetic oils that meet SAE 5W-50 and SAE 10W-60 requirements are based on PAO and esters. Polyol esters are generally chosen for their increased stability and low volatility. A comparison of the key physical properties for ILSAC GF-1 (5W-30 only) and ACEA G-4/G-5 are found in Table 6. The European ACEA requirements are more stringent than those proposed by ILSAC. As further harmonization of the auto industry occurs, pressure will be exerted to develop a clearly definable worldwide oil specification. Diesel Trends The heavy duty diesel area is extremely active in the U.S. with major restructuring of the API “C” category. Table 7 illustrates the proposed new CF, CF-2, and CF-4 categories and their performance test requirements. These revisions result from a combination of fundamental
Copyright © 1994 CRC Press, LLC
212
CRC Handbook of Lubrication and Tribology
engine design changes, oil consumption concerns, increased temperature of operation, lowering particulate emissions to meet federal environmental regulations, higher soot loading in the oil, and deposit control in cylinder area. The EPA is addressing the issue of diesel engine particulate emissions with two rounds of significant tightening in 1991 and 1994. The European Community has also adopted similar emission standards. Table 8 shows the U.S. emission standards. The major sources for particulates are: “sulfur” compounds from diesel fuel, “soot” from incomplete combustion, and lubricating oil.8,9 Sulfur content of No. 2 diesel fuel will drop from a typical level of 0.25 to 0.05% in 1994. The 1991 federal regulations were met by a combination of engine design changes, certification of low sulfur (0.1%) fuel, and by using API CF quality oils. The 1994 federal standards are so severe that the current level of particulates coming from just the oil alone would exceed the 1994 standard. A combination of engine design changes, particle traps, and catalytic converters will help to meet the standards, and improvements in oil volatility also may be required. This opens up another major area for the use of esters and PAOs as partial and full synthetic diesel engine oils. The ability of the oil to disperse high levels of soot will also become a critical issue. More burden will be placed on dispersant additives to provide the cleanliness needed to maintain the current drain intervals. Polar esters may provide assistance in soot suspension. Engine Oil-Diesel Partial Synthetics Properties for partial synthetic SAE 15W-40 engine oil containing 30% synthetic base stock are compared with a typical petroleum engine oil in Table 9. During field testing in Cummins NTC 350 engines, the partial synthetic 15W-40 oil maintained the SAE 40 viscosity, while the petroleum 15W-40 sheared to SAE 30 (fuel dilution was not a factor). Additionally, the partial synthetic 15W-40 had oil consumption averages of 2000 to 3000 miles per quart throughout the drain interval of 30,000 miles. The petroleum units averaged less man 2000 mi/qt during the 400,000-mi test. Particulate emissions were not measured in this current test; but based on the previous oil consumption
Copyright © 1994 CRC Press, LLC
213
Volume III
data and the correlation of oil consumption with particulate emissions established independently by an additive company,9 there is a strong belief that particulate reduction is possible.
Diesel Full Synthetic Engine Oils and Low Temperature With the addition of electronic assist starting and automatic ether aid injection, diesel engines will be able to start in cold climates—whether the oil flows or not. Additionally, a common practice in the northern U.S. and Canada is to use plug-in oil heaters during winter months to assist in engine starting. As a general principle, oil pumpability is more critical at low temperatures than crankability. Oil must be pumped in order to protect the engine during and immediately after cranking and firing. Based on winter field testing with Cummins NTC 300 BCII engines in Fargo, North Dakota, Table 10 compares starting performance of SAE 15W-40 petroleum oil representing the most commonly used fleet oil, and a SAE 5W-30 full synthetic oil.10 The petroleum unit at -4°F oil temperature was eventually started with a “start-all” machine. Because of high oil viscosity and pressure, the unit blew the O-ring on the lube refiner and ruptured the oil filter gasket. The synthetic unit started at -10°F oil temperature after cold soaking in a blizzard for 75 hours.
SYNTHETIC GEAR AND TRANSMISSION OILS
Commercial Gear and Transmission Oils API GL-5 gear oil service category is used as the guideline by the majority of OEMs for passenger car, light truck, and heavy duty truck service. The current MIL-L-2105D specification, essentially equivalent to the API GL-5 classification, is described in Table 11.
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
214
The continued down-sizing, aerodynamic styling, and the drive for greater axle and manual transmission efficiency have resulted in higher temperature demands on the oil. API categories of PG-1 for manual transmission oil and PG-2 for heavy duty gear oil will guide oil formulations to more thermally and oxidatively stable systems. Synthetic gear and transmission oils were first formulated in the early 1970s, and their field performance has been well documented.11 Synthetics offer advantages in extended drain intervals with less oil disposal, improved fuel efficiency, higher temperature stability, and cleanliness. Full synthetic gear and transmission oils have OEM recognition and approval and their longer life capability is reflected in the following typical drain recommendations for heavy duty truck fleet service:
• •
Petroleum gear oil—50–100M miles Synthetic 75W-90—250–500M miles
Synthetic gear and transmission oil testing has been carried to 1,000,000 miles in heavy duty truck service without a drain and with excellent results. Little or no evidence of oxidation was encountered, the systems were clean, and the parts were in like-new condition. No wear was found in the transmission shift collars and bearing. The ring and pinion integrity was excellent in the gear boxes. In the future a single oil may lubricate both the manual shift transmissions and axle gear box under sealed-for-life conditions in any application. Truck cab and chasis designs will continue to limit the amount of cool air flow in order to improve aerodynamics for better fuel economy. More thermal stress will be placed on the lubricants. Synthetics offer distinct advantage in stability and cleanliness. The improved thermal and oxidation stability of synthetic gear and transmission oils can be demonstrated by the L-60 thermal oxidation and stability test (see Table 12). Copyright © 1994 CRC Press, LLC
215
Volume III
FIGURE 2. Fuel efficiency of EP 75W-90 synthetic gear lubricant.
Fuel Savings The testing of synthetic 75W-90 gear oils by a variety of fuel-efficient tests indicates that potential fuel savings of 1 to 2% are achievable (Figure 2). The RCCC type 1 is a two-truck test method where the synthetic oil was compared to a petroleum oil control and then switched so that each truck was operated with both the synthetic and petroleum oils. The AET is the axle efficiency test designed to determine the operating efficiency of automotive axle assembly and their lubricants. The axle is a Ford 7 1/2” power divider with a 3.08 axle ratio operated to approximate the conditions found in the EPA city/highway emission and fuel economy sequence. The Maxwell dynamometer rig is a full chasis dynamometer. Three trucks with different mileage (see Figure 2) were compared using petroleum SAE 90 gear oil control. After the oils to be tested are brought up to temperature in 20 min at 56 mph, fuel consumption is measured for a set time. The Forest Engineering Research Institute of Canada (FERIC) established that significant savings were achievable with synthetic gear oil because of their improved viscositytemperature relationship and therefore less parasitic energy losses due to churning and resistant forces.12 The significance of fuel efficiency can be quite dramatic. Improvement of 1% for a fleet of 100 tracks operating at 100,000 miles per year equates to a fuel savings of approximately 20,000 gallons.
AUTOMATIC TRANSMISSION FLUIDS
OEMs have expressed concern for both high and low temperature operation of automatic transmissions. Ford has addressed the low temperature operation problem with specification M2C163A2, a synthetic DEXRON® II fluid. Ford also has established MERCON® as a worldwide specification which replaces DEXRON®, Ford type CJ, or Ford type H fluids. General Motors is addressing their concerns by their commitment to DEXRON® HE, a new fluid with improved shifting feel, enhanced antiwear, better oxidation stability, moisture resistance, and low temperature fluidity. Synthetic transmission fluids can be designed with blends of PAO/ester and commercially available additives which will meet MERCON®, DEXRON®, and M2C163A2. Table 13
Copyright © 1994 CRC Press, LLC
216
CRC Handbook of Lubrication and Tribology
compares the viscosity properties for a synthetic ATF and petroleum-based DEXRON® II. High temperature performance will be critical to the operation of future automatic transmissions due to low air flow, smaller sump sizes, and smaller, lighter components. The GM THOT (Turbo Hydromatic oxidation test) and the Ford ABOT (aluminum beaker oxidation test) were developed to test for oxidation and thermal stabilities of ATFs. A comparison of the results with typical fluids are summarized in Table 14. The tests were terminated when any one limit was exceeded. The results are summarized in Table 15 as hours to failure. Copyright © 1994 CRC Press, LLC
217
Volume III
The synthetic fluid life was twice that for the petroleum fluid. The synthetic fluid showed enhanced cleanliness even after 800 hours in the THOT. The THOT test is usually run for only 300 hours.
Lubricants and the Environment Lubricants which are designed to minimize the impact to the environment can be placed into two categories. Passive types are lubricants that exhibit extended drains, thereby requiring fewer oil changes and therefore there is less oil earmarked for disposal. This is certainly the situation today with long-drain synthetic gear and transmission oil. Passive types can also be fuel-efficient engine oils that reduce energy consumption with less emissions to environment. Synthetic engine oils and synthetic gear oils would fit into this category. Active types are lubricants that would biodegrade, be nontoxic, and have no long-term detriment to the environment. Obviously, any lubricant spill has a potentially direct insult to the environment and should be treated as such. It is inevitable that lubricants will reach the environment from either leakage or spills. Hydraulic lines will at times rupture, wheel seals and throughput shaft seals will leak. Automatic grease systems deliver grease on a time basis; where does all the grease go? In Europe the CEC (Coordinating European Council) developed the CEC L-33-T-82 biodegradability test for two-stroke engine oils which is gaining acceptance for testing other lubricants. It is a 21-day test in which the disappearance of oil is measured after being inoculated with activated sewage sludge. Lubricants giving results greater than 80% are considered to be readily biodegradable. Table 16 summarizes several synthetic and petroleum base stocks. The actual percentage can vary depending on source of the sewage sludge and other factors, but in general esters tend to be readily biodegradable, while hydrocarbons exhibit more resistance to degradation. Two-stroke readily biodegradable fluids already have been developed based on esters. Ester-based greases developed in the 1950s represent the beginnings for building a biodegradable grease. Biodegradable gear oil and engine oils will provide a more difficult challenge.
SUMMARY
The role of lubrication in automotive vehicles is going through rapid and significant changes. The equipment is being designed to be smaller and more efficient. Vehicles are becoming more aerodynamic. There are concerns for the environment. The consumer, whether an individual or a fleet operator, is demanding long, trouble-free life for his equipment. The engineers and manufacturers have recognized that the lubricant is an integral part of a total system. More OEMs are requiring or assigning lubricant specifications tailored to either full or partial synthetic lubricants for engine, drive-train, and auxiliary equipment lubricants. These changes will contribute to the increasing use of synthetic lubricants in the world market. Copyright © 1994 CRC Press, LLC
218
CRC Handbook of Lubrication and Tribology
REFERENCES
1. Boylan, J. B., Beimesch, B. J., and Schnur, N. E., Synthetic Lubricant Basestocks from Monohydric Alcohols, Ser. 159, American Chemical Society, Washington, DC, 1981. 2. Van der Waal, G., The relationship between the chemical structure of ester base fluids and their influence on elastomer seals, and wear characteristics, J. Synth. Lubr., 1(1), 1985. 3. U.S. Patent 4,376,056 and 4,584,112; Chevron Research Company. 4. U.S. Patent 4,175,047; Mobil Research and Development Corp. 5. Lestz, S. J. and Bowan, T. C, Army experience with synthetic engine oils in mixed fleet Arctic service, SAE Pap., No. 750685, 1975. 6. Lestz, S. J. and Owens, E. C, Army Arctic engine oil performance in high ambient temperatures, SAE Pap., No. 892051, 1989. 7. Cahill, G. F, Evolution of the CCMC engine lubricant sequence, in CEC Third Int. Symp. Performance Evaluation for Automotive Fuels and Lubricants, Coordinating European Council (CEC), London, 1989. 8. Richards, R. R. and Sibley, J. E., Diesel engine emissions control for the 1990’s, SAE Pap., No. 880346, 1988. 9. Cooke, V. B., Lubrication of low emission diesel engines, SAE Pap., No. 900814, 1990. 10. Margeson, M. A. and Beimesch, B. J., Cold starting capabilities of petroleum and synthetic lubricants in heavyduty diesel engine, SAE Pap., No. 890994, 1989. 11. Beimesch, B. J., Margeson, M. A., and Davis, J. E., Field performance of synthetic automotive gear lubricants, SAE Pap., No. 831730, 1983. 12. Lijbic, D. A., Forest Engineering Research Institute of Canada, Tech. Rep., No. TR-55, April 1984.
Copyright © 1994 CRC Press, LLC
POLYALPHAOLEFINS Ronald L. Shubkin
INTRODUCTION
Recent technological advances have been accompanied by a variety of problems relating to the satisfactory use of existing functional fluids and lubricants. Among these are
• • • • •
Severity of operating conditions Need for cost-effective operations Need to lessen dependence on availability of crude oil stocks Specialized end-use requirements Toxicological and biodegradability considerations
Today, mineral oil (M.O.) base stocks are being refined to yield products that are superior to those available only a few years ago. In addition, improvements in additive technology have helped push the performance capabilities of mineral oils to increasingly higher levels. In many applications, the primary function of a working fluid is lubrication, but heat transfer, power transmission, electrical insulation, and corrosion inhibition are other tasks that fluids may be expected to perform. In many instances the performance requirements exceed the capabilities of conventional fluids, and a wide variety of synthetic, high-performance functional fluids have been developed. The advantage of the synthetic fluid is often in its ability to perform in more extreme environments or more cost-effectively than conventional products in a given application. In other cases, safety, environmental, or toxicological considerations have mandated the use of particular synthetic functional fluids. For additional information, a comprehensive review has been published which contains detailed information on all of the major synthetic lubricants including their historical development, chemistry, manufacture, performance characteristics, applications, producers, and markets.1
SYNTHETIC HYDROCARBONS
Synthetic hydrocarbon fluids may be characterized by the fact that these fluids are comprised of molecules that contain only carbon and hydrogen atoms. Members of this class that are suitable for use as lubricants and functional fluids include polyalphaolefins (PAOs),2 polyintemalolefins (PIOs), alkylated aromatics,3 polyisobutenes (PIBs),4 and cycloaliphatics.5 Highly refined mineral oils (such as the high viscosity index (HVI), very high viscosity index (VHVI), and ultra high viscosity index (UHVT) fluids are sometimes referred to as synthetic hydrocarbons, but this is not an accurate description. The HVI, VHVI, and UHVT fluids are the products of a refinery. Conversion of crude oil to these products involves a variety of separation (distillation, solvent extraction, dewaxing), degradation (cracking, hydrocracking), rearrangement (isomerization), and hydrogenation (hydrotreating) techniques. The final products contain a large variety of different molecular types. By contrast, synthetic hydrocarbons are built up from specific starting materials to form molecularly well-defined products.
POLYALPHAOLEFIN CHARACTERISTICS
Polyalphaolefins (PAOs) are gaining rapid acceptance as high-performance lubricants and functional fluids because they have certain inherent, and highly desirable, characteristics relative to mineral oils. Among these favorable properties are: 0-8493-3903-0/94/$0.00 + $0.50 © 1994 by CRC Press, Inc.
Copyright © 1994 CRC Press, LLC
219
CRC Handbook of Lubrication and Tribology
220
• • • • • • • • • • • • •
FIGURE 1. Gas chromatography of typical PAO reaction product.
A wide operational temperature range Good viscometrics (High Viscosity Index) Thermal stability Improved response to conventional antioxidants Hydrolytic stability Shear stability Low corrosivity Compatibility with mineral oils Compatibility with various materials of construction Low toxicity Good to moderate relative biodegradability Low deposit formation Can be “tailored” to specific end-use application requirements
Manufacture PAOs are manufactured by a two-step process from linear alpha-olefins, which are themselves manufactured from ethylene. The first synthesis step entails oligomerization, which simply means a polymerization to relatively low molecular weight products. For the production of low viscosity (2 to 10 cSt) PAOs, the catalyst for the oligomerization reaction is usually boron trifluoride. (Note: PAOs are commonly classified by their approximate kinematic viscosity (KV) at 100°C. That convention will be used throughout this chapter. Thus, a fluid referred to as PAO 4 has a viscosity at 100°C of ca. 4 cSt.) The BF3 catalyst is used in conjunction with a protic co-catalyst such as water, an alcohol, or a weak carboxylic acid. The BF3 ROH catalyst system is unique because of its ability to form highly branched products with the oligomer distribution peaking at the trimer. Figure 1 shows a gas chromatography (GC) trace indicating the oligomer distribution of a typical reaction product Higher viscosity (40 and 100 cSt) PAO is manufactured using alkylaluminum catalysts in conjunction with organic halides. Copyright © 1994 CRC Press, LLC
221
Volume III
FIGURE 2. GC trace of PAO 4 vs. equiviscous VHVI mineral oil. The second step in the manufacturing process entails hydrogenation of the unsaturated oligomer. The reaction is carried out over a metal catalyst such as nickel or palladium. Hydrogenation gives the final product enhanced chemical inertness and added oxidative stability. One of the distinct advantages in the manufacture of PAOs is that they can be “tailormade” to fit the requirements of the end-use application.7 This customizing is done by manipulation of the reaction variables which include:
• • • •
Chain length of olefin raw material Temperature, time, and pressure Co-catalyst type and concentration Distillation of final product
The PAO manufacturer can make major alterations in the product properties by choice of starting olefin. Today, the commercial PAO market is dominated by decene-derived material because these products have the broadest temperature range of desirable properties, but a knowledgeable producer has the option of choosing other starting olefins in order to better satisfy the requirements for a particular end-use application. The physical properties of PAOs prepared from olefins other than 1-decene will be discussed in the “Physical Properties” section of this chapter. Figure 2 compares the GC trace of a commercial 4.0-cSt PAO (PAO 4) with that of a hydrotreated VHVI mineral oil base stock having the same approximate viscosity at 100°C. The trace from the mineral oil shows that it consists of a broad range of different kinds of molecules. Included are low molecular weight materials that adversely affect volatility and high molecular weight components which adversely affect low temperature properties. By comparison, the PAO 4 is primarily decene trimer, with small amounts of decene tetramer and pentamer present. The effects of the differences in composition between PAOs and mineral oils are shown in tables that are included in the “Physical Properties” section of this chapter. The fine structure in the trace on Figure 2 is attributable to isomers of the different oligomers (Note: Oligomers are low molecular weight polymers such as dimers, trimers, etc. Isomers are molecules with identical formulas and molecular weights, but with different skeletal structures.). A knowledge of reaction variables can be used to control the relative abundance of the various isomers and provides the producer another method to influence the physical properties of the final product.8
Copyright © 1994 CRC Press, LLC
222
CRC Handbook of Lubrication and Tribology
Physical Properties Table 1 lists the physical properties of various grades of PAO fluids. These products are produced from decene, and the differences in properties illustrate what can be accomplished by manipulation of the reaction parameters. Some of these products are co-produced and separated by distillation. The properties are typical of what is currently available and do not represent the specifications of any particular producer. More detailed physical property data are given in the Appendix to this chapter. Table 2 is a brief listing of the physical properties of PAO fluids prepared from different olefin raw materials. Each of these fluids was prepared using the same recipe (not necessarily a commercial recipe), which included distilling off the dimer product and hydrogenating final fluid.9 None of the fluids in Table 2 are offered commercially. Table 3 compares the physical properties of a commercial 4.0-cSt PAO with those of two conventional 100N “neutral” mineral oils, a 100NLP (low pour) mineral oil, and a hydrotreatd VHVI mineral oil. The PAO shows markedly better properties at both high and low temperatures. At high temperatures, the PAO has lower volatility and a correspondingly higher flash point. Low volatility is an important property in order for a fluid to “stay in grade” (i.e., retain original viscosity) during its working life. At the low end of the temperature scale the differences are equally dramatic. The pour point of the PAO is less than -65°C, while the three 100N mineral oils and the HVI oil are - 15, -12, -15, and -27°C, respectively. Copyright © 1994 CRC Press, LLC
223
Volume III
Table 4 compares a commercial 6.0-cSt PAO with a 160HT (hydro-treated) mineral oil, 240N oil, a 200SN (solvent neutral) mineral oil, and a UHVI fluid. The broader temperature range of the PAO is again apparent. Table 5 makes similar comparisons for 8.0-cSt fluids.
END-USE APPLICATIONS FOR PAOs
The rapid growth of PAO-based functional fluids arises from two sources. The first factor of the increasing volumes required by the conventional markets—such as automotive crankcase applications. The second factor is attributed to new areas where specific properties
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
224
of the PAOs give them a particular advantage in performance, cost-effectiveness, or environmental acceptability. Table 6 is a compilation of both established and emerging application areas for PAOs. The table indicates those properties which make PAOs especially well-suited for specific applications. In addition to those applications noted in Table 6, decene oligomers are used as lubricants in food processing and as high quality emollients in the cosmetic and personal care industry. The latter materials are specially produced under the conditions of high quality assurance appropriate for the industry. The fluids are generally known as “polydecenes” and should not be confused with the less expensive industrial grade polyalphaolefins.10
PERFORMANCE TESTING
Automotive Crankcase
While physical properties are obviously important in choosing a fluid for a particular application, it is essential that the fluid be subjected to performance testing under conditions that simulate the limits to which the final product will be stressed. Because the requirements for the wide variety of automotive applications encompass much of the broader spectrum of applications, this section will focus on tests specifically designed and conducted by the automotive industry. The following tables summarize the results of a battery of tests designed to compare the performance of PAO-based fluids and fully formulated products with their mineral oil-based counterparts. Unless otherwise referenced, the information presented in this section represents unpublished data obtained by or for the author’s company, Ethyl Corporation. Tables 7 through 10 illustrate the results of tests related to use in automotive crankcase applications. Table 7 contains data relating to the hot oil oxidation test (HOOT), which is designed to measure the thermal and oxidative stability of a bulk fluid under the severe oxidizing conditions in an automotive oil sump. A PAO and a mineral oil were compared employing identical additive packages at identical concentrations. Air is bubbled through 25 g of the test oil at a rate of 10 1/h for 5 d at 165°C. The oil contains 178 ppm iron(III) acetylacetonate and 17 ppm copper(II) acetylacetonate as oxidation catalysts. The superior performance of the PAO has three possible implications. First, the PAO-based fluid can be used for longer drain intervals, resulting in less down time and lower maintenance costs. Second, PAO can be used with lower levels of additives and other stabilizers, thus reducing the price differential between the PAO and the mineral oil. Finally, a PAO-based fluid can tolerate Copyright © 1994 CRC Press, LLC
Volume III
225
higher operating temperatures. The concept of extended drain intervals with PAO-based fluids has been controversial. Part of the problem has been the depletion of essential additives prior to significant degradation of the PAO base stock. New additive technology, however, is addressing this question. Table 8 contains the results of the Petter W1 engine test after 108 h. The test measures both the increase in viscosity of the fluid and the amount of wear, as determined by bearing weight loss. In this test, the advantages of employing a part-synthetic oil mixture are shown. When PAO is used as only 25% of the base oil, the degradation as shown by the percent viscosity increase is half that of the mineral oil fluid without PAO. The data in Table 9 were acquired from a sequence IIIE engine test, which is commonly used in North America. The sequence IIIE evaluates an engine oil for its ability to minimize
Copyright © 1994 CRC Press, LLC
226
CRC Handbook of Lubrication and Tribology
high temperature oxidation and thickening, sludge and varnish deposits, and wear. Table 10 contains the results of the more severe sequence VE test conducted on the same formulations as in the IIIE. In both the IIIE, and the VE tests, both formulations meet the specification limits for API SG classification. However, the full-synthetic oil has an SAE SW50 classification, whereas the mineral oil formulation is a 15W40, indicating a more limited operating range. Table 11 contains data relating to the VW Digifant test, which is more widely used in Europe. Another important aspect that must be considered for automotive crankcase applications is low-temperature performance. Table 12 compares the low-temperature characteristics of Copyright © 1994 CRC Press, LLC
Volume III
227
base fluid PAOs with HVI and VHVI mineral oils of comparable viscosity. The cold crank simulation test demonstrates the advantage of a PAO-based formulation in the crankcase on a cold winter morning. The results of a Caterpillar 1 -G evaluation are given in Table 13. This is a diesel detergency test for high speed, severe supercharged conditions. Both a part-synthetic and a full-synthetic PAO-based oil outperform an equivalent 10W40 mineral oil. One final advantage for the use of PAOs in the formulation of automotive crankcase oils should be noted. This might be termed the “theological advantage”, which works in several ways. PAOs have become an important factor in helping formulators meet the increasingly difficult viscosity requirements for cold-weather oils such as the 5W and even 0W grades while maintaining volatilities at the required low levels. Figure 3 is a graphical representation of the effect of blending a 4.0-cSt PAO with a 100N mineral oil. Small amounts of the PAO have a significant influence in lowering the volatility, but there is virtually no change in the viscosity regardless of the percentage of PAO in the blend. Figure 4 shows the effects of blending a 4.0-cSt PAO with a 200N mineral oil. In this case small amounts of PAO have an important influence in lowering the viscosity without increasing the volatility.
Automotive Transmissions HOOTs are also used to screen oils for use in manual transmissions and rear axles. The test is conducted at a more severe temperature (200°C), and the kinematic viscosity at 100°C is measured at specified time intervals. A comparison of the performance of mineral oil and PAO-based fully formulated oils is shown on Table 14. After 16 h, the viscosity of the PAO fluid has increased only about 19%, whereas the viscosity of the mineral oil fluid has increased nearly 500%. After 24 h, the viscosity of the PAO fluid has increased by only 21%, but that of the mineral oil product has become too viscous to measure. Similar to its use as a screening tool for manual transmission fluids, the HOOT is used as an indicator of performance for automatic transmission fluids. A lower viscosity oil is used in this application, and it in fact performs better than the manual transmission formulation at 200°C. The results of this test are presented graphically in Figure 5. The PAO-based formulation shows only an 8.6% increase in viscosity (100°C) after 24 h, the mineral oil formulation increases 550%. While the tests described above indicate that PAO-based transmission fluids show better durability and performance than mineral oil-based fluids at a given temperature, another important phenomenon has been reported. Measurement of transmission lubricant temperatures under high-speed driving conditions show that the synthetic-based oils run as much as 30°C cooler than their mineral oil counterparts.11Lower operating temperatures can lead to longer seal and bearing life. Conversely, lower operating temperature can allow equipment designs requiring less cooling capacity.
Copyright © 1994 CRC Press, LLC
228
Copyright © 1994 CRC Press, LLC
FIGURE 4. Effect of blending PAO 4 with 200n mineral oil.
CRC Handbook of Lubrication and Tribology
FIGURE 3. Effect of blending PAO 4 with 100n mineral oil.
229
Volume III
FIGURE 5. Hot oil oxidation test (automatic transmission, 200°C).
Automotive Gear The Mercedes Benz spur gear rig performance test is used to evaluate the performance of gear oils. In the test, the time to gear tooth breakage is used as the indicator of performance. A SAE 75W90 synthetic formulation showed a 60% improvement over a SAE 90 mineral oil. The data are presented in Table 15.
Industrial Gear The use of PAO-based gear oils in industrial settings can lead to important savings in energy consumption as well as decreased down-time and lower maintenance requirements. The wide range of operating temperatures allows the use of lighter oils, and the use of lighter oils results in greater energy efficiency. Table 16 is a compilation of data from ten reports on increased efficiency found when industrial transmissions were switched from mineral oil to PAO-based gear oils. The increases ranged from 2.2 to 8.8%. It is interesting that the efficiency increase observed in worm gears correlates closely with the reduction ratio. This correlation exists despite the fact that the data were reported by different companies and were collected on different types of equipment.12
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
230
The literature contains a number of reports of monetary savings directly attributable to a switch from mineral oil to PAO-based gear or bearing oils. Table 17 is a compilation of published reports.13
HEALTH AND ENVIRONMENTAL ISSUES
Toxicology Acute oral toxicity tests on 2.0-, 4.0-, 6.0-, 8.0-, and 10.0-cSt PAO were conducted on rats by an independent laboratory. The LD50 in every case was determined to be >5 gm/kg, which is considered nontoxic. Skin and eye irritation tests were negative. Biodegradability Biodegradability has become an important issue in the last few years. Massive oil spills have focused concern on the desirability of working fluids that degrade in the environment to
Copyright © 1994 CRC Press, LLC
231
Volume III
FIGURE 6. Biodegradability of PAO.
harmless end-products. Unfortunately for many fluids, those properties that make them stable to oxidative and thermal degradation in their working environment also make them stable toward biodegradation. There is a great deal of debate as to what constitutes valid testing for biodegradability. The CEC L33 T82 test, which was originally designed for outboard engine oils, appears to be winning acceptance as the most appropriate test available today for engine oils. The weight percent material biodegraded under the test conditions for all of the commercial viscosity grades of PAO are given in Figure 6. The lower grades (2.0 and 4.0 cSt) show fairly good degradation, while the higher grades degrade less rapidly. PAO fluids often show higher levels of biodegradation under the test conditions that equiviscous grades of mineral oils, but mineral oils show considerable differences depending on type. Comparative tests at the same laboratory showed only 20% biodegradation of two different stocks of 2.0 cSt mineral oil (compared to over 75% for the PAO).14
CONCLUSION
Polyalphaolefin synthetic oils have good inherent physical and chemical characteristics that make them desirable for use as lubricants and functional fluids. Although today the cost of PAOs is more than that of mineral oils, performance benefits are often sufficient to make them the preferred choice on a cost-effective basis. In addition, manufacturers of PAO fluids have a degree of flexibility in tailoring their products to meet end-use application requirements. Finally, the availability of PAO fluids is not limited by the type and availability of certain crude-oil base stocks. In a highly competitive world market that is placing increasing emphasis on efficiency and performance, continued strong growth in the use of PAO functional fluids appears certain.
Copyright © 1994 CRC Press, LLC
232
CRC Handbook of Lubrication and Tribology
REFERENCES
1. Shubkin, R. L., Ed., Synthetic Lubricants and High-Performance Functional Fluids, Marcel Dekker, New York, 1992. 2. Shubkin, R. L., Polyalphaolefins, in Synthetic Lubricants and High-Performance Functional Fluids, Shubkin, R. L., Ed., Marcel Dekker, New York, 1992, 1. 3. Dressier, H., Alkylated aromatics, in Synthetic Lubricants and High-Performance Functional Fluids, Shubkin, R. L., Ed., Marcel Dekker, New York, 1992, 125. 4. Fotheringham, J. D., Polybutenes, in Synthetic Lubricants and High-performance Functional Fluids, Shubkin, R. L., Ed., Marcel Dekker, New York, 1992, 271. 5. Venier, C. G. and Casserly, E. W., Cycloaliphatics, in Synthetic Lubricants and High-Performance Functional Fluids, Shubkin, R. L., Ed., Marcel Dekker, New York, 1992, 241. 6. Shubkin, R. L., Baylerian, M. S., and Maler, A. R., Olefin Oligomers: Structure and mechanism of formation, Ind. Eng. Chem., Product Res. Dev., 19, 15, 1980. 7. Shubkin, R. L. and Kerkemeyer, M. E., Tailor making PAOs, J. Synth, Lubr. 8(2), 115, 1991. 8. Theriot, K. J. and Shubkin, R. L., A polyalphaolefin with exceptional low temperature properties, 8th Int. Colloq. TRIBOLOGY 2000, Technische Akademie Esslingen, Esslingen, Germany, January 14 to 16, 1992. 9. Kumar, G. and Shubkin, R. L., New polyalphaolefin fluids for specialty applications, 47th Annu. Meet. Soc. Tribol. Lubr. Eng., Philadelphia, May 4 to 7, 1992. 10. Ethyl Corp. Product Brochure, ETHYLFLOR™ Polydecene for personal care applications. Ethyl Corp., Baton Rouge, LA. 11. Coffin, P. C, Lindsay, C. M., Mills, A. J., Lindenkamp, H., and Furham, J., The application of synthetic fluids to automotive lubricant development: trends today and tomorrow, J. Synth. Lubr., 7(2), I 123, 1990. 12a. Anon,, Extending compressor valve cleaning periods with a synthetic compressor lubricant, Fluid Lubr. Ideas, 6(5), 24, 1983. 12b. Anon., Synthetic lubricants reduce downtime at Midwest power plant. Fluid Lubr. Ideas, 6(5), 20,1983. 12c. Anon., Synthetic lubricant saves energy, increases oil-change interval, Eng. Min. J., 183(10), 111, 1982 12d. Facchiano, D. L. and Johnson, R. L., Examination of synthetic and mineral based gear lubricants and their effect an energy efficiency, NLOI Spokesman, 48(11), 399, 1985. 12e. Skinner, R. S., Synthetic lubricants—why their extra costs can be justified. Mar. Eng. Rev., August 1986, pp. 18,20–21. 12f. Black, P. A. and Knobel, H. E., Synthetic lube oils improve performance. Mot. Ship, 66(782), 30,1985. 12g. Faufau, J. and Nick, T. C., Synthetic lubricants can reduce downtime and increase bearing life, Pulp Pap.. 63(1), 127, 1989. 12h. Fredel, W., Synthetic lubricants help Mosinee paper overcome temperature problem and save money, Pap. Trade J.. 168(18), 16, 1984. 12i. Schlenker, H. O., Synthetic lubricants upgrade worm gear capacities, Power Trans. Des., 24(7), 34,1982. 12j. Anon., Improving industrial gear system performance with synthesized lubricants, Fluid Lubr, Ideas, 62(2) 9, 1983. 13. Edwards, D. J., Synthetic lubricants get tougher. Plant Eng., August 18, 1983, 37, 17, 59–60. 14. Carpenter, J. F., Assessment of environmental impact of PAOs, presented at STLE Annu. Meet., Calgary, May 17 to 20, 1993.
Copyright © 1994 CRC Press, LLC
Volume III
Property Color Viscosity, cSt, 200°C Viscosity, cSt, 150°C Viscosity, cSt, 125°C Viscosity, cSt, 100°C Viscosity, cSt, 40°C Viscosity, cSt, - 18°C Viscosity, cSt, -40°C Viscosity index Brookfield viscosity, cP, -40°C Cold crank simulator, cP, -25°C Pour point °C Flash point, °C Fire point, °C NOACK volatility, 250°C, 1 h, %wt. evap. Specific gravity, 15.6/15.6°C (60/60°F) Density, lb/gal, 15.6°C (60°F) Tool acid no., mg KOH/g Bromine no., g Br/100 g Moisture, ppm Molecular weight DSC, oxidation, onset temp, °C DSC, energy kj/g Aniline point, °C Evaporative wt. loss % Dielectric constant 23°C, 1.0 and 1000 KHz Vapor pressure, mm Hg 37.8°C (100°F) 93.3°C (200°F) 148.9°0C (30°F) Specific heat, cal/gm/°C, 75°C 100°C
Appendix A TYPICAL PROPERTIES OF PAOs PAO 2 <0.5 0.73 1.05 1.35 1.80 5.54 62.0 310 — — — <-65 >155 — 99
PAO 4 <0.5 1.31 1.90 2.61 3.90 16.8 341 2,460 129 2,100 490 -70 215 250 12
PAO 6 <0.5 1.75 2.65 3.77 5.90 31.0 901 7,890 138 6,950 1,300 -68 235 268 7.0
6.653 — 0.5 50 287 — — — — 2.076
6.81 <0.01 0.2 11 437 192 12.1 116.7 11.8 2.102
6.89 <0.01 0.2 13 529 192 12.4 129.6 4.6 2.118
— — — 0.551 0.576
— — — 0.567 0.597
<0.00047 0.00101 0.00062 0.0019 0.0016 0.528 0.555 0.548 0.573
0.797
Copyright © 1994 CRC Press, LLC
0.818 0.827
PAO 8 <0.5 2.06 3.26 4.88 7.80 45.8 1605 18,160 140 16,920 3,100 -63 252 286 3.0
0.832
6.93 <0.01 0.3 13 596 191 13.8 123.3 3.3 2.122
PAO 10 <0.5 2.35 3.94 5.82 9.60 62.9 2,770 32,650 134 — — -53 264 300 2.0
233
PAO 40 <0.5 — 12.9 16.6 40.0 395 40,200 — 151 — — -34 272 310 0.8
PAO 100 <0.5 — 29.0 50.0 100 1,250 203,000 — 168 — — -20 288 320 0.6
6.97 <0.01 0.4 19 632 191 13.6 132.2 1.6 2.130
7.04 <0.01 — — 1,400 189 13.8 153.3 — 2.150
7.13 <0.01 — — 2,000 188 14.6 154.4 — 2.151
— ASTM D 974 IP-129 ASTM D 1744 GC (calc) — — ASTM D 611 ASTM D 972 —
0.00030 0.00050 0.0017 0.546 0.572
0.00010 — — 0.553 0.574
— — — 0.530 0.552
— — —
0.836
0.845
0.856
Test method ASTM D 1500 ASTM D 445 ASTM D 445 ASTM D 445 ASTM D 445 ASTM D 445 ASTM D 445 ASTM D 445 ASTM D 2270 — — ASTM D 97 ASTM D 92 ASTM D 92 DIN 51581
ASTM D 1298
—
234
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
Copyright © 1994 CRC Press, LLC
DIBASIC ACID AND POLYOL ESTERS Joseph M. Perez and Elmer E. Klaus
INTRODUCTION
Esters are one of the largest classes of synthetic lubricants in the commercial market. The esters of interest in this chapter are the reaction product of an organic acid with an alcohol and typically contain about 25 to 60 carbon atoms per molecule. The acid and alcohol building blocks commercially available in adequate purity and low cost tend to contain from 5 to 13 carbon atoms per molecule. To produce lubricants of adequate molecular size and viscosity, ester lubricants tend to be based on either multifunctional acids or multifunctional alcohols. Dibasic acids have been combined with alcohols to produce dibasic acid esters. Similarly, polyol alcohols such as neopentyl glycol, trimethylol propane, and pentaerythritol have been used in conjunction with monobasic acids to produce polyol esters. The polyol alcohols and ethylene oxide or propylene oxides can be used to produce higher molecular weight polymeric esters. An outstanding characteristic of esters as a class is their excellent response to additives that are designed to control oxidation, hydrolysis and corrosion and improve lubricity. They possess excellent viscosity-volatility properties that include low volatility, high flash points, good viscosity-temperature characteristics, and are used in a wide temperature range of applications. Historically, use of synthetic esters became widespread during World War II to meet the need for low temperature lubricants for military weapons applications, and shortages of lubricants derived from petroleum led to early German work on esters.1,2 One comprehensive review of the history of synthetics is found in Synthetic Lubricants by Gunderson and Hart.3 The jet age born in the turbulence of World War II expanded the need for synthetic lubricants. Esters for both MTL-L-7808 and MIL-L-23699 military specifications were commercially available in the specialty lubricant and plasticizer markets when the aircraft gas turbine (AGT) was developed. The problem in the 1950s was tailoring these esters for the new engine market. This involved development of an understanding of the high temperature oxidation, thermal and corrosion requirements of the engines and then providing esters of adequate quality. More severe engine requirements resulted in the development of military specifications for MTL-L-7808. Dibasic acid ester base lubricants satisfied this specification and accounted for more than 3 million gallons per year by the mid-1960s. Significant use of polyol esters appeared in the 1960s as operating conditions and temperatures in the turboprops and turbines became more demanding. Specification MIL-L-23699 type fluids, made with polyol esters, were used in the more severe applications increasing the demand for these esters dramatically in the late 1960s. Typical properties for the two fluids are found on Table 1. Genetically, esters are produced by the reaction of organic acids and alcohols: In Reaction 1, the alkyl groups R and R’ can be the same or different in chain length and structure. The acid can be either mono- or dibasic and the alcohol may contain more than one hydroxyl group. The design of the molecules for a specific application is tailored to maximize stability, lower volatility, maintain low pour points, and provide good additive susceptibility. Good viscosity-temperature properties (high viscosity index) and natural dispersant/detergent tendencies are characteristic of the fluids. 0-8493-3903-0/94/$0.00 + $0.50 © 1994 by CRC Press, Inc.
Copyright © 1994 CRC Press, LLC
237
238
CRC Handbook of Lubrication and Tribology
The focus of this chapter will be on the advantages and disadvantages of the dibasic acid and polyol esters.
DIBASIC ACID ESTERS
The principal types of diesters include the adipates, phthalates, azelates, and sebacates formed from the dibasic acids and alcohols found in Table 2. In the early 1930s and 1940s significant numbers of esters were synthesized and evaluated in the U.S. and in extensive programs in Germany. The properties of a number of these esters, evaluated by M. R. Fenske and co-workers at the Petroleum Refining Laboratory of the Pennsylvania State University, are found in Table 3. Of note are the high flash points and excellent viscosity-temperature characteristics that made this class of fluids useful for the military applications. A convenient way of controlling viscosity is to use esters of a specific size. The viscosity and volatility levels of esters suitable for functional fluids and lubricants usually require a molecule containing 20 to 30 carbon atoms. When low temperature fluidity is also a consideration, esters of multifunctional acids and alcohols are more desirable than simple esters to provide asymmetry. Dibasic acid esters of monofunctional alcohols are widely used to meet these requirements. Esters of adipic, azelaic and sebacic acids with oxoalcohols of 8 to 13 carbons are examples of these base fluids. The preparation of these esters, represented generically in Equation 1, involves a condensation reaction with water formed as a by-product. Copyright © 1994 CRC Press, LLC
Volume III 239
Copyright © 1994 CRC Press, LLC
240
CRC Handbook of Lubrication and Tribology
Copyright © 1994 CRC Press, LLC
241
Volume III
FIGURE 1. Common polyols.
The water has to be removed and excess alcohol and heat are used to drive the reaction to completion. The synthesis of di-2-ethylhexyl sebacate is shown in Equation 2.
POLYOL ESTERS
In the case of the polyol esters, the alcohol is multifunctional, containing usually three or four alcohol groups. The more commonly used alcohols include polyols, such as neopentyl glycol, trimethylolpropane, pentaerythritol, and dipentaerythritol, which are shown structurally in Figure 1. Typical starting materials are found in Table 4. Technology has also resulted in the synthesis of polymeric esters through polymerization of unsaturated esters (polymethacrylates); condensation polymerization of polyfunctional acids and alcohols (polyesters); and the copolymerization of unsaturated acids and olefins such as maleic acid and polybutenes. These polymeric esters have been used to improve wear and increase viscosity and viscosity index.4 The general esterification reaction for the polyols found in Figure 1 is shown in Equation 3.
where R,R‘ can be an H or alkyl group, and x is equal to the number of alcohol groups. One advantage of the polyol esters over the dibasic acid esters is better thermal stability. Absence of the beta hydrogens increases the thermal stability by about 50°C over that of the dibasic acid esters. However, this advantage can be significantly reduced by the catalytic activity of some metals such as iron in ferrous alloys. A second advantage of polyolesters over dibasic acid esters is in the effectiveness of certain additives to produce better oxidation stability.
POLYMERIC ESTERS
Esters have been part of the backbone of many synthetic polymers used by industry over the years. Esters of acrylic and methacrylic acids are the monomers for high molecular weight viscosity index improvers, dispersants and detergents. More recent polymer type esters, used
Copyright © 1994 CRC Press, LLC
242
CRC Handbook of Lubrication and Tribology
to improve viscosity volatility characteristics of both synthetics and petroleum base oils, are the esters of α-olefin-dicarboxylic acid copolymers. These compounds have intermediate molecular weights of 1200 to 2500. The basic structure is one in which the ester groups are present in side chains and the main chain or backbone consists of carbon atoms. To extend the size and increase the viscosity of the resulting ester produced by the reaction of the polyol alcohols with a monobasic acid, ethylene oxide or propylene oxide is used to produce a polyglycol ether which is a difunctional alcohol with a series of ether linkages. The size is determined by controlling the polyfunctional acids and alcohols used to produce the polyesters which can be end-capped with either monobasic acids or mono-alcohols. Esters like methyl acrylic acid esters containing an olefinic group in the acid portion can be polymerized to form viscous polyesters. In some cases, an unsaturated dibasic acid derivative like maleic anhydride can be copolymerized with a large olefin like polybutene to make large polybasic acids which can then be esterified to produce viscous lubricant constituents with more easily controlled viscosity.
FLUID PROPERTIES
Some properties of the dibasic acid esters and polyol esters are compared to those of other synthetics base stocks in Table 5. These ratings are based on extensive studies of the properties listed for the best available examples of the fluid types shown. Oxidation and hydrolytic stability are based on the base fluids compounded with effective additives to enhance the property. High temperature thermal stability is based on an evaluation in the presence of Copyright © 1994 CRC Press, LLC
243
Volume III
FIGURE 2. Stability of oxidation-inhibited lubricants at 175°C: (a) conventionally refined mineral oils that contain N, 0, S, and polar constituents; (b) super-refined mineral oil. (From Klaus, E. E., Perez, J. M., and Fenske, M. R., Am. Chan. Soc. Div. Par. Chem. Prepr., 5(2), 59, 1960.)
ferrous alloys. The remaining properties were determined on the fluid types without formulation. The ratings are relative to 10, the best performance. The primary advantage of the ester base fluids is in the area of low temperature fluidity. They also exhibit excellent additive response to provide good oxidative and lubricity properties. Fully formulated fluids are compared in Table 6 using the same scale. Tailoring of synthetic formulations to meet the specific requirements of an application is done by taking advantage of strong points such as improved additive susceptibility and compatibility with petroleum-based products. Improved phenolic and amine type antioxidant additive susceptibility results in excellent oxidation stability of ester lubricants, (Figure 2).5
APPLICATIONS
Use of ester-base stocks is becoming more widespread in industrial applications, a few of which are found in Table 7. Military applications include engine lubricants, hydraulic oils,
Copyright © 1994 CRC Press, LLC
244
CRC Handbook of Lubrication and Tribology
and greases (Table 8). The esters are also used commercially as less flammable fluids, as transformer and capacitor fluids, plasticizers, and lubricity modifiers.
Energy-Efficient Applications The energy crisis of the 1970s resulted in a reevaluation of ester-based lubricants for industrial and automotive applications, including crankcase and differential lubrication applications.6 Friction reduction can be achieved with the ester-base stocks in both hydrodynamic and elastohydrodynamic applications. In the boundary regime, the additive package is the key to friction reduction and the superior additive response of esters provides a base fluid for the use of such additives.
Automotive Applications and Ester Properties Controlled volatility, low temperature fluidity, high temperature viscosity, and oxidation stability are desirable factors in automotive lubricants. There is an established relationship between viscosity, volatility, and viscosity temperature properties of liquids7 applicable to narrow-boiling range (30°C) mineral oils and single compound fluids such as esters. A comparison of viscosity level and flash point indicates in Table 9 that the esters are superior to the mineral oils in viscosity-volatility characteristics. The esters show a much lower viscosity at a comparable temperature for fluids in the same volatility or boiling point range. The mineral oil-base stocks in the table are good quality paraffinic oils with pour points obtained by dewaxing. Typical low cold test mineral oils with pour points in the range of the esters would have poor viscosity-temperature properties, e.g., viscosity indexes (VI) of 60 to 80. The viscosity-volatility relationship is summarized in Figure 3. Comparison of the viscosity temperature relationship of several types of fluids is shown in Table 10. At the other extreme, advanced engines under development increase the severity of operating conditions. The lubricant must reduce friction and wear and have adequate viscosity at high temperatures. Current engines are exceeding 350°C top ring reversal temperatures, Copyright © 1994 CRC Press, LLC
245
Volume III
FIGURE 3. Viscosity-volatility relationship.
and at these temperatures an adequate lubricating film needs to be maintained. In addition, the amount of lubricant on the top land and at top ring reversal needs to be minimized to reduce the lubricant’s contribution to engine emissions. A comparison of fluid viscosity as temperature increases is shown in Table 11 for several fluids having about a 2 cSt viscosity at 150°C and boiling points of 371°C. The esters are typically 125 to 150 V.I. For comparison, a mineral oil with a midpoint boiling point of 371°C is shown. The data, and the generalized data in Figure 3 for various oil types, suggest that the esters can be of considerably lower viscosity than mineral oils at 40°C and still achieve adequate film thickness and low volatility at high temperatures. As a result, ester-polymer blends can be prepared to meet SAE 5W-40 or SAE 5W-50 multigrades with volatility levels typical of conventional mineral oil formulations of SAE 10W30, SAE 20W, or even SAE 30 grades.
Copyright © 1994 CRC Press, LLC
246
CRC Handbook of Lubrication and Tribology
Some examples of ester formulations prepared with and without polymer additions are found in Table 12. Polymer-mineral oil blends do provide energy-efficient crankcase oils when compared with Newtonian oils of the same SAE grade.8 The esters provide volatility control at sufficiently low viscosities that polymer blends with much higher relative viscosities can be prepared. This means that lower molecular weight polymers could be used with esters in the preparation of crankcase oils. Many mineral oils tend to give higher than predicted low temperature viscosities from extrapolations on the ASTM viscosity temperature charts. However, viscosities of many dibasic acid esters agree with straight line extrapolations of the ASTM charts, and most polyol esters exhibit the unusual characteristic of lower than predicted viscosities at to temperatures. Several examples of molecules exhibiting this effect are found in Table 13. The pentaerythritol ester (PE) is an example of this low temperature advantage of the polyol molecules which exhibit a dense center with two to four long alkyl chains connected to a central carbon atom. The polyols in Figure 1 are examples of this molecular geometry. Dicapryl phthalate is a dibasic acid ester with a dense aromatic center and a poor VI compared to me PE. With the development of the elastohydrodynamic theory, the significance of pressure effects on lubricant films in the mechanical components of highly loaded tribological systems is important. Various methods of measuring pressure viscosity effects can be found in the Copyright © 1994 CRC Press, LLC
Volume III
247
literature.9,14 The esters exhibit excellent pressure-viscosity properties and a unique property of maintaining their fluidity at high pressure. A comparison of several fluids with a viscosity of 20 cSt at 40°C and 0.1 MPa is found in Table 14.
Magnetic Storage Systems A survey of the literature shows that the development of a lubricant for the magnetic layer in magnetic recording tape was a challenging problem with the magnetic storage representing a thin porous and permeable media needing lubrication.15–17 In general, the lubricant of choice for these products was a simple ester contained in the porous permeable layer. It appears that this simple ester lubrication system could be improved by the use of less volatile dibasic acid or polyol esters.15,17 The mechanism of tape and floppy disc lubrication delivery systems appears to be based on the principle that when there is head-tape contact, the contact pressure on the porous and permeable magnetic tape layer forces the lubricant to the surface of the tape to provide the lubrication. The low volatility and tribological properties of the dibasic and polyol esters make them attractive as topical lubricants.
Copyright © 1994 CRC Press, LLC
248
CRC Handbook of Lubrication and Tribology
FIGURE 4. Behavior of a formulated di-2-ethylhexyl sebacate base synthetic in microoxidation test 225°C. (From Klaus, E. E., New Directions in Lubrication Materials, Wear, and Surface Interactions: Tribology in the 80’s, Loomis, W. R., Ed., Noyes Publications, 1985.)
PROPERTIES
Oxidation Stability Oxidation is probably the most significant chemical factor affecting the performance of lubricants in mechanical applications, especially in automotive engines. The first step in oxidation of hydrocarbons is the formation of peroxides at the most vulnerable C-H bonds. The breaking of the peroxide bonds initiates a free radical chain mechanism that propagates the formation of peroxides. The oxidation reactions lead to the formation of other oxygencontaining molecules such as aldehydes, ketones, alcohols, acids, and esters. Polymerization and evaporation reactions also occur at rates depending on the type of products formed. Ester oxidation occurs by a similar peroxide-free radical path and, again, additives are critical to providing oxidation protection. It has been shown18 that the formation of high molecular weight products occurs in ester oxidation after the depletion of the additive package (Figure 4). Copyright © 1994 CRC Press, LLC
Volume III
249
Recent studies of ester oxidation19 show that primary oxidation produces aldehydes and ketones. These aldehydes and ketones then undergo a secondary oxidation-polymerization reaction to produce higher molecular weight condensation polymers. These polymers grow to become viscous and finally insoluble sludge and varnish. Both primary and secondary oxidation reactions are accelerated by iron in ferrous alloys, and both reactions can be inhibited by appropriate inhibitors. Oxidation can be controlled by deactivating the ferrous metal surfaces in a system which catalyzes the oxidation. An important finding in this study of oxidation kinetics20 was that the primary oxidation to form oxidation products (aldehydes or ketones) is the rate-limiting step. However, the secondary oxidation or the condensation polymerization step tends to proceed at rates that are 10 to 100 times the primary oxidation rates. The studies show that these two oxidation phenomena do not necessarily respond to the same oxidation inhibitor. In many cases, a combination of inhibitors is required for optimum oxidation control. The ability of additives to provide oxidation protection depends on their compatibility with the base fluid and whether synergistic or antagonistic effects occur. It is evident from Figure 2 that the effectiveness of oxidation inhibitors is related to the purity of the base stock. To achieve comparable additive response the mineral oils have to undergo considerable super refining to remove S-, N-, and O-containing impurities in the base oils. Esters prepared for lubricant base fluid applications, as opposed to plasticizer fluids, have shown exceptional additive susceptibility. The inhibited esters have shown excellent oxidation stability and volatility characteristics in the thin film microoxidation tests.21 In thin film oxidation tests, oxidation degradation results in increased volatility. Some of the most effective oxidation inhibitors in esters are insoluble or have limited solubility in mineral oils. The esters exhibit an autocatalytic degradation as a function of organic acid degradation.22 The use of an oxidation inhibitor from the aromatic amine class and a second additive to titrate stoichiometrically the incipient acid formation provides the optimum oxidation stability. This type of additive package is more effective in polyol esters than dibasic acid esters and has been pursued in the development of high temperature liquid lubricants for advanced engine applications.23–25 A third additive type that is beneficial to the oxidative stability of esters is a metal surface deactivator, such as the phosphates. Essentially all aircraft gas turbine oils contain a phosphate ester as a combination lubricity additive and metal deactivator. These phosphate esters (usually tricresyl phosphate) are used to serve as a metal coater or metal deactivator to reduce iron corrosion and resultant increase in the oxidation rate of the ester. Metals can have a catalytic effect on the oxidation process, resulting in a significantly lower temperature threshold of oxidation which results in larger quantities of undesirable oxidation products, such as sludge. The degree of metal catalysis depends on the surface metal composition, the rate of production of soluble metals by surface corrosion, and the effectiveness of the surface deactivator.26 In thin film oxidation studies with various metal surfaces, the polyol ester (trimethylolpropane triheptanoate) was more stable than the dibasic acid ester (di-2-ethylhexyl sebacate). The polyol ester oxidized at about two-thirds the rate of the dibasic acid ester. Figure 5 summarizes the oxidation of a polyol ester at 225°C in a thin film microoxidation test.27 Reference 27 is a general reference, comparing the performance of esters with other types of fluids and lubricants.
Hydrolysis Hydrolysis is an additional stability problem related to oxidation. It is simply a reequilibration that takes place if there is water present in the ester. The organic acids produced by hydrolysis of the ester tend to accelerate oxidation. Therefore, to prevent hydrolysis during low temperature storage of materials like fully formulated aircraft gas turbine oils, it is necessary to use additives like benzothiazole. Prior to the use of hydrolysis inhibitors. supplies
Copyright © 1994 CRC Press, LLC
250
CRC Handbook of Lubrication and Tribology
FIGURE 5. Catalytic effect of metals on oxidation of trimethylolpropane triheptanoate at 225°C in comparison with di-2-ethylnexyl sebacate. (From Lahajani, J., Lockwood, F. E., and Klaus, E. E., ASLE Trans., 25, 26; 1982.)
of Spec MIL-L-7808 gas turbine oil were discarded after storage for a specified time. The time of storage was based on tests of hydrolytic stability of the formulated product.
Electrical Conductivity Electrical conductivity of the lubricant is often a factor in industrial systems, such as hydraulic applications. The electrical conductivity of a well-refined and dry mineral oil and most synthetic esters is similar, in the 10 to 12 to 10 to 14 mho/cm2 range. In some hydraulic systems, streaming or zeta potentials have been related to erosion in servo valves. This appears to be due to the alteration of the electrical conductivity of the base fluid due to additives or impurities resulting from oxidation or contamination28 Maintaining the electrical conductance of the base fluid is critical in some electrical and hydraulic systems. This can be done by proper fluid selection for closed systems or external removal of the impurities such as water. Surface Tension Surface tension is also a consideration in lubricant applications. The surface tension values for several lubricants are found in Table 15. The major problem with surface tension is that
Copyright © 1994 CRC Press, LLC
251
Volume III
the value for the finished lubricant is sensitive to additives. The use of 0.1 wt% of a silicone in an oil will reduce the surface tension to essentially that of the silicone. The proper selection of additives to lower surface tension can improve the surface wetting of the lubricant and can be important in the lubrication of advanced materials.
LIMITATIONS
A disadvantage of esters is their selective compatibility with polymeric materials such as rubber. Excessive changes in rubber seals and hoses due to ester compatibility can result in swelling and softening of the component. In order to use esters in systems designed for mineral oils or other lubricants, it is necessary to determine the compatibility of the ester-base lubricant with the plastics used in the system. Although extensive use of esters in aircraft gas turbine engines has resulted in the development of plastic seals, hoses and parts that are safe to use with the ester-base lubricants, evaluation of system materials may be necessary. A second disadvantage of the synthetic esters is the cost of high quality fluids. However, for some applications this is not a drawback since there are no alternatives for lubrication of the system. This was certainly the reason for the introduction of synthetics in military applications in World War II. As advanced materials technology matures in engine and other industrial applications, the use of synthetic esters will likely become even more widespread to insure the performance and life requirements of more demanding systems. Ester bases and synthetic lubricants with higher costs have been replacing lower cost lubricants on the basis of unique properties, longer life, and lower energy consumption. The unique property aspects of ester base lubricants played a significant role in their domination of the aircraft gas turbine lubricants. The longer useful life of ester lubricants based on superior oxidative stability and low vapor pressure for a given viscosity can provide for longer intervals between relubrication. In many cases where relubrication is required, the manpower and time of relubrication is much more important as a cost factor than the cost of the small amount of lubricant required. The cost of energy also plays a role in the cost effectiveness of ester-based lubricants.
SUMMARY
Use of synthetic esters is continuing to grow, due to an even balance of properties. Esters possess excellent viscosity-temperature characteristics mat allow their application over a wide temperature range. Most esters possess low volatility that results in high flash and less flammable characteristics. They possess good thermal, oxidative, and hydraulic stability, and their lubricating properties include high film strength, good surface wetting, and low friction and wear rates. Additive susceptibility is excellent. As petroleum supplies dwindle, use of the synthetic esters will be a prime alternative to fill the needs of many tribological systems.
Copyright © 1994 CRC Press, LLC
252
1.
CRC Handbook of Lubrication and Tribology
REFERENCES
Zorn, H., Esters as lubricants, U.S. Air Force Transl. Rep., F-T-S-957RE, 1946–1947. 2. Zorn, H., Esters as lubricants, Erdoel Kohle, 8, 414, 1955. 3. Gunderson R. C. and Hart, A. W., Eds., Synthetic Lubricants, Rheinhold Publishing, New York, 1962. 4. Beimisch, B. J., Synthetic hydraulic fluids, ASLE Educational Course, ASLE, Las Vegas, Emery Publication, Emery Industries, Cleveland, OH, 1985. 5. Klaus, E. E., Perez, J. M.,and Fenske, M. R., Development and use of a simplified oxidation test for lubricants, Am. Chem. Soc., Div. of Petr. Chem. Prepr., 5(2), 59, 1960. 6. Klaus, E. E., Tewksbury, E. J., and Pride, E. F., The effect of base stock and additives on the formulation of lubricant efficient lubricants, Lubr. Eng., Sp. publ. SP7, 24, 1981. 7. Haviland, M. L. and Davison, E. D., Lubricant viscosity effects on passenger car fuel economy, SAE Trans., 84, Pap. No. 750675, 1975. 8. Hart, W. and Klaus, E. E., Laboratory testing of fuel efficient oils, SAE Pap.. No. 790731, 1979. 9. Wu, C. S., Klaus, E. E., and Duda, J. L., Development of a method for the prediction of pressure-viscosity coefficients of lubricating oils based on free-volume theory, J. Tribal., HI, 121, 1989. 10. Pressure Viscosity Report, American Society of Mechanical Engineers, New York, 1953. 11. Roelands, C. J. A., Correlational aspects of the viscosity-temperature-pressure relationship of lubricating oils, V.R.B., Kleine Gronigen, Holland, 1966. 12. Dixon, J. A., Webb, W. and Steele, W. A., Properties of hydrocarbons of high molecular weight, API Res. Proj. 42, Rep. to the American Petroleum Institute, Pennsylvania State University, State College, PA, 1962. 13. So, B. Y. C. and Klaus, E. F., Viscosity-pressure correlations of liquids, STLE Trans., 23(4), 409,1980. 14. Lansdown, A. R., Lubricants, Proc. Inst. Mech. Engrs., Int. Conf. Tribology—Friction, Lubrication and Wear, Vol. 1, London, 1987, 365. 15. Klaus, E. E. and Bhushan, B., Lubricants in magnetic media—a review, STLE SP Tribology and Mechanics of Magnetic Storage Systems, D, 7, 1985. 16. Klaus, E. E. and Bhushan, B., A study of the stability of magnetic tape lubricants, STLE SP Tribology and Mechanics of Magnetic Storage Systems, EG, 24, 1986. 17. Klaus, E. E. and Bhushan, B., The effects of inhibitors and contaminants on the stability of magnetic tape lubricants, STLE Trans., 31(2), 276, 1988. 18. Klaus, E. E., Status of new directions of liquid lubricants, in New Directions in Lubrication Materials, Wear, and Surface Interactions: Tribology in the 80’s, Loomis, W. R., Ed., Noyes Publications, 1985. 19. Naidu, S. K., Klaus, E. E., and Duda, J. L., Kinetic model of high temperature oxidation of lubricants, Ind. Eng. Chem. Prod., 25, 596, 1986. 20. Lockwood, F. and Klaus, E. E., Ester oxidation—the effect of an iron surface, STLE Trans., 25(2), 236, 1982. 21. Hsu, S. M., Pei, P., and Ku, C. S., Mechanisms of additive effectiveness, Lubr. Sci., Vol 1–2, No. 0954–0075, 165, 1990. 22. Lockwood, F. E. and Klaus, E. E., Ester oxidation under simulated boundary conditions, ASLE Trans., 24(2), 278, 1981. 23. Sutor, P., Bardasz, E. A., and Bryzik, W., Improvement of high temperature diesel engine lubricants, SAE Pap.. No. 900687, 1990. 24. Marolewski, T. A., Slone, R. J., and Jung, A. K., High temperature liquid lubricant for use in low-heat rejection diesel engines, SAE Pap., No. 900689, 1990. 25. Perez, J. M., Ku, C. S., and Hsu, S. M., High temperature liquid lubricant for advanced engines, SAE Pap.. No. 910454, 1991. 26. Lahijani, J., Lockwood, F. E., and Klaus, E. E., The influence of metals on sludge formation, ASLE Trans.. 25(1), 25, 1982. 27. Klaus, E. E. and Tewksbury, E. J., Liquid lubricants, in CRC Handbook of Lubrication, Vol. n, Theory and Design, Booser, E. R., Ed., CRC Press, Boca Raton, FL, 1984. 28. Beck, T. R., Wear by generation of electrostatic streaming currents, ASLE Trans., 26, 144, 1982.
Copyright © 1994 CRC Press, LLC
POLYALKYLENE GLYCOLS* William L. Brown
INTRODUCTION
Polyalkylene glycol synthetic lubricants are used in a large number of diverse applications where petroleum oil-based products do not provide satisfactory performance. This chapter begins by describing some of the physical and chemical properties of polyalkylene glycol polymers. Polyalkylene glycols have high viscosity indices, low toxicity, variable solubility characteristics, and good lubricity in a wide range of applications. Major commercial uses of polyalkylene glycol lubricants are then reviewed. These applications include gear and compressor lubricants, metalworking fluids, fiber lubricants, and fire-resistant hydraulic fluids. This chapter concludes with presentation of some typical physical properties for a number of commercial polyalkylene glycols. Tables and figures provide data on properties such as viscosity, water solubility, elastomer compatibility, and pour point.
POLYALKYLENE GLYCOL CHARACTERIZATION
Polyalkylene glycols (PAGs) are made from the reaction of alkylene oxide monomers with a nucleophilic starter, usually an alcohol. PAGs can be represented by the following structure:
where R = H or alkyl group; R’ = H, CH3, or alkyl group; and R” = H or alkyl group. The starting alcohol in most polyalkylene glycols is either a short chain linear alcohol or 2 diol. While PAGs can be made from the polymerization of any alkylene oxide, they are usually copolymers of ethylene oxide (EO) and propylene oxide (PO). The EO content can range from 0 to 100% of the oxide feed, with the balance being PO. The sequencing of the oxide monomers can either be random or blocked as shown: random PAG = M-ABBAABAAABABABBBAABAB or BAABABBBAA-D-BBBABAABBBA blocked PAG = AAAAAABBBB-D-BBBBAAAAAAA or M-BBBBBBBBBAAAAAAAAAAA
where M is a monofunctional alcohol starter, D is a diol starter, and A, B are alkylene oxide monomers. Blocked PAGs are more surface active than random polymers of equivalent molecular weights and EO to PO ratios. Therefore, blocked PAGs have better wetting properties and more surfactancy than the analogous random polymers, but they are also more prone to foaming. As the molecular weight or the percentage of polymerized EO in the PAG increases, the pour points of blocked polymers rise significantly faster man those of the random products. * ©1992 Union Carbide Chemicals & Plastics Technology Corporation. All rights reserved. 0-8493-3903-0/94/$0.00 + $0.50 © 1994 by CRC Press, Inc.
Copyright © 1994 CRC Press, LLC
253
CRC Handbook of Lubrication and Tribology
254
Polyalkylene glycols always contain at least one terminal hydroxyl group (R”—OH) as they come out of the alkoxylation reactor. If the starter is a monofunctional alcohol, there is one terminal hydroxyl group. Diol-started polyalkylene glycols have two terminal hydroxyl groups, triol-started polymers have three, and so forth. Subsequent capping reactions can be used to convert the terminal hydroxyl functionalities into alkyl groups. However, most PAGs are not capped and therefore contain at least one terminal hydroxyl group. The PAGs described in this chapter use the following nomenclature:
# = Weight percentage of polymerized EO in the PAG. The remainder of the monomer feed is PO. A = O for oil soluble; W for water soluble at room temperature. X = Represents the starter alcohol: M = methanol; B = butanol; D = low molecular weight diol, such as ethylene glycol, diethylene glycol, propylene glycol, or dipropylene glycol. * * * * = Number average molecular weight of the polyalkylene glycol. Y = Represents the monomer sequencing: R = random; Bn = normal block sequencing, (EO)x-(PO)y-(EO)x; Br = reverse block sequencing, (PO)x-(EO)y-(PO)x. Polymers made from 100% propylene oxide are not followed by a monomersequencing letter.
PAGs, like petroleum oils and other synthetic fluids, are chosen for a given application because they satisfy certain chemical, physical, and performance requirements. A number of inspection tests are used both to measure these requirements and to monitor the condition of the fluids during use.1 Standard ASTM test methods which are applicable to both petroleum oils and PAGs include:
A number of inspection methods that were developed for petroleum oils either do not apply to PAGs or must be modified in order to give meaningful results. Some analytical test methods which require modifications are: pH Water content Cloud point Neutralization number Carbon residue
Test methods commonly used for mineral oils which have no meaning when applied to PAGs include: Aniline point Boiling point Solvent precipitation tests
Copyright © 1994 CRC Press, LLC
Volume III
PHYSICAL PROPERTIES OF POLYALKYLENE GLYCOLS
255
Polyalkylene glycols differ physically and chemically from petroleum, animal, and vegetable oils. Because they are synthetic products, their manufacture can be controlled and varied to degrees not possible with natural oils and lubricants. PAGs contain no contaminants such as heavy metals, phenols, sulfur compounds, or polynuclear aromatics that may be associated with oil-based products.1 PAGs have a number of characteristics that make them desirable lubricants. These chemical and physical properties are briefly described in the following paragraphs.
Lubricity PAGs have excellent load-carrying capacity, film strength, and antiwear properties relative to mineral oil base stocks.2 Because of their good lubricity, high viscosity indices, and resistance to sludging, PAG lubricants are used in industrial applications where the long life of gears, bearings, and other vital machinery parts is important.3 The good lubricity of formulated PAG lubricants is especially evident in worm gears, where significant reductions in temperature and increases in efficiency have been demonstrated.4 Viscosity PAGs are commercially available in viscosities from 10 to 75,000 cSt at 40°C (Table 1, Figures 1 and 2). They have high viscosity indices, ranging from 180 to over 250 without the use of viscosity index improvers (Table 1). PAGs thus show significantly less change in viscosity with temperature than do petroleum oils whose viscosity indices are typically less than 100. PAGs are Newtonian fluids which exhibit good shear stability.
Pour Point Random PAGs that contain at least 50 wt% polymerized PO have low, stable pour points, typically below—20°C (Table 1). When the PAGs are made from feeds containing more than 50 wt% EO, they will solidify at higher temperatures. The further the EO content is increased above 50%, the higher the pour point. The pour point of a blocked PAG is very dependent on the size of the poly(ethylene oxide) block. As either the molecular weight or the concentration of polymerized EO in the blocked PAG is increased, the size of the poly(ethylene oxide) blocks grows, resulting in a higher pour point. Because of their long poly(ethylene oxide) segments, most blocked PAGs that contain more than 50% polymerized EO are solids or pastes at room temperature.
Chemical Stability PAGs exhibit good chemical stability. Their reactivity is similar to that of organic alcohols. PAGs do not go rancid and they resist hydrolysis in acidic, basic, or neutral aqueous solutions. Elastomer Compatibility PAGs are compatible with most natural and synthetic elastomers and gasket materials. The compatibility of PAGs with some common classes of elastomers is shown in Table 2. However, because elastomer characteristics can vary, it is important to determine the compatibility between PAGs and the specific elastomeric products that are being considered for use in critical applications.
Corrosion PAGs are noncorrosive to iron, steel, brass, bronze, and aluminum under normal operating conditions. Their use in industrial machinery or hydraulic systems is relatively unrestricted. Inhibitors are often incorporated to control the effects of corrosive moisture conditions.1
Copyright © 1994 CRC Press, LLC
256
CRC Handbook of Lubrication and Tribology
Copyright © 1994 CRC Press, LLC
Volume III
257
FIGURE 1. The viscosity of assorted oil-soluble polyalkylene glycols as a function of temperature. (From UCONRFluids & Lubricants, Union Carbide Corp., Tarrytown, NY.)
Thermal and Oxidative Stability PAGs exhibit good thermal stability when inhibited with antioxidants.2 Most PAG lubricants designed for use above 40°C or under conditions of heavy aeration are formulated with antioxidants. PAGs also resist forming sludges or varnishes because their breakdown products are usually either volatile or soluble in the base polymer.5 Cleanliness Free carbon, or coke, is not formed during most high temperature applications when PAGs are used. In the presence of air, clean burn-off is usually achieved. Ash content, Conradson carbon, and Ramsbotton carbon values are normally less than 0.01%.1
Flash Points PAGs have high flash points. Typical flash points for a variety of base polymers are shown in Table 1. The addition of an antioxidant significantly increases the flash point of a PAG.
Copyright © 1994 CRC Press, LLC
258
CRC Handbook of Lubrication and Tribology
FIGURE 2. The viscosity of assorted water soluble polyalkylene glycols as a function of temperature. (Fran UCONRFluids & Lubricants, Union Carbide Corp., Tarrytown, NY.)
Polyalkylene glycols that are inhibited with antioxidants generally have higher flash points than petroleum oils of the same standard viscosities (cSt at 40°C).1 Therefore, in instances where safety practices do not allow the use of a fluid at temperatures above its flash point, PAGs can be used at higher temperatures. This is an important characteristic in heat transfer fluids, calender lubricants, and other high temperature functional fluids.
Solubility One unique feature of PAGs is that their solubility can be varied from oil soluble to completely water soluble (Table 3). This is achieved primarily by changing the EO to PO ratio of the monomer feed. The higher the percentage of EO in the feed, the more water soluble the polymer will be (Figure 3). Conversely, the higher the percentage of polymerized PO, the more oil soluble the resulting PAG. Copyright © 1994 CRC Press, LLC
Volume III
259
Inverse Solubility The solubility of PAGs in water decreases with increasing temperature. At a temperature known as the “cloud point”, the polymer becomes insoluble in water and comes out of solution to form a milky suspension. The more polymerized EO in the PAG, the higher the cloud point (Table 1). Inverse solubility enables concentrated polymer films to form on hot metal surfaces that contact aqueous PAG solutions. This unique characteristic is utilized in synthetic metal working fluids and polymer quenchants.
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
260
FORMULATING WITH POLYALKYLENE GLYCOLS
Polyalkylene glycols are often formulated with additives to improve oxidative stability, to increase load-carrying capability under extreme pressure conditions, to reduce corrosion, or to give a particular special effect. Water-soluble additive systems can be employed in cases where dilution with water is desired. Antioxidants significantly improve the thermal and oxidative stability of PAGs and increase their flash points. Antioxidants are usually added to PAG-based products which will see extended service at temperatures above 40°C or undergo heavy aeration. Such applications include gear, bearing, and calender lubricants, heat transfer fluids, and mold release agents. Lubricity additives are added to PAGs to improve their performance in applications where boundary or extreme pressure conditions are experienced. Gear and compressor lubricants, hydraulic fluids, and metalworking fluids are formulated with lubricity additives. Corrosion inhibitors are commonly incorporated to inhibit ferrous and yellow metal corrosion and are particularly important in applications where water is present. PAGs used in conjunction with water include fire-resistant hydraulic fluids, quenchants, and metalworking fluids. Copyright © 1994 CRC Press, LLC
Volume III
261
FIGURE 3. The viscosity of aqueous solutions of assorted polyalkylene glycols. (From UCONR Fluids and Lubricants, Union Carbide Corp., Tarrytown, NY, 1987.)
Other additives such as antifoams, dyes, wetting agents, and biocides are sometimes added to PAGs to achieve specific properties.
COMMERCIAL APPLICATIONS OF POLYALKYLENE GLYCOLS
Gear Lubricants Polyalkylene glycols can be formulated into extreme pressure lubricants for enclosed industrial gears. These formulated products provide excellent lubrication, stability, and extended service life while eliminating many of the problems commonly encountered with petroleum lubricants. Their use can often result in less friction and wear, lower operating temperatures, reduced energy consumption, and extended service life.2–4 PAG gear lubricants are used in industrial bearings and in a large variety of helical, herringbone, bevel, spur, and worm gear designs.3,4,6 Gear lubricants based on PAGs are available in a wide range of viscosities and have excellent viscosity-temperature characteristics. In cold weather, their high viscosity indices and low pour points allow low start-up torques which prevent motor overload. In hot climates and under high operating temperatures, the high viscosity indices of PAGs enable them to better maintain their viscosity and thus provide more effective lubrication. The need for seasonal lubricant changeouts can often be eliminated by using PAG products.
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
262
FIGURE 4. The FDA status of various random polyalkylene glycols.7,21
Food Grade Lubricants Linear, random PAGs having number average molecular weights of 1500 and above ma; be used to lubricate food processing and packaging machinery in accordance with Regulator 21 CFR 178.3570 of the Food and Drug Administration.7 This regulation sets a limit of 10 ppm on the amount of lubricant that may get into food as a result of incidental contact. Food grade PAGs typically have viscosities of 150 cSt or higher at 40°C (Figure 4). The) can be formulated with food grade additives to form high performance products which exhibit excellent lubricity, good oxidative and thermal stability, high viscosity indices, and low pour points. In many applications these formulated PAG lubricants perform significantly better than other products identified in this regulation, namely, white oils, castor oil, and polybutenes. Copyright © 1994 CRC Press, LLC
Volume III
263
Compressor Lubricants Compressor lubricants based on PAGs are being used extensively in centrifugal, reciprocating, rotary screw, and sliding vane compressors. They are employed in the compression of a variety of gases including helium, hydrogen, nitrogen, natural gas, ethylene, air stack gases, land-fill gases, and fluorocarbon refrigerants.1,6,8 The success of PAG-based compressor lubricants is due to their excellent lubricity, high temperature stability, and resistance to sludge and varnish formation. Most important, the solubility characteristics of a PAG can be varied from hydrophobic to hydrophilic. This enables the formulation of lubricants that have low solubility in the gas being compressed. As a result, the lubricant is less susceptible to dilution by the compressed gas. Less lubricant dilution means less drop in viscosity and therefore better lubricity. It also results in reduced lubricant consumption, less leakage, and lower maintenance costs.
Water-Glycol Fire-Resistant Hydraulic Fluids Water-glycol hydraulic fluids are used when the leakage of a flammable product could result in a fire. These conditions are found in a number of places, including the steel and aluminum industries, the military, and the mining industry. If a hydraulic line breaks or develops a leak in one of these applications, the working fluid can be sprayed under high pressure onto an ignition source. Water-glycol hydraulic fluids are one of only a few types of products designed to reduce the risk of fire in case of line failure. A number of water-glycol hydraulic fluids are approved by Factory Mutual Engineering as “Group I Less Hazardous Hydraulic Fluids”.9 Water-glycol hydraulic fluids were developed to meet the demands for a relatively low cost, fire-resistant hydraulic fluid. They contain water, a glycol, a polyalkylene glycol thickener, and a proprietary additive package to improve lubricity and to provide resistance to both liquid and vapor phase corrosion.10 These products can be formulated to provide low temperature properties which allow for year-round outdoor use. However, because they contain significant amounts of water, water-glycol hydraulic fluids are usually restricted to systems where the operating temperature is below 65°C (150°F). It is important to monitor the water concentration of a water-glycol hydraulic fluid during use. Too much water leads to poor lubricity while too little will decrease the fire resistance. Water concentration can be monitored by Karl-Fisher titration or refractive index. The corrosion inhibitor package in water-glycol hydraulic fluids must also be occasionally replenished. The inhibitor concentration can be monitored through alkalinity titration. The concentrated corrosion inhibitor package is usually available from the fluid supplier. Conventional water-glycol hydraulic fluids are limited to 2000 psi in vane pumps and 2500 to 3500 psi in gear and axial piston pumps. A new type of water-glycol hydraulic fluid has recently been developed which can be used at operating pressures of up to about 5000 psi.11
Metalworking Fluids PAGs are used as lubricity additives in water-soluble or “synthetic” cutting and grinding fluids.10,12 They have also been utilized as lubricants in forming operations such as drawing, stamping, and rolling.6,13,14 Most metalworking fluids based on PAGs form true solutions in water. As a result, they are in general more resistant to microbial attack and easier to maintain than “soluble oil” metalworking fluids, which are actually oil-in-water emulsions. PAGs work in aqueous metalworking fluids by taking advantage of their inverse solubility in water.10,12 When a PAG-containing metalworking fluid comes in contact with the hot workpiece and tool or die, the temperature of the solution is raised above the cloud point of the polymer. The PAG then comes out of solution, forming finely dispersed droplets which
Copyright © 1994 CRC Press, LLC
264
CRC Handbook of Lubrication and Tribology
coat the hot metal surfaces with a thin layer of lubricant. This PAG film provides good hydrodynamic lubricity where it is needed, at the point of cut or deformation. PAGs are usually combined for synergistic response with water soluble boundary or extreme pressure additives such as fatty acids or phosphate esters. Combinations of PAGs and fatty acids or phosphate esters have been shown to provide better lubricity than equivalent concentrations of either type of additive by itself.12 As a result of this synergy, these aqueous metalworking fluids provide excellent lubricity as well as the good cooling properties of water. Other additives such as corrosion inhibitors, antifoams, and biocides are often incorporated into the final metalworking fluid formulation as needed.
Mill and Calender Lubricants The high viscosity indices of PAGs and their good thermal stability in the presence of antioxidants make them excellent base fluids for mill and calender lubricants.10,15 They are used in large mills and calenders by the rubber, textile, paper, and plastic industries. Properly formulated petroleum oils meet the requirements for use in calenders operating at moderate temperatures. However, when calender roll temperatures exceed 350°F, petroleum lubricants tend to develop carbonaceous residues that may contribute to lubrication problems. Formulated PAG lubricants are recommended for this high temperature service because of their excellent thermal stability, high flash and fire points, low sludging tendencies, and good lubricating properties (Tables 4 and 5). The high viscosity indices of polyalkylene glycol lubricants enable them to better maintain their viscosity at elevated temperatures, resulting in enhanced lubricity when compared to petroleum oils. Textile Lubricants Polyalkylene glycols are utilized in practically every phase of textile fiber lubrication.2,10,16 Depending on the application, fiber type, and yam structure, the desired level of friction control
Copyright © 1994 CRC Press, LLC
Volume III
265
can be obtained by selecting the appropriate PAG lubricant. These lubricants have outstanding resistance to discoloration from heat and light and do not stain or discolor fibers. They also do not become rancid or gummy during storage or use and are thoroughly removed in conventional scouring processes. Their high smoke points and low volatility are also beneficial. Textile machine lubricants can be formulated from water-soluble PAGs.1 These products are used to lubricate weaving machinery in applications where oil contamination of the fabric is a problem. They provide excellent lubrication and are completely removed from many fabrics during most scouring processes. The water washability of these PAG-based textile machine lubricants can significantly reduce the number of seconds caused by oil staining. These formulated products exhibit good oxidative stability and corrosion protection. They also do not separate and will retain their water washability even after extended goods storage.
Rubber Lubricants PAGs, with their negligible solvent and swelling effects on most natural and synthetic rubber, are ideally suited as antistick, coating, and parting agents for the rubber industry.10,17,18In most cases, the lubricant is diluted with water or some other solvent (Table 6). Wetting agents, fine particle solids, glycerol, or silicone emulsions may also be added to produce desired properties. Water-soluble PAGs are widely used as mandrel release agents in the manufacture of EPDM and NBR (nitrile) cured hoses. Poly alkylene glycols are also used as antistick agents for uncured rubber, as mold release agents, as machining lubricants for hard rubber, and as lubricants for rubber packings, O-rings, and seals.
SELECTION OF THE APPROPRIATE POLYALKYLENE GLYCOL LUBRICANT
The selection of the appropriate polyalkylene lubricant is based on the physical and chemical properties of the product. Properties that are commonly considered include viscosity, pour point, water or gas solubility, lubricity, thermal and oxidative stability, and flash point. Previous experience and data from pertinent experiments should be used when available to aid in the decision.1 When selecting a PAG lubricant to replace a petroleum product, the viscosity comparison should be made at the system operating temperature rather than at SAE or ISO standard temperatures. This is necessary because of the high viscosity indices of PAGs relative to petroleum oils.
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
266
The solubility characteristics of PAGs should also be considered when choosing the appropriate lubricant. The significant oil solubility of PAGs made with high concentrations of PO makes them useful lubricants in applications where contamination from petroleum oils is likely. Water-soluble PAGs are often used as process fluids or as lubricants where water washability is important. PAGs also have unique gas solubilities, making them ideal lubricant candidates for a variety of compressor applications. The selected PAG lubricant should be formulated with the additives that are necessary to achieve satisfactory performance in a specific application. These additives may include antioxidants, corrosion inhibitors, extreme pressure additives, and dyes.
HANDLING AND STORAGE OF POLYALKYLENE GLYCOLS
In general, PAGs are stable, noncorrosive materials that can be stored in carbon steel tanks.1 Since many of these fluids have a viscosity greater than 1000 cSt at temperatures below 0°F, heated tanks are usually provided for outside storage. Either hot water or low-pressure steam (15 psig or less) can be used in external heat transfer panels or internal coils to heat the stored PAG lubricant. Electrical heating by means of heating tape or cable is also satisfactory. PAGs exhibit solvency characteristics different from those of petroleum oils. PAGs will soften and lift many industrial coatings.1,10 Catalyzed epoxy, epoxy-phenolic, and modified phenolic coatings have performed well. Alkyd and vinyl coatings are unsatisfactory. If possible, PAGs should be stored in clean carbon steel tanks or unpainted lubricant reservoirs. If internal coatings are present and cannot be removed, it is important to clean all filters and strainers frequently, especially during the initial period of use. PAGs exhibit only limited solubility in mineral oils; therefore, contamination with petroleum oils should be avoided. Storage tanks previously used for petroleum products should be cleaned before PAGs are introduced. PAGs, especially those containing more than 70% polymerized EO, are hygroscopic. If water content is critical, precautions should be taken to prevent atmospheric moisture from entering the storage tank. A desiccant unit can be installed on the vent line or the tank can be blanketed with dry air or nitrogen.1 Centrifugal pumps are adequate to transfer most PAGs.1 However, if viscosities exceed 500 cSt, a rotary or gear pump is preferable. Transfer lines should be carbon steel and of adequate size to handle the desired flow and viscosity with a reasonable pressure drop. A 3in. line should be provided for unloading bulk shipments.
PRODUCT SAFETY
When considering the use of a PAG or any other chemical, the product’s material safety data sheet should be obtained from the manufacturer and carefully reviewed to ensure safe use.
SUMMARY
Polyalkylene glycols represent a diverse family of fluids and lubricants. They exhibit a wide range of both gas and liquid solubilities, good inherent lubricity, high viscosity indices, and low toxicity. Polyalkylene glycols also have high flash points, burn cleanly, and resist sludging or varnish formation. Polyalkylene glycols are advantageously used in a wide variety of industrial operations. Their excellent stability, high viscosity indices, and good lubricity make them effective gear and bearing lubricants, heat transfer fluids, and calender lubricants. The unique solubility Copyright © 1994 CRC Press, LLC
267
Volume III
properties of polyalkylene glycols has lead to their use as compressor lubricants, fire resistant hydraulic fluids, metal working fluids, and textile lubricants. While some of the major applications for polyalkylene glycols have been described, the number of uses for these versatile products continues to grow.
1. 2. 3. 4. 5. 6.
7. 8.
9. 10. 11. 12. 13. 14.
15. 16. 17. 18. 19. 20. 21. 22.
REFERENCES
Technical Literature, UCONR Fluids & Lubricants, Union Carbide Corp., Tarrytown, NY, 1987. Klamann, D., Lubricants and Related Properties, Verlag Chemie, Deerfield Beach, FL, 1984. McCabe, M. A., Chemical plant applications for polyalkylene glycols, Lubr. Eng., 38, 749, 1982. Technical Literature, Triboschlussel, 2nd ed., Kluber Lubrication, Munich, 1977, 4. Moreton, D. H., Review of synthetic lubricants, Lubr. Eng., 10, 65, 1954. Sweat, C. H. and Langer, T. W., Some industrial experiences with synthetic lubricants, Mech. Eng., 73, 469, 1951. US FDA, Code of Federal Regulations, 21 CFR 178.3570, 1990. Garg, D. R., Polyalkylene glycol based compressor lubricants, in 6th Annu. Reciprocating Machinery Conf. Proc., #91–509, Pipeline and Compressor Research Council, Dallas, TX, 1991. Factory Mutual Research Corporation, 1151 Boston-Providence Turnpike, Norwood, MA. Mueller, E. R. and Martin, W. H., Polyalkylene glycols: uniquely water soluble, Lubr. Eng., 31, 348, 1975. Lewis, W. E. F., Energy transmitting fluid, U.S. Patent No. 4,855,070, 1989. Brown, W. L., The role of polyalkylene glycols in synthetic metalworking fluids, Lubr. Eng., 44, 168, 1988. Marx, J., Synthetic lubricant for machining and chipless deformation of metals, U.S. Patent No. 3,980,571, 1976. Felton, G. F., Jr., Low smoking lubricating composition for cold heading operations, U.S. Patent No. 3,983,044, 1976. Technical Literature, UCONR Calender Lubricants, Union Carbide Corp., Tarrytown, NY, 1978. Trevor, J. S., New glycols as textile lubricants and conditioning agents, Textile Recorder, 67, 86, 1950. Russ, J. M., Jr., “UCON” synthetic lubricants and hydraulic fluids, ASTM, Tech. Pap., No. 77:3–11, 1947. Gunderson, C. G, and Hart, A. W., Synthetic Lubricants, Reinhold Publishing, New York, 1962. Technical Literature, Technical Data on PLURONICR Polyols, BASF Wyandotte Corp., 1978. Technical Literature, Technical Data on PLURONICR R Nonionic Surface Active Agents, BASF Wyandotte Corp., Wyandotte, MI, 1978. Technical Literature, UCONR Fluids & Lubricants, Union Carbide Corp., Tarrytown, NY, 1992. Technical Literature, UCONR Fluids & Lubricants, Union Carbide Corp., Tarrytown, NY, 1979.
Copyright © 1994 CRC Press, LLC
PHOSPHATE ESTERS
Michael P. Marino and Douglas G. Placek
INTRODUCTION
Phosphate esters were first used in the lubrication field as antiwear additives in automotive crankcase and gear oils and aircraft engine oils in the late 1930s. Developments in aircraft design during World War II featured the increased use of hydraulic control systems, and phosphate esters were introduced as less flammable hydraulic fluids. This chapter will review commercially available triaryl, trialkyl, and dialkyl/aryl phosphate esters, and will discuss their use as fluid base stocks, lubricant base stocks, and antiwear additives.
CHEMISTRY
Trisubstituted, or tertiary, esters of orthophosphoric acid are represented by the general structure:
Commercially available trialkyl phosphate esters are symmetrical (R‘ = R‘’ = R‘’‘), and include tri-n-butyl, tributoxyethyl, and trioctyl phosphate. The first commercially significant triaryl phosphate esters produced were tricresyl phosphate (TCP) and trixylenyl phosphate (TXP), which are referred to as “natural” phosphate esters because the cresols and xylenols used as raw materials are derived from petroleum oil or coal tar. Commercial TCP is synthesized from a mixed isomer feedstock, resulting in a product with a variety of isomeric components. Cresol, xylenol, and phenol can be blended in different ratios and used as raw materials to create a mixed natural triaryl phosphate. None of these mixed esters are commercially significant in the U.S. at present, but products such as cresyl diphenyl have been used in the past. “Synthetic” analogs of the natural phosphate esters were developed in the late 1960s, which reduced toxicity concerns and lowered product costs. Isopropylphenyl phosphates and tertiarybutylphenyl phosphates are now commercially available in a variety of viscosity grades. Typical chemical structures for two “synthetic” phosphate esters are given below:
0-8493-3903-0/94/$0.00 + $.50 © 1994 by CRC Press, Inc.
Copyright © 1994 CRC Press, LLC
269
270
CRC Handbook of Lubrication and Tribology
Commercial products are a mixture of compounds of varying molecular weights and isomer content, as well as triphenyl phosphate. The only alkyl/aryl phosphates which find use in the lubrication industry are dibutyl phenyl phosphate and isodecyl diphenyl phosphate. Other alkyl/aryls find use in the plastics industry as flame retardant plasticizers.
Triaryl Phosphate Ester Production Commercial triaryl phosphate esters are produced by reacting phosphorus oxychloride with phenolic compounds in the presence of magnesium chloride or aluminum chloride catalyst:
An excess of aromatic alcohol is maintained in order to avoid the presence of intermediates or acid esters in the final product. The family of “synthetic” phosphate esters is produced from a mixture of phenol and alkyl substituted phenols, as in Equation 2. Thermodynamics and steric hindrance determine which alcohols will react with greater speed, and thus define the molecular weight and isomeric distribution of the final product.
The crude phosphate ester is refined in a series of filtration, distillation, or washing steps to remove catalyst, excess alcohols, and impurities.
Trialkyl and Alkyl Aryl Phosphate Ester Production Equation 3 describes a commercial process for production of trialkyl phosphate esters, which involves the reaction of phosphorus oxychloride with a sodium alkoxide.
Commercial trialkyl phosphate esters are symmetrical; however, an unsymmetrical mixed ester could be produced from a mixed alcohol feed stream. Mixed alkyl/aryl phosphate esters are best prepared by a stepwise reaction scheme. An intermediate chloridate is created by conducting the reaction in Equation 1 with an aliphatic alcohol and excess POCl3. The neutral phosphate ester is completed by reacting the intermediate chloridate with excess sodium arylate, as in Equation 3. Copyright © 1994 CRC Press, LLC
Volume III
271
PHYSICAL PROPERTIES Tables 1 and 2 provide a review of the basic physical properties of commercially available phosphate esters.1.5 In order to simplify the presentation, the following abbreviations have been used throughout the text: TPP—Triphenyl phosphate TCP—Tricresyl phosphate TXP—Trixylenyl phosphate IPPP—Isopropylphenyl phosphate TBPP—Tertiary-butylphenyl phosphate DAAP—Dialkyl aryl phosphate TBP—Tributyl phosphate TBEP—Tributoxyethyl phosphate TOP—Trioctyl phosphate
The abbreviation TBPP/46 describes a t-butylphenyl phosphate ester in the ISO 46 viscosity range.
Copyright © 1994 CRC Press, LLC
272
CRC Handbook of Lubrication and Tribology
Lubrication Phosphate esters provide boundary lubrication by chemically binding with metal surfaces at frictional contact points to form thin layers of iron phosphates and/or iron phosphides. These low melting eutectic surface films fill the valleys between surface asperities, increase the surface area of the contact zones, and reduce pressure points. Under friction, the iron phosphate films are sheared, preserving the integrity of the lubricated surface and reducing wear. The good antiwear protection offered by neat phosphate ester fluids is demonstrated by four-ball wear data in Tables 1 and 2, and the vane pump wear data in Table 3. Phosphate esters can also be used synergistically with other lubricity additives to improve the overall wear protection offered by an additive package. At heavy loads, the mild EP (extreme pressure) characteristics of phosphate esters reduce friction and ease the transition into the EP additive regime where high temperatures and pressures activate chlorinated and sulfurized additives. Copyright © 1994 CRC Press, LLC
Volume III
273
In addition, phosphate esters contain no metals or ash-forming constituents; they are noncorrosive, colorless, and odorless. Ashless additive technology is critical for applications where severe operating environments may cause degradation and abrasive deposits cannot be tolerated.
Fire Resistance Hydraulic fluids used in the proximity of electric sparks, molten metal, open flames, or high temperature surfaces represent a serious fire hazard. Phosphate esters are inherently fire resistant, will resist ignition under severe conditions, and will not support combustion if ignited. Phosphate esters are significantly harder to ignite than synthetic hydrocarbons with equivalent viscosities and flash points.6 Mineral oils, synthetic hydrocarbons, and synthetic esters all have heats of combustion that are significantly higher than phosphate esters (~10,000 kcal/kg vs. 7,500 kcal/kg), which allows them to sustain combustion. Because flash point, fire point, and auto ignition temperature tests do not fully assess safety properties, other tests (Table 4) have been developed to evaluate the potential hazard posed by a hydraulic fluid. Some of the tests identify chemical properties, while others attempt to simulate actual industrial crisis situations. Commercial phosphate ester hydraulic fluids will pass all of the tests listed in Table 4. Phosphate esters are classified as HF-D type fluids, or as “less flammable fluids”, group II, by Factory Mutual Research Corporation.7
Oxidative Stability The central P-O-C bond structure of a phosphate ester has excellent stability. Studies8 have demonstrated that the molecule’s organic groups determine its ultimate stability. In general, the oxidative stabilities of triaryl phosphate esters are superior to trialkyl phosphates.8 Table 5 presents data9 describing the onset of oxidation for a variety of commercial products, as determined by DSC (differential scanning calorimetry). Triphenyl phosphate is the most oxidatively stable, showing no degradation at temperatures as high as 340°C. The TBPP esters exhibit the best oxidative stability of any commercially available liquid base stocks. IPPPs, TCP, and TXP exhibit good oxidative stability, while trialkyl phosphate esters begin to oxidize in the same temperature range as paraffinic mineral oil. Functional range may also be limited by thermal or volatility characteristics. Table 5 demonstrates that phosphate esters are oxidatively stable up to and above the point where significant evaporation occurs.
Copyright © 1994 CRC Press, LLC
274
CRC Handbook of Lubrication and Tribology
Hydrolytic Stability Hydrolysis can be the most significant threat to phosphate ester stability. The reaction yields acid esters which can be corrosive and also catalyze further degradation. Unformulated natural phosphate esters perform better than most synthetic phosphate esters in the standard test for hydrolytic stability (ASTM D-2619, beverage bottle hydrolytic stability test), although all phosphate esters can be formulated to pass industry requirements. Normal maintenance procedures will ensure long fluid life.
Viscosity Characteristics Commercial triaryl phosphate esters are manufactured in ISO viscosity grades 22 through 100. The pour points of these fluids fall between -30 to -5°C, and their viscosity indices are typically less than 50. Trialkyl phosphate esters are typically used in low temperature applications due to their low pour points (< -70°C). The trialkyl esters have viscosity indices comparable or superior to paraffinic mineral oils. Specific viscosity-temperature data are listed in Tables 1 and 2; viscosity profiles are shown in Figure 1. Fluids with wide operating ranges can be created by blending trialkyl and triaryl phosphate esters. Mixtures generally take on the desirable characteristics of both components, having excellent load-bearing properties, high VI, low pour point, and fire resistance. Aircraft hydraulic fluids (see Table 10) are prime examples of blended phosphate ester fluids.
Solubility/Compatibility Characteristics Phosphate ester fire-resistant hydraulic fluids are miscible with all organic esters, but are not compatible or miscible with water-containing hydraulic fluids such as water glycols, waterin-oil emulsions, and high water-base fluids (HF-A, HF-B, and HF-C types). Phosphate ester fluids can be used in contact with all standard construction metals including cast iron, all steels, aluminum, copper, brass, bronze, silver, zinc, cadmium, titanium, and magnesium. Copyright © 1994 CRC Press, LLC
Volume III
FIGURE 1. Viscosity vs. temperature relationship for commercial phosphate esters.
275
Copyright © 1994 CRC Press, LLC
276
CRC Handbook of Lubrication and Tribology
Note: Key—R, recommended; A, acceptable, check with supplier, U, unsatisfactory. Materials not listed here should be considered unsatisfactory for use with phosphate esters. Always consult suppliers for specific recommendations.
The strong solvency exhibited by phosphate esters can lead to compatibility problems with standard seals, hoses, paints, and coatings. Elastomers typically used with hydrocarbon lubricants may not be satisfactory, however, a wide variety of materials are available which are compatible (see Table 6). Phosphate esters can be used as additives in a wide variety of lubricant base stocks due to their excellent solubility. In addition, phosphate esters are neutral and are not known to interfere with the performance of other lubricant additives like corrosion inhibitors, antioxidants, metal passivators, detergents, dispersants, defoamers, surfactants, wetting agents, or demulsifiers. Phosphate esters are miscible with all other carboxylic esters, and are typically soluble at over 10% in hydrocarbon oils. The use of phosphate esters in polyalphaolefins, mineral oils, and silicone fluids can also make it possible for these fluids to accept other additives. PAOs and silicones are otherwise difficult to formulate. The solvency properties of trialkyl and alkyl/aryl esters can sometimes be exploited as seal swelling agents. Trialkyl phosphate levels of 1 to 2 wt% can give a hydrocarbon fluid the ability to impart a controlled degree of swell to standard nitrile seals, in order to reduce oil leakage or to extend the life of aging, shrinking, or deteriorating seals. Volatility Phosphate ester boiling points are significantly higher than other organic compounds with similar viscosities. Their low vapor pressures contribute to high flash points and fire resistance. Table 7 presents the boiling points and vapor pressures of a variety of phosphate ester fluids.
Copyright © 1994 CRC Press, LLC
Volume III
277
Their low volatility characteristics translate into very low evaporative losses and excellent performance under conditions of vacuum. Thermal Properties Heat transfer properties of a fluid are critical when assessing a system’s heating or cooling capacity. Phosphate ester specific heats (Cp) are presented in Table 8. Other thermal properties which can be applied to the entire class of triaryl phosphate esters include:
Compressibility Triaryl phosphate esters exhibit lower compressibility than mineral oils, and demonstrate excellent performance as hydraulic fluids in high and low pressure systems. Hydraulic fluids
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
278
must have low compressibility and high bulk modulus to transmit power with low heat buildup and energy losses. Table 9 presents typical compressibility and bulk modulus values for triaryl phosphate esters.
FLUIDS AND LUBRICANT APPLICATIONS
Military and commercial performance requirements for safety, fire resistance, and wide operating temperature range provided the impetus for the development of phosphate esters as synthetic lubricant base stocks in the 1940s and 1950s. Although triaryl phosphates are widely used as antiwear lubricant additives, the following section outlines only uses of phosphate esters as fluid base stocks. Formulation requirements and additive selection will be discussed.
Industrial Hydraulic Fluids Fire-resistant industrial hydraulic fluids represent the largest volume commercial use of phosphate esters. Phosphate ester fluids provide lubrication that is equal to formulated hydrocarbon fluids and superior to water-based fluids, and they permit operation at higher temperatures and pressures than water based products. The examples of industrial hydraulic fluid performance in Table 10 supplement the physical property data in Tables 1 and 2. These fluids are commercially available in viscosity ranges from ISO 22 through ISO 100. Patent literature indicates mat antioxidants, metal passivators, antifoams, VI improvers, and acid acceptors may be incorporated in phosphate ester fluid formulations. Phosphate esters have also been blended with other synthetic and mineral oil-base stocks to extend performance or complement the properties of the primary base. Commercial fluids which are blends of triaryl phosphates and mineral oils give a lower degree of fire resistance but offer cost savings compared to straight phosphate fluids. Phosphate esters have also been successfully blended with silicones13 and polyol esters.14 Military Hydraulic Fluids The U.S. Navy uses a phosphate ester fluid in hazardous areas, chiefly in flight deck elevators of aircraft carriers. This ISO 46 fluid is specified under MIL-H-19457D to be 100% t-butylphenyl phenyl phosphate ester.
Aircraft Hydraulic Fluids Phosphate ester hydraulic fluids meeting SAE AS 1241A and airframe manufacturers’ specifications are used in almost all commercial aircraft worldwide. These fluids are mixtures,
Copyright © 1994 CRC Press, LLC
Volume III 279
Copyright © 1994 CRC Press, LLC
280
CRC Handbook of Lubrication and Tribology
Copyright © 1994 CRC Press, LLC
Volume III
281
chemical or physical, of alkyl and aryl phosphates and contain antioxidants, corrosion and erosion inhibitors, and VI improvers.15–17 Through blending, these fluids achieve the low pour points and high viscosity indices of trialkyl phosphate esters, but maintain the higher viscosity and antiwear performance of triaryl phosphate esters. Table 10 lists physical properties and specifications of aircraft hydraulic and other low temperature hydraulic fluids.
Electrohydraulic Control Fluids Steam flow through a power plant turbine is regulated by an electrohydraulic (EHC) control system for which all major turbine manufacturers specify a fire-resistant phosphate ester fluid. Conversion to phosphate ester EHC fluids reduced the frequency of fire incidents in power plants by more than 50%. EHC fluids, ISO 46 phosphate esters, are distinguished from industrial fluids by low chlorine and particulate content and high electrical resistivity, specifications aimed at reducing servo valve erosion.
Turbine Lubricants Phosphate ester fluids find commercial use as bearing lubricants in gas turbine18–19 and large steam turbines,20–21 uses in which fire resistance and long term stability are required. These lubricants, based on triaryl phosphates, are usually formulated with antioxidants, corrosion inhibitors, and antifoam agents.22–23 Eleven General Electric Frame 7000 gas turbines used as drives for 60 to 70 MW generators have been in operation since the 1970s on phosphate esters and have run between 15,000 and 30,000 hours of operation between lubricant changes.24 A large number of small gas turbines, used as pump drives on gas pipelines in the U.S. and Canada, have been running on phosphate ester main bearing lubricants since the mid 1970s.19 More recently, in the Confederation of Independent States, seven 220 to 1000 MW steam turbine-generators have been operating for long periods on phosphate ester lubricants, some for well over 30,000 hours,25 with excellent fire safety and lubrication performance. Gas turbine lubricants are typically ISO 32 viscosity grade products. Tables 1, 2, and 10 give physical and performance data.
Other Lubricants Use of phosphate ester lubricants for air compressors dates from the 1960s. Where high operating temperatures of certain air compressors might cause ignition of mineral oil lubricants, phosphate esters offer low volatility, high flash points, and fire resistance. Limited use has also been reported of phosphate esters as gear lubricants26 and metalworking lubricants.27 Environmental concerns have prompted research into low heat rejection diesel engines with low emissions and greater fuel efficiency. These engines will impose demands beyond the capability of petroleum oils, which has led to the development of lubricants based on blends of polyol esters and phosphate esters.28 Considerable development effort is also being directed at a unique lubrication system where phosphate ester vapor is delivered to the piston ring zone of the cylinder liner and immediately consumed.29–32
FORMULATING PHOSPHATE ESTER FLUIDS
Since the most widely used triaryl phosphates are inherently stable thermally, hydrolytically, and oxidatively, additives may not be necessary for satisfactory performance in standard industrial hydraulic systems. However, additives are used to meet more difficult performance specifications. For example, antioxidants may be used when normal operating temperatures are above 130°C, or an appropriate additive will be used when antirust performance, high viscosity index, or low foaming tendency are called for.
Copyright © 1994 CRC Press, LLC
282
CRC Handbook of Lubrication and Tribology
Hydrolytic Stabilizers A variety of compounds reportedly improve hydrolytic stability of phosphate esters. Basic nitrogen compounds such as urea, semicarbazones, 33 and aliphatics diimides34 have been used in recent work. Epoxy compounds, which received earlier attention, appear to still be the most commonly used. Orthoformate esters35 reportedly react preferentially with water in a phosphate ester system, minimizing the amount available for phosphate hydrolysis reactions. Antioxidants Hindered phenols are widely used as antioxidants in a variety of chemical systems including phosphate esters.5,36,37 These act as free radical inhibitors which disrupt the oxidation chain reaction. Tris(tributylphenyl) chlorotitanate38 has also been reported to be useful, as have been esters of thiopropionic acid22 and certain amine antioxidants.
Metal Passivators Phosphate esters themselves have been used as metal passivators, for ferrous metals.39 Since copper and its alloys can be attacked by acidic compounds, passivators for these “red” metals are often used in phosphate systems. Benzotriazole is probably the most common passivator (see Reference 22, for example) but other sulfur and nitrogen compounds have been reported, including quinizarin, thiadiazoles, sarcosines, napthyl amines, and aliphatic aminomethyl methane compounds. Metal passivators also can reduce potential fluid deterioration since copper and iron are catalysts for phosphate ester degradation reactions.
Erosion Inhibitors Modern aircraft systems were found to be particularly susceptible to servo-valve wear40 from electrical “streaming current” potential generated by rapid flow of fluid over valve surfaces. This problem has been essentially eliminated through control of chloride or chlorine content in the fluid, or by increasing the fluid conductivity so as to dissipate electrical charges through the fluid rather than through the metal of the valve. The most common erosion inhibitors include metallic salts of fluorinated surface active agents,41 pyridine derivatives of phosphate esters,42 and ethyl acetate.43
Viscosity Index Improvers Despite the relatively poor viscosity index (VI) of triaryl phosphate esters, VI improvers are not added to most industrial hydraulic fluids. In aircraft fluids, common polymeric VI improvers are used, such as polyacrylates and methacrylates, polypropylene glycol polyesters of azelaic acid, polyol esters,14 and polyoxyethylene and polyoxypropylene polyols.36 These additives can be used to formulate fluids with a VI of over 250 with good shear stability.
Corrosion Inhibitors Phosphate esters will not protect iron components from rust if water contamination becomes a problem. Effective additives for eliminating hydrated iron oxide in phosphate ester fluids include acid phosphate esters, neutral metal sulfonates, amines, and amine sulfonates. Because they can contribute to potential gel formation, calcium and magnesium sulfonates are best avoided; if used, they should not be overbased.
Defoamers Some phosphate ester applications may use standard silicone or organic polymer defoamers in order to overcome the effects of contamination or to avoid carrying entrained air into the system which might disrupt lubricant supply or cause sluggish hydraulic system response.
Copyright © 1994 CRC Press, LLC
Volume III
OPERATION AND MAINTENANCE OF PHOSPHATE ESTER FLUID SYSTEMS
283
Successful use of phosphate ester fluids depends upon proper system maintenance, including equipment preparation, good filtration, and periodic monitoring. The Electric Power Research Institute published an excellent review24 of worldwide experience with phosphate esters used in lubrication and EHC fluid areas. Factors involved in the design or conversion of phosphate ester fluid systems have been outlined by B. J. Wiggins44 and can also be obtained from fluid and equipment suppliers. Materials of construction must be reviewed to assure that all seals, hoses, filters, and paints are compatible with phosphate esters (see Table 6). If the system is being converted to phosphate esters, it should first be flushed and drained in consultation with the fluid supplier and refilled with the proper viscosity phosphate fluid. Phosphate esters are thermally and oxidatively stable for extended periods at temperatures up to 135°C, and temperatures as high as 230°C for short intervals. Thus, fluid degradation due to thermal stress is unlikely. Since the most significant threat to phosphate ester stability is hydrolysis, systems should be monitored for water contamination. Excessive water contamination and subsequent hydrolysis will lead to increased fluid acidity which in turn can lead to corrosion, fluid gelation, clogged filters, and sluggish system operation if left unchecked. Moisture accumulation can be controlled by equipping reservoir breather tubes with desiccant traps. Small quantities of free water (which will float on the surface) can be removed by skimming the surface of the reservoir, or by increasing the temperature to 60 to 70°C to evaporate the water. Water can also be controlled by constant by-pass filter systems with paper blotter or Fuller’s earth filters. If water levels are excessive, vacuum dehydration and/or complete fluid change-outs are alternatives. Contaminated or degraded phosphate ester fluids can usually be restored with filtration.45 Fuller’s earth and other clay media act by neutralizing acids to form calcium and magnesium phosphate salts. These tend to precipitate as gels if solubility limits are exceeded and excessive water is present. Activated alumina and ion exchange resins have shown promise as alternatives. Excessive contamination with a mineral oil will require a fluid change-out since complete separation is not possible and residual mineral oil will reduce the fire resistance. A good fluid maintenance program should include quarterly (monthly, if possible) monitoring of the parameters listed in Table 11.46
TOXICOLOGY
All commercial triaryl phosphate esters are classified as practically nontoxic by oral, dermal, or inhalation routes of exposure, according to OSHA standard 29CFR1910.1200. This classification is based on the very high LD50 and LC50 levels typical of commercial triaryl phosphates. Triaryl phosphates are not mutagenic or irritating to the skin or eyes. They are not listed as carcinogens by ACGIH, OSHA, or NIOSH. Certain organic phosphates, phosphites and phosphonates inhibit enzyme activity associated with the nervous system, resulting in disruption of neural transmissions and a condition known as OPBDN, organophosphorus induced delayed neurotoxicity. OPIDN occurs days or weeks after exposure, and results in numbness and motor dysfunction in the limbs. The damage is reversible if the exposure is mild, but long-term, high-dose exposure can result in permanent nerve damage. Several thorough reviews of OPIDN history and mechanisms have been published.47,48
Copyright © 1994 CRC Press, LLC
284
CRC Handbook of Lubrication and Tribology
Triorthocresyl phosphate (TOCP) has been clearly identified as a neurotoxin to humans. It and other ortho isomers are usually associated with most neurotoxic test results on phosphate esters. The neurotoxicity of current commercial TCP has been significantly reduced by manufacturing the product to contain over 98% meta and para isomers and virtually no TOCP. In addition to lower cost, reduced neurotoxicity was a major reason for the development of synthetic phosphates. The isopropylphenyl phosphates and t-butylphenyl phosphates show little to no neurotoxicity even at extremely high doses. Epidemiology studies on current and former phosphate ester production plant employees have not demonstrated any unusual pattern of toxicity, mortality, or disease. OPIDN has not been associated with any commercial alkyl phosphate esters. These compounds are not known to be carcinogenic or mutagenic, but several are skin irritants. Extended contact with tributyl or trioctyl phosphates can result in dermatitis.
ACKNOWLEDGMENT
The authors wish to thank Mr. Sundeep Shankwalkar and Ms. Dede LaMarche of the FMC Corporation for their expert assistance in the development of technical data and historical perspective that were not previously available. The authors also thank FMC Corporation for supporting this effort. REFERENCES
1. Durad® Lubricant Additives, FMC Corp., Philadelphia, PA, 1988. 2. Reolube® HYD Fire Resistant Fluids, FMC Corp., Trafford Park, U.K., 1993. 3. Fyrquel® Fire Resistant Hydraulic Fluids, Tech. Bull. No. 88–151, Akzo Chemicals Inc., Chicago, H, 1986. 4. Houghto-Safe® 100 Series Phosphate Ester Fluids, Tech. Bull. No. 2–276-F 2M, E. F. Houghton Co., Valley Forge, PA, 1990.
Copyright © 1994 CRC Press, LLC
Volume III
285
5.Schulz, W. W., Navratil, J. D., and Bess, T., Science and Technology of Tributyl Phosphate, Vol. 2, Part B, CRC Press, Boca Raton, FL, 1987, 89. 6. Placek, D. G. and Shankwalkar, S. G., Flammability limits of synthetic lubricant basestocks as measured by a modified oxygen index test, Lubr. Eng., 49, Jan 1993. 7. Factory Mutual System Approval Guide, Factory Mutual Research Corp., Norwood, MA, 1993. 8. Cho, L. and Klaus; E. E., Oxidative degradation of phosphate esters, ASLE Trans., 24(1), 119, 1979. 9. Shankwalkar, S. G. and Placek, D. G., Oxidation and weight loss characteristics of commercial phosphate esters, Ind. Eng. Chem. Res., 31, 1810, 1992. 10. Kelly, D. J., Resistance of materials to hydraulic fluids, Mach. Design, Jan. 1971, p. 100–103. 11. Hatton, R. E., Phosphate esters, in Synthetic Lubricants, Gunderson, R. C. and Hart, A. W., Eds., Reinhold, NY, 1962, chap. 4. 12. Chevron HyJet® IV-A, Phosphate Ester Aircraft Hydraulic Fluid, Tech. Bull., Chevron International Oil Co., San Francisco, CA, 1989. 13. Quaal, G., U.S. Patent 3,634,246 (to Dow Corning Corp.), 1972. 14. Ohba, K., Izumi, K., and Yasuda, S., U.S. Patent 4,298,489 (to Kao Soap KK), 1981. 15. Romano, J. and Shatynski, J. G. B., U.S. Patent 1,231,458 (to Stauffer Chemical Co.), 1971. 16. Forty, N., Godfrey, D., and Peeler, R., U.S. Patent 3,583,920 (to Chevron Research Co.), 1971. 17. Burrous, M. and Furby, N., U.S. Patent 3,649,721 (to Chevron Research Co.), 1971. 18. Farmer, E. P., Jr., Fire resistant lubricants in gas turbines, Lubr. Eng., 27(3), 83, 1971. 19. Gardner, L. and G. Moon, Evaluation of fire resistant lubricants for industrial gas turbines, Mech. Eng. Rep. MP56 (NRC No. 12433), National Research Council of Canada, November 1971. 20. Staley, C. and McGuigan, B., The European use of phosphate esters in steam and gas turbines, Lubr. Eng., 33(10), 527, 1977. 21. Wolfe, G. F. and Whitehead, A., Experience with phosphate ester fluids as industrial steam turbine lubricants, Lubr. Eng., 34(8), 13, 1978. 22. Wright, R. M., U.S. Patent 4,171,272 (to FMC Corp.), 1979. 23. Doanchis, H., U.S. Patent 4,169,818 (to FMC Corp.), 1979. 24. Evaluation of fire retardant fluids for turbine bearing lubricants, Final Rep. NP-6542, Project 2969–2, Electric Power Research Institute, Palo Alto, CA, Sept. 1989. 25. Phillips, W. D. and Vilanskaya, G. D., Recent operating experience in Europe and the Soviet Union with fire resistant turbine lubricants, in Proc. Am. Power Conference, Chicago, EL., 1990. 26. Wildersholm, M., Denmark Patent 3,829,610 (to UK Mineralol. Wenze.), 1990. 27. Allsop, B. et al, European Patent 297,046 (to Ciba Geigy Corp.), 1989. 28. Mullin, G., U.S. Patent 4,879,052 (to Akzo America Inc.) 1989; U.S. Patent 4,780,229 (to Akzo America Inc.), 1988. 29. Klaus, E. E., Jeng, G. S., and Duda, J. L., Study of tricresyl phosphate as a vapor delivered lubricant, Lub. Eng., 45, 717, 1989. 30. Gunsel, S., Klaus, E. E., and Bruce, R. W., Friction characteristics of vapor deposited lubricant films, SAE Tech. Pap., No. 890148, SP 785, Worldwide Progress on Adiabatic Engines International Congress, Detroit, MI, Feb. 27 to March 3, 1989. 31. Klaus, E. E., Duda, J. L., Jeng, G. S., Hakim, N. S., Groeneweg, M. A., and Belnaves, M. A., Vapor phase tribology for advanced diesel engines, U.S. Department of Energy Proceedings—Coatings for Advanced Heat Engines Workshop, Castine, ME, 1987. 32. Placek, D. G. and Freiheit, T., Progress in vapor phase lubrication technology, in Proc. ASME Internal Combustion Engine Symp., Houston, TX, Jan. 1993. 33. Miles, P., U.S. Patent 4,919,833 (to Ciba Geigy Corp.), 1990. 34. Huebner, J., Denmark Patent 3,010,669 (to Mobil Oil AG.), 1981. 35. Jayne, G. J. J. et al., German Patent 2,703,110, 1976. 36. Fischer, R. et al., German Patents 279,028 and 279,027 (to VEB Hydrierw Zeitz), 1990. 37. Tsuboi, A. and Oyoshi, H., Japanese Patent 6,268,894 (to Mitsui Petrochemical Co.), 1987. 38. Dannels, B. and Shepard, A., U.S. Patent 3,418,348 (to Hooker Chemical Co.), 1968. 39. Cottington, R. L. and Ravner, H., U.S. Gov’t. Res. & Dev. Report 1968, Naval Research Laboratories, Washington, DC, 68(14), 81. 40. Beck, T. R., Wear by generation of electrokinetic streaming currents, ASLE Trans., 26, 144, 1982. 41. MacKinnon, H., U.S. Patent 4,324,674 (to Chevron Research Co.), 1982; U.S. Patent 4,302,346 (to Chevron Research Co.), 1981. 42. Marolewski, T. A. and Jaffe, F., U.S. Patent 4,252,662 (to Stauffer Chemical Co.), 1981. 43. Flaring, J., U.S. Patent 3,629,114 (to Monsanto Co.), 1972. 44. Wiggins, B. J., System conversions for fire resistant hydraulic fluids, Lubr. Eng., 43(6), 467, 1987. 45. Phillips, W. D., The conditioning of phosphate ester fluids in turbine applications, Lubr. Eng., 39 (12), 766, 1983.
Copyright © 1994 CRC Press, LLC
286
CRC Handbook of Lubrication and Tribology
46. Anzenberger, J. F., Evaluation of phosphate ester fluids to determine stability and suitability for continued use in gas turbines, Lubr. Eng., 43, 528, 1987. 47. Johnson, M. K., Organo phosphates and delayed neuropathy—is NTE alive and well?, Tox. Appl. Pharm., 102, 385, 1990. 48. Abou-Donia, M. B. and Lapadulla, D. M., Mechanisms of organophosphorus ester induced delayed neurotoxicity: type I and II, Annu. Rev. Pharm. Tox., 30, 405, 1990.
Copyright © 1994 CRC Press, LLC
PERFLUOROALKYLPOLYETHERS Thomas W. Del Pesco
INTRODUCTION
Perfluoroalkylpolyether (PFPE) fluids are composed entirely of carbon, fluorine, and oxygen. They are colorless, odorless, completely inert to most chemical agents including oxygen, compatible with most other materials, and liquid over a wide temperature range. Gumprecht1 first disclosed the use of the perfluoroalkylpolyethers (PFPE) as lubricants in an ASLE/ASME lubrication conference in the fall of 1965. Since then, the following four distinct types of PFPE oils have become commercially available. Although all PFPE types exhibit similar physical and chemical properties, there are small and sometimes significant differences.
Both PFPE-1 and PFPE-2 are nonlinear molecules because the polymer chains contain pendant trifluoromethyl groups, (-CF3). PFPE-4 and PFPE-3 contain no pendant groups and are linear. The linear PFPE structures show less change of viscosity with temperature and pressure when compared to nonlinear PFPE. Pendant trifluoromethyl groups immediately adjacent to the ether, (-0-), linkage provide some shielding to protect that linkage from acid-catalyzed cleavage. PFPE-1 has a fully shielded polymer chain because every ether linkage is adjacent to a carbon with a pendant trifluoromethyl group. PFPE-2 has a partially shielded polymer chain because less than half of all ether linkages are protected with a pendant trifluoromethyl group. PFPE-4 and PFPE-3 have nonshielded polymer chains and are thus not protected from acid-catalyzed cleavage. PFPE-1 and PFPE-4 contain one type of ether group and are homopolymers, while PFPE2 and PFPE-3 contain two types of ether groups. PFPE-2 and PFPE-3 both contain the difluoroformyl (-CF20-) linkage as well as the perfluoropropyl or perfluoroethyl group, respectively. Even though PFPE-2 Mid PFPE-3 appear to be copolymers, they are prepared using only one monomer.
PREPARATION OF PFPE TYPES
Perfluoroalkylpolyethers, PFPE-1, are prepared by anionic polymerization of hexafluoropropylene epoxide (HFPO) at low temperatures. The preparation of HFPO has been described.2,3 HFPO can be polymerized in solvents such as aliphatic hydrocarbon polyethers or nitriles using cesium fluoride as the source of fluoride ions. The reaction temperature is -24°C, the boiling point of HFPO. The acid fluoride end group of this polymer is much too reactive for use in lubricants. The polymer is stabilized4 by reaction with elemental fluorine. Polymer with molecular weight range of 435 to 13,500 is then fractionated into grades by vacuum distillation. PTFE-2 is prepared by the photochemical catalyzed polymerization of hexafluoropropylene in the presence of oxygen at low temperatures.5 Treatment with UV light or elemental fluorine or both causes decomposition of the peroxide linkages, giving a polymer containing two 0-8493-3903-0/94/$0.00 + $0.50 © 1994 by CRC Press, Inc.
Copyright © 1994 CRC Press, LLC
287
288
CRC Handbook of Lubrication and Tribology
types of perfluoroalkyl linkages in the polymer chain, [-OCF2-] and [-OCF(CF3)CF2-]. The molecular weight range of this polymer is 1000 to 10,000. PFPE-3 is produced when tetrafluoroethylene is used instead of hexafluoropropylene,6and the crude polymer is treated with elemental fluorine to destroy the peroxide sites. The PFPE3 crude polymer has a higher molecular weight distribution than PFPE-2, ranging from 8000 to 70,000. During the stabilization process, however, the molecular weight decreases because chains are cleaved into smaller pieces. The reaction mechanism and kinetic equation for the photooxidation of fluoroolefins to PFPE-2 and PFPE-3 is described by Sianesi et al.7 PFPE-4 is obtained by Lewis acid-catalyzed ring-opening polymerization reaction of 2,2,3,3-tetrafluorooxetane.8 The hydrogen-carbon bonds are subsequently converted to fluorine-carbon bonds by direct fluorination with elemental fluorine. Ultraviolet light catalyzes the direct fluorination. The crude polymers are usually purified by contact with absorbing agents to remove polar materials and distilled under reduced pressure. The reaction mass is fractionated by distillation into specific molecular weight and viscosity ranges. These ranges usually correspond to different grades or product types available commercially.
PROPERTIES
PFPE oils are colorless and odorless fluids. Physical properties such as those given in Table 1 vary with molecular weight. Chemical properties and stability usually depend more on chemical structure than on molecular weight.
Thermal and Oxidative Stability Table 2 summarizes the temperature stability of PFPE by type. PFPE oils are very stable in pure oxygen. Tests with PFPE-1 (Table 3), under a variety of conditions demonstrate the inertness of the PFPE oil to reaction with oxygen.
Compatibility of PFPE Fluids with Metals and Metal Compounds A summary of metals compatible with PFPE oils at various temperatures is given in Table 4. In the presence of oxygen, oxidative-corrosion may occur with some metals. PFPE-1 oils are inert to most metals at temperatures up to approximately 288°C in an oxygen atmosphere. Compatibility results of PFPE-1 oils with a number of metals and alloys, using the “micro oxidation-corrosion test” developed by the Air Force Materials Laboratory, are shown in Figure 1. In general, nickel and cobalt alloys exhibit the greatest resistance to oxidative corrosion and are suitable for use with PFPE oils up to at least 370°C. Ordinary steels are not suitable above 288°C. Specific stainless steels are satisfactory at 316°C. Certain alloys cause catalytic depolymerization of PFPE oils at high temperatures. Titanium alloys which contain aluminum can decompose PFPE oils above 136°C. Aluminum 2024 appears to do the same thing at 370°C. These problems are greatly minimized in the absence of oxygen, indicating that the reactions involved are between the oil and oxide coating on the metals. Additives which markedly decrease the reaction rate of PFPE oils with many metals at high temperatures are under development.9 Some offset the catalytic depolymerization of the oils caused by titanium alloys.10–14,19 Metal halides, such as AlCl3, catalyze the decomposition of all types of PFPE oils. PFPE2 and PFPE-3, which contain the difluoroformyl (-CF2O-) group, decompose more rapidly than PFPE-1 or PFPE-4, which do not contain that group.
Copyright © 1994 CRC Press, LLC
Volume III 289
Copyright © 1994 CRC Press, LLC
290
CRC Handbook of Lubrication and Tribology
Metal oxides can also lower the decomposition temperature of PFPE-2. Sianesi et al.15 showed that cuprous oxide had a small effect on rate of decomposition at 360°C, whereas aluminum oxide had 100 times that rate, as summarized in Table 5.
Hydrolytic and Chemical Stability Hydrolytic stability of PFPE oils is exceptional. Long-term contact with steam or boiling water produces no adverse effects on the PFPE fluids. All PFPE oils are essentially inert to most chemicals. No reaction is observed with boiling sulfuric acid, fluorine gas at 200°C, molten sodium hydroxide, chlorine trifluoride at 10 to 50°C, uranium hexafluoride gas at 50°C, or any of the following materials at room temperature: JP-4 turbine fuel, unsymmetrical dimethyl hydrazine, hydrazine, diethylenetriamine, ethyl alcohol, aniline, 90% hydrogen peroxide, inhibited red fuming nitric acid and nitrogen tetroxide. PFPE-1 oils are slightly soluble in hydrazine and have moderate (25 to 30%) solubility in nitrogen tetroxide. PFPE-1 does not react with nitrogen tetroxide, nor inhibited red fuming nitric acid in impact tests.
Copyright © 1994 CRC Press, LLC
Volume III
291
PFPE-2 is reported to react with liquid (100%) and gaseous ammonia, alkaline metals, and finely divided powder of light metals such as aluminum, magnesium, and their alloys.16 The stability of PFPE-2 towards reactive chemicals is shown in Table 6. Solubility PFPE oils are not soluble in common solvents, acids, and bases, but some solvents will dissolve in PFPE oils. Solubility data for PFPE-2 are shown in Table 7. PFPE oils are completely miscible in highly fluorinated solvents such as trichlorotrifluoroethane, hexafluorobenzene, perfluorooctane, and hexafluoropropylene dimer. Table 8 summarizes gas solubility data.
Compatibility with Elastomers and Plastics Elastomer compatibility with PFPE-1 oil is summarized in Table 9.17 Most elastomers are affected only slightly by contact with the oil at 93°C. The inherent stability of the elastomers themselves and not compatibility with PFPE limits their use at higher temperatures. No significant changes in dimension, hardness, or color were noted when many elastomers were immersed in PFPE oil at 70°C for 6 months.18 Elastomers used were: neoprene, butyl, buna N and natural rubber, polyfluorosiloxane, polychlorotrifluoroethylene, and vinylidene fluoridehexafluoropropylene copolymers. PE oils have no significant effect on plastics. The following showed no effect when tereph-
Copyright © 1994 CRC Press, LLC
292
CRC Handbook of Lubrication and Tribology
FIGURE 1. Corrosion of metals by PEPE at elevated temperatures (20-ml sample in Inconel test tube, 20 1/h dry flow, 24 and 48 h test duration; no reflux condenser used).
Copyright © 1994 CRC Press, LLC
Volume III
293
thalate, polybutylene terephthalate, polystyrene, polyethylene low and high density, acrylonitrile-styrene copolymer, polymethylmethacrylate, acrylonitrile-butadiene-styrene polymer, polyamide 66, polyvinylchloride, and polycarbonate.
Flammability PFPE-1 oils are not flammable under any conditions likely to be encountered. They show no autogenous ignition, flash, or fire points up to 649°C (1200°F) in standard ASTM tests. A sample of PFPE-1 did not flash or burn when contacting a manifold at temperatures in excess of 649°C (1200°F). In a high pressure spray ignition test (MIL-F-7100), PFPE-1 oil did not flash or fire at up to 2 feet from the spray orifice.
Copyright © 1994 CRC Press, LLC
294
CRC Handbook of Lubrication and Tribology
Radiation Resistance PFPE-1 oils are stable to radiation when compared with many materials used as lubricants or power fluids. In general, irradiation of PFPE-1 oils only causes minor changes in its physical properties. No insoluble solids or sludge are formed and the viscosity decreased 21% when a
Copyright © 1994 CRC Press, LLC
Volume III
295
sample of PFPE-l was exposed to 108 rads of electron bombardment at ambient temperature in air. Mori and Morales19 studied the effects of X-ray irradiation on the degradation of PFPE4, PFPE-3, and PFPE-1. They concluded that PFPE-4 cross-linked more easily than the other three fluids.
Shear Stability PFPE-1 oil did not break down when subjected to high rates of shear. Exposure at 10 kHz in a sonic shear tester at room temperature for 1 h resulted in viscosity changes of less than 0.5%.
Physical Properties Physical properties that show little change with molecular weight are summarized in Table 10.
Thermal Properties Thermal conductivity of PFPE-l varies slightly over a wide temperature range (Table 9). These values are a little lower than in many hydrocarbon lubricants with similar viscosity. The specific heat of PFPE-1 with number average molecular weight of 6000 amu is a linear function of temperature varying from 0.23 to 0.24 cal/(g-C at 38°C to 0.29 to 0.30 cal/ (g.C) at 204°C. Thermal coefficient of expansion for two PFPE-l oils is given in Figure 2. The thermal coefficient of expansion and its change with temperature decrease with increasing molecular weight. Density The densities of PFPE fluids are nearly twice that of hydrocarbon lubricants. Density increases slightly with increasing molecular weight and decreases linearly with temperature.
Electrical Properties Electrical properties (given in Table 11) are affected by the presence of even trace amounts of moisture present in the PFPE oil.
Copyright © 1994 CRC Press, LLC
296
CRC Handbook of Lubrication and Tribology
FIGURE 2. Thermal coefficient of expansion for PFPE-1 oils.
Vapor Pressure and Volatility The vapor pressure and volatility of PFPE oils at a specific temperature vary inversely with number average molecular weight and, thus, the higher viscosity oils generally have lower volatility losses. Volatility, as measured by ASTM D-972, is shown in Table 12. This test is useful in determining the light-end content of the PFPE oil. As the lower molecular weight polymers evaporate, the viscosity of the oil will increase.
Viscosity The viscosity of PFPE oils and their change with temperature for each type of PFPE are shown in Figure 3. The absolute or centipoise (mPa-s) viscosity will be 1.9 times higher than the kinematic viscosity because the density of PFPE oils is about 1.9 g/ml. The temperature dependence of viscosity is similar to that of high grade petroleum oils, with the exception of PFPE-3, which is more favorable than petroleum oils.
Copyright © 1994 CRC Press, LLC
297
Volume III
FIGURE 3. Viscosity-temperature properties: comparison of different PFPE types.
Compressibility PFPE oils are considerably more compressible with lower bulk modulus man are conventional petroleum hydraulic fluids. Adiabatic tangent bulk modulus data (obtained by sonic methods) for PFPE-1 are given in Figure 4 for pressures up to 34.5 MPa (5000 psig) and temperatures to 204°C (400°F). Figure 5 compares the compressibility of PFPE-1 fluorinated oil with that of atypical hydrocarbon-based hydraulic fluid. At 38°C (100°F) and an applied pressure of 34.5 MPa (5000 psi), volume of the hydrocarbon oil is reduced by about 2% while PFPE-1 is compressed almost 3½%.
Lubrication PFPE oils are excellent lubricants under normal, severe, and starved operating conditions; under heavy loads; at high speeds; and at elevated temperatures. In general, when failure does occur, it occurs over a period of time, giving the operator warning of a failure. Test data in the following lubricity tests were obtained on PFPE-1 oil which contained no additives. Four-Ball Wear and Extreme Pressure Tests Four-ball wear tests were conducted on PFPE-1 at 75°C and 316°C and at loads of 1 to 40 kg, using balls made of different alloys. Typical results at 620 rpm and 1280 rpm are given in Figure 6. Table 13 shows the load-carrying ability of PFPE oils in four-ball EP (extreme pressure) tests and how it increases with viscosity.
Copyright © 1994 CRC Press, LLC
298
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
FIGURE 4. Adiabatic tangent bulk modules of PFPE-1 fluorinated oil.
299
Volume III
FIGURE 5. Compressibility of hydrocarbon oil and PFPE-1 at 38°C (100°F).
Falex Extreme Pressure Tests Extreme pressure tests with the Falex pin and V-block tester (Table 14), using M-10 or 52100 steels with PFPE oils, exceeded the limits of the test equipment. Using the standard shaft and V-block metals, the load at failure for PFPE oil far exceeded that of the other test oils.
Bearing Fatigue Tests Fatigue test results in a rolling contact bearing rig in Table 15 showed PFPE oil to provide excellent fatigue life at room temperature, 218°C, 260°C, and 316°C. The test pieces were made from M-50 steel; the stress was 4826 MPa (700,000 psi) maximum Hertz and the speed was 25,000 stress cycles per minute. Oil was fed at the rate of 20 drops per minute.
PFPE GREASES
A large number of solids are suitable thickening agents for the formation of greases from the PFPE oils, many the same as those used for high temperature hydrocarbon greases. Thickening agents commonly used are: finely divided silica, “Attapulgus clay”, ammeline, boron nitride, copper phthalocyanine, metal-free phthalocyanine, montmorillonite, the PTFE family of polymers, and zinc oxide. Some thickeners or fillers can impart thermal conductivity to the grease.20 Some of the PFPE greases thickened with fluorinated ethylene-propylene copolymer or PTFE contain antioxidants21 such as tris(fluoroalkoxyphenyl)phosphine or corrosion inhibitors such as perfluoroalkyl- and perfluoroalkyl ether-substituted benzoxazoles, benzothiazoles, bisbenzoxazoles, bis-benzothiazoles,13 and benzimidazole derivatives.22 PFPE greases can also be prepared by polymerizing tetrafluoroethylene in the PFPE itself.23 Ball bearing performance tests on the PTFE-thickened PFPE-1 greases using FTMS-791, method 333 show that PTFE-thickened PFPE greases give excellent high temperature performance. Results in Table 16 are reported in hours to failure as evidenced by excessive running torque with a large increase in spindle input power, excessive bearing noise or screech, or increase of more than 20°F in bearing temperature.
Copyright © 1994 CRC Press, LLC
300
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
FIGURE 6, Wear characteristics of PITFE-1.
301
Volume III
APPLICATIONS
The chemical and thermal oxidative resistance of the PFPE lubricants (oils and greases) has led to their widespread use. Table 17 gives many applications by type and the special properties of the PFPE oil which make the oil or grease especially suited for the application.
ACKNOWLEDGMENTS
The author thanks Robert Kelly, Jimmie Patton, John Graham, Gerry Madden, Greg Bell, and Nandan Rao for many helpful conversations and contributions to the contents and for reading and proof-reading this chapter.
Copyright © 1994 CRC Press, LLC
302
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
Volume III 1. 2. 3. 4.
5. 6. 7.
8. 9.
10. 11.
12.
13.
14.
15. 16. 17.
18.
19.
20.
21.
22.
23.
REFERENCES
303
Gumprecht, W. H., PR-143—A new class of high-temperature fluids, ASLE Trans., 9, 24, 1966. Carlson, D. P. and Milian, A. S., 4th Int. Symp. Fluorine Chemistry, Estes Park, Colorado, July, 1967. British Patent 904, 877, Sept. 5, 1962. Gumprecht, W. H., The preparation and thermal behavior of hexafluoropropylene epoxide polymers, in 4th Int. Symp. Fluorine Chemistry, Estes Park, Colorado, July, 1967. U.S. Patent 3442942, May 5, 1969. U.S. Patent 3715378, Feb 6, 1975. Sianesi, D., Pasetti, A., Fontanelli, R., Bernardi, G. C, and Caporiccio, G., La chimica e l’industria. Wear, 55(2), 208, 1973; U.S. Patent 4,451,646, May 29, 1986. European Patent Application 0148482, Dec. 20, 1984. Dolle, R. E. and Harasacky, F. J., New high temperature additive systems for PR*-143 Fluids, U.S. Air Force Materials Laboratory Tech. Rep. AFML-TR-65–349, Jan. 1966. Corti, C. and Savelli, P., Perfluoropolyemer lubricants, Proc. Conf. Synth. Lubr., Zakar, A., Ed., Hungarian Hydrocarbon Institute, Szazhalombatta, Hungary, 1989, 128. Strepparola, E., Gavezotti, P., and Corti, C, Antirust additives for lubricants or greases based on perfluoropolyethers, Eur. Pat. Appl., EP 337425 Al, Oct. 18, 1989; CA111 (26):236482v. Jones, W. R., Jr., Paciorek, K. J. L., Ito, T. I., and Kratzer, R. H., Thermal oxidative degradation reactions of linear perfluoroalkylethers, Ind. Eng. Chem. Prod. Res. Dev., 22(2), 166, 1983. Christian, J. B. and Tamborski, C., Benzoxazole and benzothiazole antirust greases, Lubr. Eng., 36(11), 639, 1980. Snyder, C. E., Tamborski, C., Gopal, H., and Svisco, C. A., Synthesis and development of improved high-temperature additives for polyperfluoroalkylether lubricants and hydraulic fluid, Lubr. Eng., 35(8), 451, 1979. Sianesi, D., Zamboni, V., Fontamelli, R., and Binaghi, M., Wear, 18, 85, 1971. PFPE manufacturer bulletin MA-816E, Montedison USA, New York, NY. Dolle, R. E. et al., Chemical, physical and engineering performance characteristics of a new family of perfluorinated fluids, U.S. Air Force Materials Laboratory Tech. Rep. AFML-TR-65–358, Sept. 1965. Messina, J., Perfluorinated lubricants for liquid fueled rocket motor systems, ASLE Prepr., No. 67 AM8A-4, May 1 to 4, 1967. Mori, S. and Morales, W., Degradation and crosslinking of perfluoroalkyl polyethers in ultra vacuum, NASA Tech. Pap., No. 2910, 1989. Mizushima, S., Nakahara, H., and Yamada, H., Perfluoropolyether compound compositions with excellent thermal conductivity, JP 63251455, Oct. 18, 1988. Christian, J. B., Oxidation stable polyfluoroalkyl ether grease compositions, U.S. Pat. Appl.; NTIS Order No. PAT-APPL-6–418 106., US 418106 A 15 Apr. 1983; CA99(10):73640n. Christian, J. B., Grease compositions, U.S. Pat. Appl., NTIS Order No. PAT-APPL-225 546. US 225546 A0 31 Jul 1981, CA96(4):22263x. Tohzuka, T., Kataoka, Y., Ishikawa, S., and Fujiwara, K., Fluorine-containing grease and its preparation, EP 341613, Al 15 Nov. 1989, CA112(12):101951x.
Copyright © 1994 CRC Press, LLC
SILICONES E. D. Brown
INTRODUCTION By the middle of the 19th century, a complete industry was based on inorganic silicates, and scientists of the time attempted to develop a new chemical system based on silicon, duplicating that based on carbon. The most promising work was done by Stock, who thought he had succeeded in making a compound of silicon with a doubly bonded oxygen analogous to an organic ketone. Given the name “silicone”, this was soon seen to be simply an Si-O-Si linkage. The name, however, was retained as a popular, convenient shorthand for a complex series of products. With the availability of silicon chemicals, silicon tetrachloride in particular, and reactive organometallic compounds, Friedel, Crafts, and Landenberg all were successful, late in the 19th century, in creating the first organo-silicon monomers. In 1904, F. S. Kipping,2 using these newer silanes, was able to synthesize di- and tri-silanols (materials which can be condensed to form siloxanes), the starting point of silicone chemistry. It was not until the mid 1940s that further advances were made by Hyde at Coming Glass and a team at General Electric which included Rochow,3 Patnode, and Sauer. Although researchers were looking for a high temperature resin for electrical insulation, the first product developed was a low viscosity oil which found almost immediate commercial acceptance. Before serious commercial production could proceed, an efficient, inexpensive process for making the starting material was needed. This was supplied by Rochow through the invention of the “direct process”.4 Today there are seven major producers and a number of smaller specialty companies who compound products into greases, emulsions, and chemicals.
THE NATURE OF SILICONES
To produce silicones, elemental silicon is reacted under catalysis with an organic chloride to form a mixture of organochlorosilanes, which are separated by distillation and reblended in the proper portion to obtain the desired product.5,6 This mixture is hydrolyzed to form silanols, which condense quickly to form the desired siloxane. There are two major ways in which building blocks are formed. In the first, a series of chlorosilanes are made, either by the direct process as noted above or through conventional chemistry, primarily Grignard reactions. In the second, the standard chlorosilanes are modified by appropriate reactions to impart specific properties: lubricity, solvent resistance, surface reactivity, etc.
The Chlorosilanes I (CH3)3SiCl—Trimethylchlorosilane Trimethylchlorosilane is the “mono” chain stopper which limits the length of the polymer and, in this way, controls the viscosity and to a lesser extent the shear stability and compressibility of the finished product.
II (CH3)2SiCl2—Dimethyldichlorosilane By far the most common of the silanes, dimethyldichlorosilane, “di”, is the chain extender. It can hydrolyze with “mono” to form linear chains of any practical length or with itself to form cyclic rings. Copyright © 1994 CRC Press, LLC
305
CRC Handbook of Lubricants and Tribology
306
III (CH3)SiCl3—Methyltrichlorosilane Used primarily in resins, methyltrichlorosilane, “tri”, is primarily a cross linker and, as such, of little use in lubrication. In small quantities and properly end stopped, it can contribute to low temperature properties in silicone fluids. IV (C6H5)2SiCl2—Diphenyldichlorosilane
V (C6H5)CH3SiCl2—Methylphenyldichlorosilane Both of these silanes are used in the same way, in small quantities to contribute low temperature properties and in high amounts to improve oxidative stability.
VI (C6H5)SiCl3—Phenyltrichlorosilane Used primarily in resins, phenyltrichlorosilane serves as starting point for production of a lubricating unit when used in fluids.
VII (CH3)HSiCl2—Methylhydrogendichlorosilane Methylhydrogendichlorosilane is one of the more versatile building blocks. As well as forming chains and cyclics which are useful themselves, it is the starting point for many other materials. The prime feature is the ability of the hydrogen to add across the double bond of organic molecules and even across the double bond of such silanes and siloxanes as those containing vinyl groups. Some of these modified silanes and siloxanes are important in lubrication.
Modified Silanes and Siloxanes Although modified silanes and siloxanes represent a small percentage of the silicone market, they include a number of important products designed for specific purposes. The most common of these are the reaction products of methylhydrogen silanes and siloxanes with organic olefins,7,8 particularly allyl units with functional groups. With a combination of routine steps and, in some cases, complex chemistry, a series of methylpropylsiloxanes with such functional groups as chloro, amino, mercapto, cyano, and fluoro are made. By far the most important in lubrication are the fluoropropyl siloxanes. In other cases methylhydrogen polysiloxanes are reacted with straight chain or branched olefins, either alkyl or aryl, to make methyl alkyl and methyl styryl (or substituted styryl) fluids and their mixtures. All of these have a use in lubrication. Other units can also be modified. The phenyltrichlorosilane can add four chlorine atoms to the phenyl unit, and the methylphenyldichlorosilane can add two chlorines to the phenyl unit. Both of these are used as hydraulic fluids and as general purpose lubricants. Finally, it is possible by using proper techniques to make dimethyl polysiloxane-polyglycol block copolymers which have also found lubrication applications.
MANUFACTURE OF SILOXANES
Rochow5 and Noll6 both describe the process thoroughly, and shorter descriptions can be found in Meals and Lewis,9 Hardman and Torkelson,10 and McGregor.11 Essentially the chlorosilanes are blended and hydrolyzed in excess water, the siloxanes separated from the water and neutralized. Although every possible combination is formed during the hydrolysis, the majority of the molecules will be short chain linears and cyclics. Fortunately, it is possible to catalytically break and remake the silicon oxygen bonds until an equilibrium mixture is formed which is represented by a standard probability curve. The volatile light ends are removed, and the desired product results. Copyright © 1994 CRC Press, LLC
Volume III
PROPERTIES OF THE SILICONES
307
By far the most common of the silicones are the various grades of dimethylpolysiloxane, but unfortunately it is these which have given the family the reputation of being extremely poor lubricants. This reputation is only partly deserved since many surfaces can be lubricated well by these materials and several silicones are excellent lubricants for all surfaces. Some will lubricate surfaces that no other material can protect. There is a second and important consideration. Dimethyl silicones must be used in innumerable cases because of their unique properties: viscosity-temperature characteristics, low and high temperature stability, shear resistance, compressibility, low toxicity, and many others. In these it is important that ways be found to use them without damage to the systems since they have no boundary or extreme pressure characteristics and only limited hydrodynamic ability to enable steel on steel lubrication. In describing physical properties of the silicones, the major emphasis will be on the dimethyls. When characteristics of special lubricating silicones deviate appreciably, that will be noted.
SURFACE PROPERTIES
The same features that make silicones valuable in so many fields are those that are largely responsible for the poor lubricating ability of the dimethyl siloxanes, and it is the function of the lubricating silicones to overcome these obstacles. The most obvious of these features is the ability of silicone fluids to spread quickly to oriented monomolecular films, a function of the low Van de Waals forces as shown by surface tension measurements (q.v.). This allows individual textile fibers to be coated to give water repellancy while retaining their porous nature and allows thin films to act as release agents for molding operations. In lubrication, thin films deposited on glass containers reduce damage in handling, small amounts incorporated in plastic composites bleed to the surface to make a self-lubricating plastic product, and the spread film provides a superior lubricant for rubber surfaces. For the most part, however, the surface properties do not favor lubricity. Early work by Fox et al.12 and more recently by Steinbach and Sucker13 and Brown14 give a reasonably clear picture of the structure of these molecules, leading to an explanation of their behavior. With thin films of dimethylpolysiloxanes spread on water, using a film balance of the same type as used to show the orientation and thickness of such boundary lubricants as stearic acid, the thickness of the monomolecular film was found to be 5.9 Å, commensurate with a molecule in which the Si-O-Si unit lies on the surface. This thinness, by itself, would not rule the class out as an effective lubricant; but when the film is compressed it passes through a fairly stable stage at a little over 7 Å and finally reaches 12.5 Å. This agrees with the measurements of a helix of 6 units per turn. Further compression has no effect since the helices simply tumble over one another with no tendency for entanglement. There is no compression force which will allow a film of fluid to support a lens of the same fluid. (See Figures 1 and 2.) Two obvious ways of rectifying the inability to form a thick boundary layer are used with the lubricating silicones. The chlorophenyl silicones modify the surface, and the methyl alkyl silicones increase the film thickness. Chlorophenyl Silicones When a small amount of tetrachlorophenylsesquisiloxane is incorporated in a dimethyl silicone chain, the fluid is capable of lubricating tool steels as well as any of the conventional
Copyright © 1994 CRC Press, LLC
308
CRC Handbook of Lubricants and Tribology
FIGURE 1. Force/area diagram, dimethyl siloxane and stearic acid. (From Steinbach, H. and Sucker, C, Ber. Busenges Phys. Chem., 67, 407, 1963.)
FIGURE 2. Siloxane surface orientation.
lubricants. The mechanism is a variation of “extreme pressure” lubrication in which chlorine is the active ingredient. Careful analysis has shown that during service a film of ferrous chloride dihydrate is formed.
Methyl Alkyl Silicones When one of the methyl groups is replaced by a longer chain alkyl group, the thickness of an expanded film when spread on water in a monomolecular layer increases dramatically. Under compression, the higher alkyl film heights exceed the best of the boundary lubricants and do not collapse. In addition, as shown by the surface tension, lateral adhesion is increased and considerable entanglement occurs. With this arrangement it makes little difference what metal or nonmetal pairs are in sliding contact. Steel, bronze, aluminum, glass, plastic, and monel are all lubricated effectively.
PHYSICAL PROPERTIES
Surface Tension As expected from their structure, the surface tension of the polysiloxanes is very low. For most of the fluids, this value will not vary greatly from 21 dynes/cm with practically no change with increasing molecular weight. Bulky side groups do have an effect. By replacing one methyl with a phenyl unit, surface tension is increased to 25 dynes/cm, and with a long chain alkyl group this value can nearly match hydrocarbons at the lower end, 31 dynes/cm. The practical effect in lubrication of the low surface tension and, consequently, the low work of cohesion, is unique spreading ability. Silicones will spread quickly, approaching a Copyright © 1994 CRC Press, LLC
Volume III
309
FIGURE 3. Viscosity/temperature relationships for dimethyl siloxanes. Centistokes viscosities at 25°C given at right. (From Demby, D. et al., Synthetic Lubricants and High Performance Fluids, Marcel Dekker, New York, 1992, 183.)
monomolecular film if possible. While the bulkier side groups only increase the surface tension by some 25%, the spreading is slowed greatly.
Viscosity-Temperature Properties One of the most dramatic properties of the silicones is the small change in viscosity with temperature. This change is so small that the old standard measurement, the viscosity index, has little meaning. In place of this a new standard was introduced, the viscosity-temperature coefficient, defined as viscosity at 38°C (100°F) - viscosity at 99°C (210°F)/viscosity at 38°C (100°F). In general, the coefficient for silicones is about 0.6, while that of most hydrocarbons is 0.8 or higher. Among the several explanations for this phenomenon, the most reasonable relates it to the helical structure. As the temperature increases, the spacing between molecules increases, but this action is countered by expansion of the helix. Figure 3 shows the viscositytemperature relation for a number of silicone fluids and a comparative curve for organic fluids. Note that when the molecule is modified by adding phenyl, tetrachlorophenyl, or alkyl units in Figure 4, a rather large amount of the bulkier group is required to affect the temperature dependence.
Low Temperature Properties The pour points of the silicones are extremely low. In general, the dimethyl silicones will flow at temperatures within a few degrees of - 51°C (-60°F), but this can be lowered substantially by slight modification of the molecule. Warrick et al.17 published the effect of adding phenyl units to a dimethyl polysiloxane polymer (see Figure 5). Addition of any unit which will break the symmetry of the molecule will have the same effect at about the same mole percent. The pour point drops fairly sharply as the percentage of the added material increases until this amount reaches about 5 to 7%, and then starts to rise slowly, eventually becoming considerably higher than the base fluid. This occurs with phenyl, tetrachlorophenyl, and branched chain methyl groups among the commercially available silicone fluids. Copyright © 1994 CRC Press, LLC
310
CRC Handbook of Lubricants and Tribology
FIGURE 4. Viscosity/temperature relationships for a number of siloxanes. (From Barnes, J. E. and Wright, J. H., 55th NLGI meeting, 1988.)
FIGURE 5. Pour points of methylphenyl siloxanes as a function of mol% phenyl. (From Warrick, E. L., Hunter, M. J., and Barry, A. J., Ind. Eng. Chem., 44, 2196, 1952.)
Fortunately, this optimum percentage for pour point is precisely the percentage of tetrachlorophenyl units required to make one of the lubricating silicones. Unfortunately, the methyl alkyls and the fluoropropyl lubricating silicones require far more of the nondimethyl units to be effective. Typical pour points for various silicone fluids are given in Table 1.
Shear Stability Another unique characteristic of the silicones is their resistance to shear. Fluids with viscosities under 1000 centistokes show almost no reduction in viscosity, even at very high rates of shear. As the nominal viscosity increases beyond 1000 cSt, a temporary reduction in viscosity occurs as the molecules line up in the direction of flow (see Figure 6). When the shear force is removed, the fluid returns to its original viscosity. Zisman and co-workers subjected a low viscosity (70 cSt) fluid to 105,000 cycles over a 500-h period at 105 kg/cm2 and found a change of viscosity of less than 2%, while a polymerthickened mineral oil tested for comparative purposes lost more than 50%. Copyright © 1994 CRC Press, LLC
Volume III
311
FIGURE 6. Effect of shear on temporary viscosity of silicone fluids of 1,000 to 100,000 cSt viscosity at 25°C. (From Demby, D. et al., Synthetic Lubricants and High Performance Fluids, Marcel Dekker, New York, 1992, 183.)
Using a radial piston pump, the tetrachlorophenyl methyl polysiloxane was run for 1200 h at 4000 psig. This allowed recirculation for 44,000 cycles. The viscosity loss was less than 4%. A mineral oil (5606) run for comparison lost over 50% in leakage in less than half this time and showed a viscosity decrease of greater than 25%.22 The same shear stability characteristics are shown by all of the silicones.
Compressibility and Bulk Modulus Although silicones are more compressible than organic fluids, only the very lowest viscosity materials have the unusually high compressibility that makes them useful as liquid springs and in other devices which depend on this property. As soon as the viscosity reaches 50 cSt, there is very little change with increasing molecular weight. The bulk modulus of dimethyl fluids is about 150,000 psi at room temperature and 6000 psig, while that of the silicone most often considered for hydraulic applications (tetrachlorophenyl methyl polysiloxane) is a little less than 200,000 under the same conditions. Two considerations counter the disadvantages of high compressibility in hydraulic applications. The bulk modulus changes with temperature much less than does that of organic Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
312
FIGURE 7. Mechanisms for high temperature attack.
fluids, and silicone fluids do not solidify even at very high pressures, as is often the case with organic materials.
High Temperature Properties Thermal attack may come in the two ways shown in Figure 7 with entirely different results. The oxidative attack occurs at the silicon-carbon bond with the decomposition products being initially formaldehyde and a cross-linked silicone polymer. Under strictly thermal conditions, the silicon-oxygen bond is ruptured with the formation of low molecular weight silicones. The latter is somewhat puzzling since the silicon-oxygen bond is stronger than the silicon-carbon bond. The most logical explanation is that the breakage of the bond is catalyzed by trace impurities. Oxidative Stability Table 2 lists the oxidative threshold for a variety of silicone fluids. The accepted threshold temperature for the basic dimethyl silicones is listed as 204°C (400°F). This is reasonably true for low and medium viscosity fluids since the gel times will be from 350 to 500 h. This decreases with higher molecular weights, but in all cases they will withstand at least 190°C (375°F) for an indefinite period.19 Gel times observed in exposure in beakers in an air circulating oven are given in Table 3. Addition of phenyl groups to the molecule will show an almost straight line relation between phenyl content and oxidative threshold until at a phenyl/methyl ratio of 0.75 a value of about 520°F is reached.
Copyright © 1994 CRC Press, LLC
Volume III
313
At the other end of the scale, when long chain alkyl groups are incorporated, the best that can be hoped for, even with oxidation inhibitors, is about 350°F. Since attack on the longer carbon chains is at the second carbon from the silicone, the threshold is about the same for all added alkyl groups. Traditional oxidation inhibitors have little effect because of both their lack of solubility and the high temperatures involved, usually well above the stability level of the inhibitors. Fortunately nonconventional inhibitors such as iron soaps, particularly iron octoate and ferrocene derivatives, are effective although requiring special processing.
Thermal Stability The standard method for determining the threshold of thermal breakdown is the isoteniscope.20 With this instrument, vapor pressure is measured against temperature, usually resulting for pure compounds in a straight line with a definite break at some temperature. Since the silicones are polymers, the results are not quite as clear; but, for all silicones the break occurs within a few degrees of 315°C (600°F). The breakdown products are, in general, low molecular weight cyclic silicones, considerably more volatile than the base fluid. Meals and Lewis,9 by thermodynamic calculations, set the upper limit for the siloxane bond at 538°C (1000°F), a value never reached by the silicones. For the silicone most likely to be used in hydraulic applications with occasional surges in temperature over 315°C (600°F), the breakdown is partially compensated for by the pressure at which hydraulic fluids operate. This will, to some degree, reverse the process.
Hydrolytic Stability In general at moderate temperatures, the silicones are completely hydrolytically stable. At temperatures of 204°C (400°F) and higher, water will act as a catalyst for thermal degradation. At temperatures of 315°C (600°F) and higher, water will attack the chlorine in the chlorophenyl lubricating fluid to cause more rapid degradation.
COMPATIBILITY
With Other Fluids While none of the silicones are soluble in water, lower alcohols, glycols, fatty acids, vegetable oils, or higher hydrocarbons, they are completely miscible with common aliphatic and aromatic solvents, higher alcohols, low viscosity petroleum fractions, and the base fluids of most natural and synthetic hydraulic fluids. 1. Sufficient silicone can be dissolved in medium weight petroleum products, and in some cases dispersed in higher fractions, to serve their purpose as antifoams, V.I. improvers, and similar additive functions. Copyright © 1994 CRC Press, LLC
314
CRC Handbook of Lubricants and Tribology
2. The silicone most adapted to use as an hydraulic fluid, the tetrachlorophenyl methyl siloxane, is completely miscible with the base fluid of most other hydraulic fluids. It is definitely not miscible with the additives. When mixed with MIL-H-5606A, for example, a gummy residue is formed. More serious than this, however, is the lubricity incompatibility of this fluid with diester lubricants. The mechanism of lubrication is completely different and there is a competition for the surface. Bearings run in with MIL-L-7808 should be lubricated with the same fluid; likewise if the bearing is run in with the chlorophenylmethyl silicone, this should be the lubricant. 3. Methyl alkyl fluids and the mixed aryl, alkyl methyl fluids are an exception to these rules. They are completely compatible with most organic materials. 4. At the other end of the scale are the fluorosilicones and the trifluoropropyl dimethyl copolymers. These are designed to be resistant to other fluids and solvents and are used in areas such as chemical industry compressors and in solvent-resistant greases.
With Rubber When used as a thin film lubricant for O-rings and rubber sleeves, or when used as a release agent, all of the silicones perform as desired with no ill effect on any rubber, natural or synthetic. When rubber is in intimate contact with large amounts of the fluid, particularly at elevated temperatures, those rubbers containing an appreciable amount of plasticizer will stiffen due to the leaching of these materials. To avoid this, rubber with little or no plasticizer (neoprene, butyl, nitrile and natural) should be used. The silicones are serviceable up to the temperature limit of the rubber. For most systems using silicones, the low surface tension encourages leakage, and seals are designed to compensate for this. Occasionally, particularly when the properties of the silicones are needed for existing systems, silicones are required to provide a certain amount of seal swelling to avoid leakage. It is possible in these cases to use conventional rubber swell additives tailored to the particular rubber. All of the above discussions concern low and moderate temperature applications within the operating range of the elastomers. High temperature, high pressure systems require much different treatment. Although the operating limits of the silicones are above that which the best of the elastomers can tolerate, the 400°F limit of fluorocarbon elastomers (Viton, Kel-F, etc.) is sufficient for many uses. Beyond this, other systems will be required, i.e., springloaded Teflon, carbon seals, split rings, etc.21 With Carbon When properly designed, carbon seals can operate under extreme conditions with or without a lubricant and thus would seem ideal for high temperature silicone sealing. This is true with one proviso: for a carbon seal to operate with silicones, there must be a positive pressure across the seal resulting in a slight leakage, perhaps as little as a half drop per day. Otherwise the fluid will oxidize under the seal and damage it. Fortunately most seals fulfill this requirement.22
With Plastics With the exception of polyacetal and polyvinyl chloride, the silicones are completely compatible with common plastics. Under long-time exposure, there will be some stress cracking of polyethylene and polypropylene unless these have been previously stress-relieved. With Metals All of the suppliers will provide a long list of metals of construction suitable for use with silicones. Most common metals are suitable with the exception of those containing appreciable
Copyright © 1994 CRC Press, LLC
Volume III
315
amounts of lead which is an excellent catalyst for condensing silanols which form in small quantities in any system. Such condensation will eventually gel the fluid. The chlorophenylmethyl silicones will tend to dezincify alloys containing this metal. At the high temperatures in which this and other silicone fluids are often required to operate, the tendency for some metals to degrade (stripping of cadmium plate, sloughing of chromium plate at sharp corners, etc.) will be aggravated.
With Paint and Solder Only methyl alkyl and methyl alkyl aryl silicones are paintable and do not interfere with soldering. All others require a thorough cleaning following the manufacturer’s direction before attempting to paint or solder a treated surface.
FLASH AND FIRE POINTS
For the silicones, the results are strictly related to the volatility and to the thermal breakdown temperature. As the molecular weight and viscosity increase, the flash point also increases until the viscosity reaches 50 cSt. From this point on, the open cup flash point will be within a few degrees of 315°C (600°F), (closed cup about 302°C [575°F]) and the fire point will be within a few degrees of 371°C (700°F), values reached by few conventional lubricants.
AUTOGENOUS IGNITION TEMPERATURE
Using a quartz flask heated by blast burners, drops of fluid are introduced until a sudden flash appears. Various instrumental attachments have been used to give the test better credibility, but in all cases there is considerable error. For all silicones over 50 cSt the values range from 454 to 482°C (850 to 900°F).
RADIATION RESISTANCE
Both dimethyl and chlorophenylmethyl silicones have quite poor radiation resistance; however, the addition of phenyl units improves the resistance markedly. The moderate phenyl fluids are quite suitable for use in seismic snubbers for nuclear power plants and have been used in this application for several years.
LUBRICITY
Silicones have never been considered as general purpose lubricants, partially because of their intrinsic price and partially because of the need to be careful in the selection of both the fluid and the systems in which they are to be used. As the lubrication requirements became more severe and the equipment manufacturers more demanding, over 30 lubricant wear testers were devised. In order to prevent confusion, it is necessary to understand what is being tested and why there is sometimes little correlation between testers. For example: the Ryder gear test ranks the phenyl silicones as one of the best lubricants while most other tests rate it as one of the worst. Tables 4, 5, and 6 give representative results from three widely used testers.
Falex Tester The Falex tester consists of two steel blocks with “V” grooves which are pressed against a spinning steel pin at steadily increasing loads until the pin shears or the wear becomes so severe that the blocks touch. The test was designed primarily to evaluate “extreme pressure” additives and not necessarily to duplicate any particular service condition. Copyright © 1994 CRC Press, LLC
316
CRC Handbook of Lubricants and Tribology
Ryder Gear Test Using a “four square” configuration, this test measures the surface damage on hardened steel gears. The gear meshing action involves more than thin film quantities which are subjected to high pressure. Great increase in viscosity with pressure helps cushion the load.
Four-Ball Wear Tests Probably the most widely used, versatile, and most easily interpreted test for lubricity is the Shell four-ball wear test. In it a single half-inch ball is spun in a nest of three balls clamped immobile in a cup. As currently designed, speed, load, temperature, and ambient atmosphere can be varied at will. A side arm is connected to a transducer and friction measurements are made continuously. At the completion of the test, usually 1 hour, the bottom balls are removed and the circular wear scar is measured. Since there is little, if any, hydrodynamic component in this type of testing, both boundary and extreme pressure characteristics are measured. A wide variety of test pieces are available, and at one time or another practically every bearing metal has been evaluated.
SPECIFIC APPLICATIONS
Dimethyl Silicones The dimethyl silicones have been used for years as lubricants for rubber against a wide variety of materials: metals, glass, and plastic. Rubbers in the form of O-rings, sleeves, and manufactured parts are all lubricated well with no deleterious effect on the rubber even after long-time contact. Sewing thread, needles, and textiles in general have all been lubricated with the dimethyl silicones for a number of years. Both of these applications have led to a more recent use in the lubrication of plastics. As Steijn23 has stated, the friction and wear of plastics is of a different nature than that of metals, behaving sometimes as a rubber and sometimes as a metal. Silicones have been tried and found to be effective lubricants, either applied to the plastic surface or incorporated in the Copyright © 1994 CRC Press, LLC
Volume III
317
plastic itself. Silicones are often the lubricants of choice where nontoxic, nonstaining, longlife, and water resistant characteristics are important. The inability of dimethyl silicones to lubricate steel surfaces leaves a few problems. There are certain metal/metal contacts that can not be avoided since these fluids are essential in many industries. Pumps, actuators, sliding pistons, etc. all must operate without damage. In 1946 Zisman and co-workers examined nearly 100 metal combinations lubricated with dimethyl silicones, finding a reasonable number satisfactory for light duty work.24 Although many were impractical, several combinations including brass and bronze journals with a chrome-plated steel shaft gave adequate performance. Later tests with pumps, gear shafts, and antifriction bearings have borne this out. For over 40 years these parts have operated with a minimum of trouble. Clocks and timers for household use require lubricants which will last for many years without change of physical properties. Industrial versions of these require that, in addition to long life, the lubricant will withstand harsh environments of low and/or high temperatures. The metal combinations in most of these fit nicely with those recommended by Zisman, and the dimethyl silicones are often the lubricant of choice. Although additives (primarily chlorine compounds) will improve the lubricity of the dimethyl fluids to some extent,25 in the unusual circumstances where extra lubricity is required, the chlorophenyl methyl fluids are suggested. Whereas lubricity additives give only marginal results, other additives have a profound effect. By using rubber swell control compounds, corrosion protectors, and rust inhibitors, a complete hydraulic brake fluid system has been developed to replace such water-sensitive materials as the glycols, thereby avoiding vapor lock and associated problems.26
Methyl Phenyl Fluids Many of the same comments will apply to methyl phenyl fluids as well as the dimethyl silicones. Considering the greater cost of the phenyl-containing fluids, they will only be used
Copyright © 1994 CRC Press, LLC
318
CRC Handbook of Lubricants and Tribology
where the greater heat resistance is required. Again, it is absolutely essential that metal combinations be chosen carefully because of their serious problems in lubricating steel on steel.
Chlorophenylmethyl Silicone This fluid was designed to supply a silicone with essentially the physical properties of the dimethyl fluids and improved lubricity. The small amount of tetrachlorophenyl improved the pour point, increased the oxidative stability slightly, and supplied lubricity nearly equal to organic lubricants. As a result, it became an excellent candidate for hydraulic fluid applications where the fluid must operate at temperature extremes of - 54 to + 316°C (- 65 to 600°F). Extensive testing of aircraft hydraulic pumps and systems has been done under the most severe conditions of temperatures at both extremes. All gear pumps work well and any properly designed piston pump performs satisfactorily. Complete systems, actuators, servos, piping, valves, filters, and reservoirs have all performed satisfactorily at these temperatures. Miniature bearings lubricated with the chlorophenyl silicones will perform better at 400°F than those lubricated with the best organic fluids at 167°F. In some cases these fluids have replaced the dimethyl as clock and timer lubricants. For the more difficult applications in this Held, additives which had been used with the dimethyl fluids are unnecessary with the chlorophenyl silicones. The low vapor pressure and the stability of the fluid has led to its use in vacuum pumps where the low miscibility with organics and water allow an easier clean-up of the fluid.
Fluoropropyl Silicones There are several copolymers of trifluoropropyl and dimethyl units, all very expensive. The greater the amount of trifluoropropyl units, the better the lubricity, the higher the cost, and the poorer the heat stability. Good lubricity and load-carrying capacity make them useful in many areas. Resistance to other fluids and solvents gives them an added advantage in special applications. They are used as base fluids for greases, as lubricants in corrosive environments, as compressor fluids, and in very special cases as hydraulic fluids.
Methyl Alkyl Silicones These unique fluids, operating in a very different way than normal boundary or e.p. lubricants, allow the use of metal combinations which have been difficult to lubricate: steel on aluminum, aluminum on aluminum, steel and bronze on steel, glass on glass, and many others. In addition, these fluids are excellent release agents for aluminum the casting and serve as cutting fluids and general purpose household lubricants.
GREASES
There are at least two distinct types of filled fluids: compounds which are usually filled with silica or other inorganic fillers and “greases” which use soaps, carbon black, molybdenum disulfide, PTFE, and the like. The manufacture of greases in general and silicone greases in particular requires years of experience and is as much an art as it is a science. Both greases and compounds consist of some filler or thickening agent dispersed in a silicone oil. Every possible way of dispersing a filler in the oil has been used. These include simple stirring, cooking, milling, preparing in situ, and combinations of these. In order to obtain a more stable material, it is usually necessary to add a stabilizing agent. For the silicones this is usually a highly hydrophilic compound, alcohol, glycol, or water. This prevents the Copyright © 1994 CRC Press, LLC
Volume III
319
filler from becoming completely wet out by the oil and allows a structuring to take place.16 Since the properties of the grease or compound will depend to a large extent on the properties of the base oil, enhanced or limited by the properties of the filler, selection of base fluid is determined by the final use of the product.
Compounds Usually consisting of dimethyl silicones or, in rare cases, methyl phenyl or methyl alkyl silicones with an aerosol silica thickener and some stabilizing agent, these mixtures are used as dielectric material, protectors for insulators (see Table 7), optical compounds, lubricants for light duty antifriction bearings, and for both sliding and rolling applications with plastic or rubber journals and bearing surfaces. With metal oxides they form thermal conductive layers, and with proper additives they are used as corrosion inhibitors. The consistency varies greatly, and with sufficient shear they are likely to be thixotropic. There is little change of properties with temperature, and they are useful in approximately the same temperature range as the fluids from which they are made. As would be expected, there is no “drop point” for any of these materials.
Greases The greases are often designed for specific applications, to government specifications, or to answer critical industrial requirements. Since the requirements are usually too severe or too specialized for the dimethyl-based greases, by far most of the greases are made from phenyl-, chlorophenyl-, trifluoropropyl-, or alkyl-containing silicones. The most common fillers are organic soaps, usually lithium for temperature stability and graphite. Less common but equally important fillers are aryl ureas, organic dyes, PTFE, and molybdenum disulfide. (See Table 8.)
APPLICATIONS
Methyl Phenyl Greases The lithium soap-filled greases are used to lubricate ball and roller bearings, electric motors, clocks and timers, and any place where high temperatures may be expected. Various grades are used to meet government specifications, MIL-L-15719 and MDL-G-46886 among others. Some are highly radiation resistant and can be used in nuclear power stations. Copyright © 1994 CRC Press, LLC
320
CRC Handbook of Lubricants and Tribology
The graphite-filled greases are used on conveyor belts, ovens, roller and sleeve bearings, and as antiseize compounds.
Chlorophenyl Methyl Grease This is a general purpose grease for both sliding and rolling contact in bearings, gears, and other mechanisms where high temperatures are expected. As with the base fluid, it is a good lubricant for steel on steel, steel on brass or bronze, and on most common bearing materials except where both members are soft. Fluoropropyl Methyl Greases Primarily used where resistance to other fluids and solvents is required, these greases serve in the chemical industry and in such applications as lubricants for the bearings in automotive fan clutches.
Methyl Alkyl Greases Where high loads are anticipated and with unusual metal combinations, particularly aluminum and bimetallic contacts, these lubricants are outstanding. In addition these are among the very few silicone products that are both paintable and solderable.
CONCLUSIONS
In spite of their continuing high cost, there are many applications where nothing else will serve and an equal number where the silicones perform so much better than other products that the price is economically feasible. As an example of the latter, bearings lubricated with either silicone greases or oils have a life 3 to 20 times longer at high temperatures than their organic counterparts. In addition, some applications such as the lubrication of miniature bearings require so little lubricant that the contribution of the silicone to the final cost is negligible. As hydraulic fluids operating over a wide range of temperatures, the chlorophenyl methyl fluids have served and will serve admirably. As lubricants for traditionally difficult metal pairs such as aluminum on steel or aluminum, the methyl alkyls may be the only answer. For lightly Copyright © 1994 CRC Press, LLC
Volume III
321
loaded sliding and rolling contact with plastics where nonstaining, nontoxic, lifetime lubrication is needed, the dimethyl silicones will serve. Wherever lubricants or mechanical fluids are required to operate at low or high temperature extremes, or in other hostile environments, the silicones may be able to prevent costly failures.
REFERENCES 1. Friedel, C, Landenberg, A., and Crafts, J., Justus Liebig. Ann. Chem., 127, 31, 1863. 2. Kipping, F. J., J. Chem. Soc, Volumes 79 through 128: numerous pages (1901–1924). 3. Rochow, E. G., J. Am. Chem. Soc, 67, 963, 1945. 4. Rochow, E. G., U.S. Patent 2,280,995, (to General Electric), Aug. 7, 1945. 5. Rochow, E. G., An Introduction to the Chemistry of the Silicones, 2nd ed., John Wiley & Sons, New York, 1951. 6. Noll, W., Chemistry and Technology of Silicones, Academic Press, New York, 1968. 7. Barry, A. J. et al., J. Am. Chem. Soc, 69, 2916, 1947. 8. Wagner, G. H. and Struther, C. O., U.S. Patent 2,632,013 (to Union Carbide and Carbon Corp.) March 17, 1953. 9. Meals, R. N. and Lewis, F. M., Silicones, Reinhold Publishing, New York, 1950. 10. Hardman, B. and Torkelson, A., Silicones, John Wiley & Sons, New York, 1989. 11. McGregor, R. R., Silicones and Their Uses, McGraw-Hill, New York, 1959. 12. Fox, H. W., Taylor, P. W., and Zisman, W. A., Polyorganosiloxanes surface active properties, Ind. Eng. Chem.. 39, 1409, 1947. 13. Steinbach, H. and Sucker, C, Ber. Busenges Phys. Chem., 67, 407, 1963. 14. Brown, E. D., Methyl Alkyl Silicones. A New Class of Lubricant, ASLE Trans., 9, 31, 1966. 15. Demby, D. H. et al., in Synthetic Lubricants and High Performance Fluids, Skubkin, R. L., Ed., Marcel Dekker, New York, 1992, 183. 16. Barnes, J. E. and Wright, J. H., Silicone greases and compounds: their components, properties and applications, presented at NLGI 55th Meeting, Tampa, Florida, 1988. 17. Warrick, E. L., Hunter, M. J., and Barry, A. J., Ind. Eng. Chem., 44, 2196, 1952. 18. Fitzsimmons, V. G., Pickett, D. L., Militz, R. O., and Zisman, W. A., Trans. ASME, 68, 361, 1946. 19. Murphy, C. M., Saunders, C. E., and Smith, D. C, Ind. Eng. Chem., 42, 2462, 1950. 20. Ballentine, O. M., Method for determining thermal stability of synthetic oils, WADC Tech. Rep. 54–417, 1958. 21. General Electric Silicones Technical Data Book S-9 G.E. Silicones, Waterford, NY. 22. General Electric Silicones Historical Technical Data Book S-10 G.E. Silicones, Waterford, NY. 23. Steyn, R. P., Friction and wear of plastics, Met. Eng. Quart., May 1967. 24. Fitzsimmons, V. C, Pickett, D. L., Miller, R. O., and Zisman, W. A., Trans. ASME, 68, 361, 1946. 25. Agens, M. G., U.S. Patent 2,837,482, June, 1958. 26. Brown, E. D., U.S. Patent 4,005,023, 1981. 27. Holdstock, N. G., Brown, E. D. et al., WADC Tech. Rep., 56–168, 1956, 1957, 1958. 28. Visilox Div., Locktite Corp. Technical Data Folder, Poestenkill, NY.
Copyright © 1994 CRC Press, LLC
Volume III
TRIBOLOGY OF MAGNETIC STORAGE SYSTEMS
325
Bharat Bhushan
INTRODUCTION
The basic principle of magnetic recording (writing) and playback (reading) shown schematically in Figure 1, involves the relative motion between a magnetic medium and a read/write ring head.1–5 In writing (recording), a varying signal current passing through the coil of the magnetic head generates a corresponding variable flux field. This field then magnetizes the medium moving past the write head to give a permanent magnetic record of the original signal current variations. When the recorded medium passes over the head, each magnetic particle in the medium supplies flux lines, proportional in magnitude to the medium magnetization. The flux lines from the medium permeate the core and the induced voltage in the head winding, after suitable amplification, reproduces the original signal. A single head can be used for both read and write functions. More recently, miniaturized read heads have been introduced of the magnetoresistive (MR) type in which a strip of a ferromagnetic alloy (for example, Ni80Fe20) is mounted perpendicular to the plane of the medium. The variation of the magnetic field component in the magnetic medium (perpendicular to the plane of medium) causes variation in the electrical resistance of the MR strip which can be readily measured.6 One disadvantage is that both read and write functions cannot be combined in one head. In 1977, Iwasaki and Nakamura proposed perpendicular (vertical) recording for ultrahigh density magnetic recording.7 In vertical recording, magnetization is oriented perpendicular to the plane of medium rather than in its plane. Single-pole heads are generally used for recording on vertical media. Ring-shaped heads, previously discussed, have also been used with some design changes. Vertical recording has the advantage of reduced self-demagnetization.8 For high areal recording density, the linear flux density (number of flux reversals per unit distance) and the track density (number of tracks per unit distance) should be as high as possible. Reproduced (read) signal amplitude decreases with a decrease in the recording wavelength and/or the track width. The signal loss occurs from the magnetic coating thickness, head-to-medium spacing (clearance or flying height), and read gap length. The spacing loss of interest here was first described by Wallace,9 where λ is the recording wavelength and d is the head-medium separation. We note from Equation l that the spacing loss can be reduced exponentially by reducing the separation between the head and the medium. The noise in the reproduced signal needs to be minimized. Thus, signal-to-noise ratio (SNR) must be as high as possible. The wide-band SNR is also dependent upon head-to-medium spacing. The heads in modern magnetic storage systems are designed so that they develop hydrodynamic (self-acting) air bearings under steady operating conditions to minimize headmedium contact. Physical contact between the medium and the heads occurs during starts and stops. In modern tape and disk drives, the head-to-medium separation ranges from 0.1 to 0.3 µm, and roughness of the head and medium surfaces ranges from 1.5 to 10 nm rms.8.10 The need for ever larger recording-bit densities requires that surfaces be as smooth as possible and that flying height be as low as possible. Smoother surfaces lead to increase in adhesion, friction, and interface temperatures, and closer flying heights lead to occasional rubbing and Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
326
FIGURE 1. Principle of horizontal magnetic recording and playback.
increased wear with increased demands on interface materials and lubricants, control of dynamics of the head and medium, and the environment. A fundamental understanding of the tribology of this magnetic head-medium interface becomes crucial for the continued growth of this $80 billion a year magnetic storage industry. This chapter initially defines the construction and materials used in different magnetic storage devices. It then describes the theories of friction and adhesion, interface temperatures, wear, and solid-liquid lubrication relevant to magnetic storage systems.
MAGNETIC STORAGE SYSTEMS
Tape Drives The linear analog technique is most commonly used for domestic audio recorders which typically use a 4-mm wide tape. There is always physical contact between the head and the tape. Rotary single-track heads developed for video recording are also used for high recording density. For professional audio recording, dedicated digital technologies have emerged based on either a multi-track stationary head (S-DAT) or a rotating head (R-DAT). A video recorder uses a helical-scanning rotating head. Because a rotating head can be moved at a greater speed than a heavy roll of tape, much higher data rates can be achieved than with a linear tape drive. (Multi-track stationary heads can also provide high data rates.) Tapes with a width of 12.7 or 8 mm are most commonly used. Digital video recorders yield a high signal-to-noise ratio. Digital tape drives are used for data processing (computer) applications. Figure 2 shows a schematic of the high-density, high-data-rate, IBM 3480/3490 tape drive for mainframe computers. For this drive, a 165 m long, 12.7 mm wide, and approximately 17-µm thick particulate tape wound on a reel is housed inside a rectangular cartridge (100 x 125 x 25 mm). The 18-track read-write thin film head shown in Figure 2 uses bleed slots to reduce flying height for maximum reproduced amplitude. Edge slots provide flying uniformity. The write head is an inductive type and the read head is a MR type. Copyright © 1994 CRC Press, LLC
Volume III
FIGURE 2. (a) Schematic tape path and (b) schematic of magnetic thin-film head (with a radius of cylindrical contour of about 20 mm) in an IBM 3480/3490 dataprocessing tape drive. (From Bhushan, B., Tribology and Mechanics of Magnetic Storage Devices, Springer-Verlag, New York, 1990.)
327
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
328
FIGURE 3. Schematic of the interface for flexible disk drives.
The helical scanning rotary head configuration in a 8-mm tape format similar to video recorders and in a 4-mm tape format the same as R-DAT audio recorders is also used in dataprocessing tape drives; 8-mm format is used for very high volumetric density for midrange computers (work stations). Drives using 130-mm full height form factors are commonly used.
Flexible Disk Drives A flexible recording disk, also called a floppy disk or diskette, is physically a thin (~82 µm) pliable disk contained in a removable cartridge. The disk is clamped at its center in the drive and rotated at relatively low speed, while the read-write head accesses the disk through a slot in the jacket. In most designs, the accessing arms traverse above and below the disk and with read-write head elements mounted on spring suspensions. Head positioning is usually accomplished by a stepping motor (Figure 3). The shape of flexible disk heads is either spherical or flat. The most commonly used disk drives today are in 90-mm (3.5 in.) and 130mm (5.25 in.) form factors which use disk diameters of 85.8 mm and 130.2 mm, respectively. The 50-mm (2 in.) form factor drives use 47-mm or 50.8-mm diameter disks, and these drives are used in “notebook” computers. The 90-mm form factor disks use a metal hub and are encased in a hard plastic jacket which does not bend like the 130-mm form factor soft jackets and incorporates a shutter to protect the disk surface.
Rigid Disk Drives The rigid disk-drive “Winchester” technology utilizes a nonremovable stack of rigid disks rotated at constant angular speed. Typically one or two heads (for fast accessing) are positioned by an actuator over concentric data tracks on each disk surface. While all heads are actuated together, only one head is selected at a time to read or write. The schematic of the head-disk interface of a high-density, high-data rate IBM 3390-type disk drive with a linear actuator driven by a voice coil motor is shown in Figure 4. A fairly large brushless DC motor used to drive the precision spindle must overcome static friction from often many heads resting on the disks. The motor is mounted inside the spindle or underneath the base plate casting. Preloaded and sealed ball bearings are used to support the spindle shaft in both ends. Conventionally, a slider is mounted on a flexture in the orientation optimal for linear actuators. The longitudinal axis of flexure points is in the direction of carriage actuation, and Copyright © 1994 CRC Press, LLC
Volume III
329
FIGURE 4. Schematic of the head-disk assembly in IBM 3390 rigid-disk drives consisting of the voice-coilmotor driven head-arm assembly and disk stack.
the slider mounted at a right angle. However, current trends are toward smaller, more compact disk storage devices using compact, low-mass, rotary actuators to save space in the drive. In rotary actuators, the slider is mounted along the rotary arm. The actuator is operated by a stepping motor or voice-coil motor (VCM) as in Figure 4. The VCM is very much like a loudspeaker coil/magnet mechanism which provides the desired linear or the rotary motion directly, whereas circular motion provided by the stepping motor needs to be converted by a lead screw or capstan-band method. Furthermore, the VCM provides faster and smaller stepping than that by stepping motor. The small drives use both linear and rotary actuators driven by either a stepping or a VCM motor. The large drives use a linear actuator driven by a VCM. The actuator connected to the VCM rides on a set of ball bearings on the tracks as shown in Figure 4. The head slider used in high-end rigid disk drives (IBM 3380K/3390) is a two-shaped rail, taper-flat design supported by a nonmagnetic steel leaf spring (flexure) suspension to allow motion along the vertical, pitch, and roll axes, Figure 5. The front taper pressurizes the air lubricant, while some air leaking over the side boundaries of the rail results in a pitch angle. In the shape-rail design, the leading-edge rail width is greater than that of the trailing end, to attain increased pitch angles, independent of air-film thickness.” The inductive-type thin-film read-write elements are integrated in the Al203-TiC slider at the trailing edge of each rail where the lowest flying height occurs. Suspension supplies a vertical load of either 100 mN (10 g) or 150 mN (15 g), which is balanced by the hydrodynamic load when the disk is spinning. Stiffness of the suspension (~25 mN mm-1) is several orders of magnitude lower than that of the air bearing (~0.5 kN mm-‘) so that most dynamic variations are taken up by the suspension without degrading the air bearing. Small disk drives commonly use inductive coil-type heads of one of two types: minimonolithic (mini-Winchester) and minicomposite. A minimonolithic head slider consists Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
330
FIGURE 5. Schematic of IBM 3370/3380/3390 type suspension-slider assembly.
of a slider body and a core piece carrying the coil, both consisting of monolithic magnetic material (typically Mn-Zn ferrite). The taper-flat bearing area is provided by the outer rails of a trirail design. The center rail defines the width of the magnetic element in the trailing edge where a ferrite core is formed. A minimonolithic head slider consists of a Mn-Zn ferrite core and read-write gap, glass bonded into the air-bearing surface of a nonmagnetic, wear resistant slider (typically calcium titanate). The 3380-type suspensions normally used for heads in small drives apply a 95-mN (9.5-g) load onto the slider. The 3380-K/3390-type sliders are about 4.045 mm long by 3.200 mm wide by 0.850 mm high with a mass of 0.45 mN (45 mg), as opposed to about 4.1 mm long by 3.1-mm wide by 1.4 mm high and mass of 0.7 mN (70 mg) for the mini-Winchester. More recently, micro- and nano-sliders using thin-film inductive and MR heads and Al2O3-TiC as a substrate material have been introduced both for large and small drives which are about 70% and 50% of the regular slider sizes, respectively, and vertical loads on the sliders are 50 mN (5 g) and 30 mN (3 g), respectively. The surface roughness of the air-bearing rails is typically 1.5–2.5 nm rms. Schematic representations of head-medium interfaces for tape, flexible, and rigid-drive drives are shown in Figure 6. Environment, usage time, and contamination (external and wear debris) play a significant role in the reliability and usable lifetime of the interface.
Magnetic Heads Magnetic heads used to date consist either of conventional inductive or of thin-film inductive and magnetoresistive (MR) devices. Film head design capitalizes on semiconductorlike processing technology to reduce fabrication costs, and thin-film technology allows production of high-track density heads with accurate track positioning control and high reading sensitivity. Conventional heads consist of a body forming the air bearing and a magnetic ring core carrying the wound coil with a read-write gap. In the film heads, the core and coils or MR strips are deposited by thin-film technology. Air-bearing surfaces of tape heads are cylindrical in shape. The tape is slightly underwrapped over the head surface to generate hydrodynamic lift during read-write operations. For inductive-coil tape heads, the core materials have been typically permalloy and Sendust. However, since these alloys are good conductors, it is sometimes necessary to laminate the core structure to minimize losses due to eddy currents. The air-bearing surfaces of most inductive coil-type heads consist of plasma-sprayed coatings Copyright © 1994 CRC Press, LLC
Volume III
331
FIGURE 6. Schematic diagrams of (a) head-tape interface, (b) head-flexible disk interface, and (c) head-rigid disk interface.
of hard materials such as Al2O3-TiO2 and ZrO2. Read and write heads in modern tape drives (such as IBM 3480/3490) are miniaturized using thin-film technology (Figure 2b). Film heads are generally deposited on Ni-Zn ferrite (11 wt% NiO, 22 wt% ZnO, 67 wt% Fe2O3) substrates. Flexible disk heads are inductive coil-type composite devices which are either spherically contoured or flat. Mn-Zn ferrite (30 wt% MnO, 17 wt% ZnO, 53 wt% Fe2O3) is
Copyright © 1994 CRC Press, LLC
332
CRC Handbook of Lubrication and Tribology
generally used for head cores and barium titanate for the magnetically inert support structures. Material used in construction of the thin-film (Winchester-type) head in large disk drives is generally (nonmagnetic) Al2O3-TiC (70–30 wt%) or yttria-stabilized zirconia/aluminatitanium carbide composite. Small rigid disk drives for low-end applications use heads with magnetic ring core and a wound coil, and one of the two types of head sliders described previously: minimonolithic (mini-Winchester) or minicomposite. Typical physical properties of hard head materials are presented in Table 1.
Magnetic Media Magnetic media fall into two categories: (a) particulate media, where magnetic particles are dispersed in a polymeric matrix and coated onto the polymeric substrate for flexible media (tape and flexible disks) or onto a rigid substrate, such as aluminum, or more recently introduced glass for rigid disks; or (b) thin-film media, where continuous films of magnetic materials are deposited onto the substrate by vacuum techniques. Requirements of higher recording densities with low error rates have resulted in increased use of thin films which are smoother and considerably thinner than the particulate media. Thin-film media are extensively used for rigid disks and are beginning to be used for high-density audio/video and dataprocessing tapes. Cross sectional views of a particulate and a thin-film (evaporated) metal tape are shown in Figure 7. Flexible disks are similar to tapes in construction, except that these have magnetic coating on both sides and the substrate is generally about 76.2 µm in thickness. Base film for flexible media is almost exclusively poly(ethylene terephthalate) (PET) film, while polyimides and polyamide-polyimide copolymers have been explored for better dimensional stability of flexible disks. Typically 6.35 to 36.07-µm thick PET substrate with rms roughness of about 1.5 to 2.5 nm for particulate media and 1.5 to 2.5 nm rms for thin-film media is used for tapes and 76.2 µm thick for flexible disks. The base film is generally coated on both sides with a magnetic coating, typically 2 to 4 µm thick and containing 70 to 80% by weight (or 43 to 50% by volume) acicular magnetic particles (such as γ-Fe2O3, Co-modified -γ-Fe2O3, CrO2 [only for tapes], and metal particles for horizontal recording or hexagonal platelets of barium ferrite for both horizontal and vertical recording). These magnetic particles are held in polymeric binders such as polyesterpolyurethane, polyether-polyurethane, nitrocellulose, poly(vinyl chloride), poly(vinyl alcoholvinyl acetate), poly(vinylidene chloride), VAGH, phenoxy, and epoxy. To reduce friction, the coating consists of 1 to 7% by weight of lubricants (mostly fatty acid esters, e.g., tridecyl stearate, butyl stearate, butyl palmitate, butyl myristate, stearic acid, and myrstic acid). Finally, the coating contains a cross-linker or curing agent (such as functional isocyanates); a dispersant or wetting agent (such as lecithin); and solvents (such as tetrahydrofuran and methyl isobutylketone). In some media, carbon black is added for antistatic protection if the magnetic particles are highly insulating and abrasive particles (such as Al2O3) are added as a headCopyright © 1994 CRC Press, LLC
333
Volume III
FIGURE 7. Sectional views of (a) a particulate, and (b) a thin-film magnetic tape.
cleaning agent and to improve wear resistance. The coating is calendered to a surface roughness of 8 to 15 nm rms. For antistatic protection and for improved tracking, most magnetic tapes have a 1- to 3µm-thick backcoating of polyester-polyurethane binder containing conductive carbon black and TiO2, typically 10% and 50% by weight, respectively. Flexible disks are packaged inside a soft polyvinyl chloride (PVC) jacket or an acrylonitrile-butadiene-styrene (ABS) hard jacket (for smaller 90-mm form factor). Inside the jacket, a protective fabric liner is used to minimize wear or abrasion of the medium. Wiping action of the liner on the medium coating removes and entraps particulate contaminants which may originate from diskette manufacturing, the jacket, head-disk contact (wear debris), and the external environment. The liner is made of nonwoven fibers of PET, rayon, polypropylene, or nylon, thermally or fusion bonded to the plastic jacket at spots. The soft jacket near the data window is pressed with a sponge pad to create slight friction and hence stabilize disk motion under the heads. The hard cartridge is provided with an internal plastic leaf spring for the same purpose. Thin-film (also called metal-film or ME) flexible media consist of polymer substrate (PET or polyimide) with an evaporated film of Co-Ni (with about 18% Ni) and, less commonly, evaporated/sputtered Co-Cr (with about 17% Cr) for vertical recording, which is typically 100 to 300 nm thick. Electroplated Co and electroless plated Co-P, Co-Ni-P, and Co-Ni-Re-P have also been explored but are not commercially used. Since the magnetic layer is only 100 to 300 nm thick, the surface of the thin-film medium is greatly influenced by the surface of the substrate film. Therefore, an ultrasmooth PET substrate film (rms roughness ~ 1.5–2 nm) is used to obtain a smooth tape surface (~ 5–6 nm rms) for high-density recording. Several undercoatings, overcoatings, and oxidation treatment (by adding oxygen into the vacuum chamber during evaporation) are used to increase corrosion resistance and durability. An undercoating layer has been employed consisting of very fine particles, each protruding a few nanometers. Various organic (such as acrylic polymer in 50 to 80 nm thickness) and inorganic overcoats (such as diamondlike carbon in about 10 to 20 nm thickness, SiO2 and ZrO2) have Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
334
FIGURE 8. Sectional views of (a) a particulate, (b) a thin-film magnetic rigid disk.
been proposed to protect against corrosion and wear. Addition of alumina (~0.2 µm) particles in polymeric overcoats has been proposed. In the manufacturing of commercial ME tapes, oxygen is leaked into the vacuum chamber at a controlled rate during the Co-Ni evaporation to form an oxide top layer of 10 to 30 nm, for improved durability, corrosion resistance, and decreased magnetic noise. Surface oxidation of evaporated Co-Cr coating by annealing in air has also been reported to improve coating durability. In addition to an oxide layer, generally a thin layer (2 to 10 nm in thickness or 4 to 20 mg/ m2 in weight) of liquid lubricant is applied on a medium surface to further reduce friction, wear, and corrosion. Fatty acid esters or perfluoropolyethers are most commonly used.10 Figure 8 shows sectional views of two types of rigid disks—a particulate disk and a thinfilm disk. The substrate for rigid disks is generally non-heat-treatable aluminum-magnesium alloy AlSI 5086 (95.4% Al, 4% Mg, 0.4% Mn, and 0.15% Cr) with an rms surface roughness of about 15 to 25 nm rms and a Vickers hardness of about 90 kg/mm2. For particulate disks, the Al-Mg substrate is sometimes passivated with a very thin (<100 nm) conversion coating based on chromium phosphate. The finished substrate is spin coated with the magnetic coating and burnished to a surface roughness of about 7.5 to 15 nm rms. The binder is generally made of a hard copolymer of epoxy, phenolic, and polyurethane constituents. About 30 to 35% by volume of acicular magnetic particles of λ-Fe2O3 are interdispersed in the binder. A small percentage of A12O3 particles 0.2 to 0.5 µm in size (2 to 8% by volume) are added to improve wear resistance. A thin film of perfluoropolyether lubricants is applied topically. The magnetic coating is made porous for lubricant retention.8 For high-density recording there is a trend to use thin-film disks, with the typical construction shown in Figure 8. For thin-film disks with metallic magnetic layer, the Al-Mg substrate is electroless plated with nickel-phosphorous (90–10 wt%) layer to improve its surface hardness to 600 to 800 kg/mm2 (Knoop) and smoothness. The coated surface is polished with an abrasive slurry to a surface roughness of about 2 to 4 nm rms. For a thin-film Copyright © 1994 CRC Press, LLC
335
Volume III
disk with an oxide magnetic layer, a 2- to 20-µm thick alumite layer is formed on the Al-Mg substrate through anodic oxidation in a CrO3 bath. Ni-P cannot be used because it becomes magnetic when exposed to high temperature during preparation of -γ-Fe2O3 film. The start-stop zone of substrates for thin-film disks are generally textured mechanically in the circumferential or random orientation to 6 to 8 nm rms roughness to minimize static friction at the head-disk interface. For convenience, the entire disk is generally textured. Circumferential direction of texturing in the data zone (if textured) is preferred in order to keep the magnetic orientation ratio as high as possible. The finished substrate is coated with a magnetic film 25 to 150 nm thick. Some metal films require a Cr undercoat (10 to 50 nm thick) as a nucleation layer to improve magnetic properties, such as coercivity. Typically, magnetic films that have been explored are metal films of cobalt-based alloys, with sputtered iron oxide being the principal exception. Magnetic films that are used to achieve the high recording density have weak durability and poor corrosion resistance. Protective overcoats with a liquid lubricant overlay are generally used to provide low friction, low wear, and corrosion resistance. Protective coatings are typically sputtered diamond-like carbon, spin-coated or sputtered SiO2, sputtered yttria-stabilized zirconia, and plasma-polymerized protective (PPP) films. In most cases, a thin layer of perfluoropolyether lubricant is used.8 Typical materials used for various magnetic media and operating conditions for dataprocessing applications are shown in Table 2. Typical physical properties of components of magnetic media are presented in Table 3.
FRICTION AND ADHESION
When two surfaces come in contact, the load is supported by the deformation at the tips of the contacting asperities. The proximity of the asperities results in adhesive contacts caused by interatomic attractions. In a broad sense, adhesion is considered to be either physical or chemical in nature. Experimental data suggest that adhesion is primarily due to weak van der Waals forces. When the two surfaces (in contact) move relative to each other, frictional force, commonly referred to as “intrinsic” or “conventional” frictional force, is contributed by adhesion and deformation (or hysteresis) of these asperities. For most practical cases, adhesional friction is the primary contributor.8 In addition, “suction” can occur due to meniscus/viscous effects, microcapillary evacuation, and changes in surface chemistry.12,15 Here we will concentrate on the meniscus/viscous effects only. Generally, any liquid that wets or has a small contact angle on surfaces will condense from vapor in the form of an annular-shaped capillary condensate in the contact zone. The pressure of the liquid films of the capillary condensates or preexisting film of lubricant can significantly increase the adhesion between solid bodies. Liquid-mediated adhesive forces can be divided into two components: meniscus force (FM) due to surface tension and a rate-dependent viscous force (Fv). The total tangential force F required to separate the surfaces by sliding is equal to an intrinsic force (FA) and stiction force FA (combination of friction force due to meniscus effect and the peak viscous force) where fr is “true” static coefficient of friction and W is the normal load. The normal force required to move two flat, well-polished surfaces (such as magnetic head and medium surfaces) in the presence of liquid medium and/or sticky substance can be large (up to several N in extreme cases). Therefore, we define the difference between stiction and conventional static and kinetic friction being that stiction requires a measurable normal force (normally several mN or higher) to pull the two surfaces apart from the static conditions. Copyright © 1994 CRC Press, LLC
336
CRC Handbook of Lubrication and Tribology
Copyright © 1994 CRC Press, LLC
Volume III
337
Conventional Friction From Tabor’s classical theory of adhesion,16 frictional force due to adhesion (FA) in dry contact is: and for lubricated contact, where where A, is a real area of contact, a is the fraction of unlubricated area, τa and τe are the shear strengths of the dry contact and the lubricant film, respectively, η is the absolute viscosity of the lubricant, V is the relative sliding velocity, and h is the lubricant film thickness.
Greenwood and Williamson Contact Model — The contacts can be either elastic or plastic which primarily depend on the surface topography and the mechanical properties of the mating surfaces. The classical model for a combination of elastic and plastic contacts between rough surfaces, that of Greenwood and Williamson17 (G & W), assumes the surface is composed of hemispherically tipped asperities of uniform size with their heights following a gaussian distribution about a mean plane. The radius of these asperities is assumed to be equal to the mean radius of curvature obtained from roughness measurements. The real areas of contact for elastic (e) and plastic (p) contacts are,18 Copyright © 1994 CRC Press, LLC
338
CRC Handbook of Lubrication and Tribology
FIGURE 9. Influence of plasticity index on the real area of contact in metals/ceramics. (From Bhushan, B., J. Lubr. Tech., Trans. ASME. 106, 26, 1984.)
and
where Aa is apparent area of contact; pa is apparent pressure (W/Aa) Ec is composite modulus of elasticity; H and Y are the hardness and yield strength of the softer material;
Volume III
339
FIGURE 10. NOP and AFM images of (a) particulate tape A, (b) particulate disk A, and (c) circumferentiallytextured thin-film (sputtered Co-Pt-Ni) disk B1. The wireframe NOP images are of a 250 µm x 250 µm region, and the AFM images are solid gray level images (white is high and black is low) of 2.5 µm x 2.5 µm regions of the same disk. (From Bhushan, B. and Blackman, G. S., J. Tribol., Trans. ASME, 113, 452, 1991.)
Copyright © 1994 CRC Press, LLC
340
CRC Handbook of Lubrication and Tribology
coatings used in the construction of the thin-film disk, Figures 10(c). Table 4 indicates that topography and contact statistics predictions are a strong function of lateral resolution of the roughness measurement tool.21,22Here η is the density of summits per unit area, n is the number of contact spots, and pr is the real pressure. Surface topography statistics show that the average summit radius (Rp) for the AFM data is two to four orders of magnitude smaller than that for the NOP data, whereas summit density for the AFM data is two to four orders of magnitude larger than for the NOP data. The plasticity index (Ψ) calculated using the AFM data suggests that all contacts made of nanoasperities are plastic; NOP data suggest that all contacts made of microasperities are elastic (see Figure 14, to be described later).
Fractal Model of Elastic-Plastic Contact — The contact analyses developed over the last quarter century consider only an averaged surface with a single scale of roughness to be in contact with another surface. However, due to the multiscale nature of surfaces, it is found that the surface roughness parameters depend strongly on the resolution of the roughnessmeasuring instrument or any other form of filter, hence not unique for a surface.23 Variances of surface height, slope, and curvature shown in Figure 11 suggest that instruments with different resolutions and scan lengths yield different values of these statistical parameters for the same surface. Therefore, the predictions of the contact models based on these parameters may not be unique to a pair of rough surfaces. However, if a rough surface is characterized in a way such that the structural information of roughness of all scales is retained, then it will be more logical to use such a characterization in a contact theory. In order to develop a contact theory based on this motivation, it is first necessary to quantify the multiscale nature of surface roughness. A unique property of rough surfaces is that if a surface is repeatedly magnified, increasing details of roughness are observed right down to nanoscales. In addition, the roughness at all magnifications appears quite similar in structure as qualitatively shown in Figure 12. Such a behavior can be characterized by fractal geometry.24,25 The main conclusions from these studies were that a fractal characterization of surface roughness is scale-independent and provides information of the roughness structure at all the length scales that exhibit the fractal behavior. Based on this observation, Majumdar and Bhushan26,27 developed a new fractal theory of contact between rough surfaces. Consider a surface profile z(x) such as shown in Figure 12, which appears random, multiscale, and disordered. Mathematical properties of such a profile are that it is continuous, nondifferentiable and statistically self-affine. The nondifferentiability arises from the fact that a tangent or a tangent plane cannot be drawn at any point on the surface since more and more details of roughness will appear at the point. In short, a rough surface is never smooth at any length scale. Statistical self-affinity is due to similarity in appearance of a profile under different magnifications. The Weierstrass-Mandelbrot (W-M) function satisfied all these properties and is given as,28
where the parameter D is the fractal dimension, G is a characteristic length scale of the surface, and γn are the discrete frequency modes of surface roughness. The theory of fractal geometry and the concept of fractal dimension is well described by Mandelbrot,29 Ling,30 and Majumdar and Bhushan.2 The parameters which characterize the W-M function are G, D, and η1, where Copyright © 1994 CRC Press, LLC
Volume III 341
Copyright © 1994 CRC Press, LLC
342
CRC Handbook of Lubrication and Tribology
FIGURE 11. Variation of the ratio of the rms values at a surface magnification β > 0 to the corresponding rms values at a magnification of unity, with magnification β. The data for β < 10 was obtained by NOP measurements and for β β 10 by AFM. (From Oden, P. I. et al., J. Tribol., ns. E, 114, 666, 1992.)
γ = 1.5 was found to be suitable for high spectral density and for phase randomization. Since measurement of a rough surface is a nonstationary random process, the lowest cut off frequency is related to sample length L as γη = 1/L. Parameters G and D can be found from the power spectrum of the W-M function
where S(ω) is the power of the spectrum and ω is the frequency, the reciprocal of the roughness wavelength. If S(ω) is plotted as a function of ω on a log-log plot, the power law behavior results in a straight line whose slope is related to fractal dimension D of the surface roughness and parameter G is related to the location of the spectrum along the power axis. The value of D ranges from 1 (smooth surface) to 2 (rough surface). To verify whether surfaces do follow a power-law fractal behavior and to obtain the parameters D and G of a surface, the power spectrum of a real surface profile has been compared with that of the W-M function in Equation 6. Figures 13(a) and 13(b) show the averaged spectrum of a surface profile of an untextured thin-film magnetic rigid disk. The spectrum in Figure 13(a) corresponds to the surface that was measured by an optical profiler and the spectrum in Figure 13(b) corresponds to the surface which was measured by a scanning tunneling microscope (STM). The spectrum in Figure 13(a) follows S(ω> ~ ω-2.24 corresponding to D = 1.38; whereas the spectrum in Figure 13(b) follows S(ω) ω ω-2.35 corresponding to D = 1.33 and G ~ 10-16 m. These data show that the surface of an untextured thin-film rigid disk follows a fractal structure for three decades of length scales. Fractal analysis of a magnetic tape surface at different resolutions by NOP and AFM revealed two regimes of roughness demarcated by a scale of 0.1 µm, corresponding to the size of magnetic particles.22,23,27 The fractal model of elastic-plastic contact has been developed by Majumdar and Bhushan.26,27 An interface between a statistically isotropic rough surface and a flat plane was Copyright © 1994 CRC Press, LLC
343
Volume III
FIGURE 12. Quantitative description of statistical self-affinity for a surface profile.
considered with contact spots of different sizes and spread randomly. Depending on radius of curvature and height (or deformation) of the asperity, the contact spot will be either in elastic or plastic deformation. Limit of elastic deformation (propensity of yielding) is governed by the Tresca or Huber-Mises yield criteria in which plastic flow will occur when maximum shear stress equals half of the tensile yield strength of the material. Whether contacts go through elastic or plastic deformation is determined by a critical area
This shows that small contact spots (a < ac) are in plastic contact, whereas large spots are in elastic contact. This result is in contrast with that of the G & W model where small ones are in elastic deformation, a direct implication of the assumption of uniform asperity radii. For a magnetic tape, typical values of the parameters are D = 1.97,G = 5.15 x 10-9 m and H/E = 0.14.22 Since critical contact area for inception of plastic flow for a magnetic tape is ac = 1014 m2 (contact diameter ~ 100 nm), all contact spots larger than 100 nm would deform elastically. For an untextured thin-film rigid disk, typical values of the parameters are D = 1.38, G = 10-16 m and H/E = 0.06.26 The critical contact area for inception of plastic deformation is ac = 10-27 m2, so small that all contact spots can be assumed to be elastic at moderate loads. The question remains as to how large spots become elastic when they must have initially been small plastic spots. The possible explanation is graphically shown in Figure 14. As two surfaces touch, the nanoasperities (detected by AFM type of instruments) first coming into contact have smaller radii of curvature and are therefore plastically deformed instantly, and the contact area increases. When load is increased, nanoasperities in the contact zone merge, and the load is supported by elastic deformation of the larger scale asperities or microasperities (detected by NOP type of instruments). Cumulative size distribution of the contact spots is assumed to follow the power law relation26 where the distribution is normalized by the area of the largest contact spot ae. Since the power spectra of surface indicates that a surface can be fractal even at nanoscales, as → 0 where as is the area of the smallest contact spot. Note that in the distribution of Equation 8, the number Copyright © 1994 CRC Press, LLC
344
CRC Handbook of Lubrication and Tribology
FIGURE 13. Surface topography and average power spectra at different length scales of a magnetic thin-film rigid disk C surface (a) NOP data at left, (b) STM data on the right. (From Majumdar, A. and Bhushan, B., J. Tribal., Trans. ASME, 112, 205, 1990.)
of the largest spot is unity, whereas the number of spots of area a → 0 would tend to infinity. The real area of contact, Ar is given as
Copyright © 1994 CRC Press, LLC
345
Volume III
FIGURE 13 (continued).
For the case ae > ac, the portion of the real area of contact in elastic deformation is
Total elastic-plastic load W(ae > ac) is related to the real area of contact as Copyright © 1994 CRC Press, LLC
346
CRC Handbook of Lubrication and Tribology
FIGURE 14. Schematic of local asperity deformation during contact of a rough surface (upper profile measured by NOP and lower profile measured by AFM, typical dimensions shown for a polished thin-film disk C against a flat surface. The vertical axis is magnified for clarity. Firm lines show the surfaces before contact and dotted lines show surfaces after contact. (From Bhushan, B. and Blackman, G. S., J. Tribol., Trans. ASME, 113,452,1991.)
and the total plastic load (ae < ac) For the special case of ac > 0 (e.g., in thin-film disk C), in the elastic-plastic regime Here, the load-area relationship depends on the fractal dimension, whereas G&W predict a linear relationship. Fractal model verifies the load-area relationship26 observed both for the magnetic tape A19 and for the thin-film disk C31
Measurement of Contact Area — The real area of contact of magnetic tapes and rigid disks has been measured using the optical-interference technique.9,31 A loading-unloading experiment was conducted to determine if most contacts in the range above 0.7 µm in diameter were elastic.18 Photographs of tape contacts were taken at 28 kPa; then higher pressure (1.38 MPa) was applied for short durations, and the tape contact was brought back to 28 kPa and rephotographed (Figure 15). Lack of changes in the real area of contact before loading to 1.38 MPa and after unloading to 28 kPa suggests that the contacts were elastic, an observation in agreement with fractal model predictions.
Copyright © 1994 CRC Press, LLC
Volume III
347
FIGURE 15. Optical interference photographs of tapes taken at 28 kPa; then subjected to higher pressure (1.38 MPa) for about 5 minutes and brought back to 28 kPa and rephotographed. We see no change in the real area of contact implying an elastic contact. (From Bhushan, B., J. Lubr. Tech., Trans. ASME, 106, 26, 1984.)
If the contacts are elastic, then real area of contact and friction are governed by the Ec and σ/Rp of the magnetic medium surface. Figure 16(a) shows an example that friction of various magnetic tapes studied by Bhushan et al.12 depend significantly upon the complex modulus and the surface roughness. Stable frictional behavior was exhibited only by those tapes which displayed a complex modulus of greater than 1.2 to 1.5 GPa. Figures 16(b) and 17 illustrate that coefficient of friction also strongly depends on the surface roughness and doubles for CrO2 particulate tape for rms surface roughness below about 50 nm. Figure 17 shows an example of surface roughness dependence on the coefficient of friction for a thin-film rigid disk.32 Typical contact diameters for tapes and rigid disks were found to be about 6 and 1.5 µm respectively. Bhushan and Dugger31 reported a significant increase in the contact diameter, number of contacts, and total real area of contact of the thin-film rigid disk as a function of loading time (Figure 18). Viscoelastic and viscoplastic deformations not only increase the size of existing asperities but also bring the two surfaces closer to allow contact of additional asperities. To minimize the rate of increase in the real area of contact as a function of loading time, attempts should be made to select materials for disk coatings with low creep compliance, to reduce normal stress at the head-disk interface, and to use methods (such as load/unload mechanisms) to minimize or avoid the storage of the head in contact with the disk. In the case of magnetic tapes, creep compliance and hydrolytic degradation of the binder also need to be optimized for sustained low friction after storage at high pressure (e.g., near end-of-tape on a reel) and high temperature/humidity.8 Liquid-Mediated Adhesion (Stiction) A smooth magnetic medium (especially thin-film disk) tends to adhere or stick strongly to the smooth magnetic head. Liquid-mediated adhesion, commonly referred to as “stiction”
Copyright © 1994 CRC Press, LLC
348
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
FIGURE 16. Coefficient of friction (a) during start at 30°C/85% RH measured on a commercial tape drive versus complex modulus at 50°C. The CrO2 tapes were stored at 50°C/60% RH for 14 days before the tests; (b) effect of surface roughness on coefficient of friction for CrO2 particulate tapes. (From Bhushan, B. et al., ASLE Trans.. 27. 33. 1984.)
Volume III
349
FIGURE 17. Coefficient of friction as a function of degree of disk texturing for a thin-film (metal) rigid disk with sputtered carbon overcoat against ferrite slider. (From Doan, T. Q. and Mackintosh, N. D., in Tribology and Mechanics of Magnetic Storage Systems, Vol. 5, Bhushan, B. and Eiss, N. S., Eds., SP-25, STLE, Park Ridge, IL, 1988, 6.)
FIGURE 18. Log normal distribution of asperity contact diameters for a thin-film disk C loaded by 500 mN (9.27 MPa) at two loading durations, (a) initial and (b) after loaded for 60 h. (From Bhushan, B. and Dugger, M. T., Wear, 137,41, 1990.)
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
350
FIGURE 19. Regimes of different liquid levels in the bead-medium interface.
in the computer industry, is especially pronounced when liquid lubricants and adsorbed moisture are present at the interface. Liquid-mediated adhesion can be divided into two significant components: a meniscus term which depends on surface tension of the liquid and a viscous term. The viscous term does not depend on surface tension and may be observed even when the surfaces are completely surrounded by the liquid. If the surfaces are submerged in the liquid they may be separated easily, provided the separation is carried out very slowly. However, if the rate of separation is rapid when the surfaces are pulled apart, liquid must flow into the space between them and the viscosity of the liquid will be the determining factor in stiction. For analysis purposes, we consider a model of contact region between smooth surfaces with different level of “fills” of the interface and it depends on the mean interplanar separation and the liquid levels (Figure 19). In two extreme regimes, either a small quantity of liquid bridges the surfaces around the tip of a contacting asperity (the “toe-dipping” regime) or the liquid bridges the entire surface (the “flooded” regime); in a third intermediate “pillbox” regime, the liquid bridges a significant fraction of the apparent area. The different regimes can be modeled and the expression for FM and Fv can be obtained.8,33 In the toe-dipping regime, the liquid adhesion force between a single asperity and a surface can be modeled by a sphere of composite radius of curvature in contact with a flat surface with a liquid bridge in between. Total meniscus and viscous forces of all wetted asperity contact can be calculated by multiplying the number of contacts by meniscus and viscous forces at a typical contact. The flooded regime can be modeled by a liquid bridge between two flat surfaces. The pillbox regime can be modeled by two flat surfaces. If we assume that surface asperity radii are constant and their heights follow a gaussian distribution, the true coefficient of friction fr can be obtained from the following expressions: For toe-dipping regime:
For flooded regime:
Copyright © 1994 CRC Press, LLC
Volume III
351
FIGURE 20. Effect of humidity on adhesion of CrO2 tape A in contact with a Ni-Zn ferrite pin. (From Miyoshi, K. et al., in Tribology and Mechanics of Magnetic Storage Systems, Vol. 5, Bhushan, B. and Eiss, N. S., Eds., SP-25, STLE, Park Ridge, IL, 1998, 12.)
where F is the friction force, γe and ηe are the surface tension and viscosity of the liquid, θ2 are the contact angles of the liquid on the two surfaces, h is the average thickness of the liquid bridge, L is the distance surfaces need to slide to become unstuck, α is the start-up linear acceleration, and Aa is the apparent area. In the toe-dipping regime, adhesion force is independent of the apparent area and proportional to the normal load (i.e., number of asperity contacts). However, the flooded regime shows the opposite tendencies. The pillbox regime is intermediate and can exhibit either behaviour at the extremes. In the three regimes, adhesion force decreases with an increase in σp and a decrease in Rp and is independent of η. Relative humidity of the environment, rest period, head-slider area, surface roughness, lubricant viscosity and its thickness, and relative velocity affect the liquid-mediated adhesion.8,14,34–40 Miyoshi et al.36 found that adhesive force (normal pull-off force) of a Ni-Zn ferrite pin in contact with a flat of Ni-Zn ferrite or of magnetic tape A, Figure 20, remained low below 40% RH, but nearly doubled with increasing relative humidity to 80%. Changes in the adhesion of contacts were reversible on humidifying and dehumidifying. The adhesive forces for a liquid bridge between a spherical surface with radius the same as of the pin and a flat surface calculated using surface tension and contact angle values for water compared well with measured values. The concluded that ferrites adhere to ferrites or tapes in a saturated atmosphere primarily form meniscus effects of a thin-film of water adsorbed on the interface. Bhushan and Dugger37 found that for a 3380-type Al2O3-TiC slider in contact with an untextured thin-film (metal) disk, measurable adhesion (>0.1 mN) at 90% RH for an unlubricated disk was observed only after 90 min of exposure, Figure 21. Adhesive force then increased with exposure time up to about 5 h, after which there was no further significant increase. The effect of humidity on the lubricated and unlubricated disks is shown in Figure 22. The adhesive force (normal pull-off force) of head-disk contact remained low (below resolution of the measurement technique ~ 50 mN) below 75% RH, but increased greatly above 75%. Adhesive force increased with an increase of relative humidity for both disks; however, the increase was slightly larger for the unlubricated disk than for the lubricated one. Copyright © 1994 CRC Press, LLC
352
CRC Handbook of Lubrication and Tribology
FIGURE 21. The effect of exposure time (before contact) of specimen surfaces to 90% RH nitrogen on the adhesive force for an unlubricated polished thin-film (σ = 2.11 nm) disk. Adhesive force was measured after a 5-min contact with Al203-TiC slider at 150 mN load, followed by separation at a normal velocity of 80 µm/s. (From Bhushan, B. and Dugger, M. T., J. Tribol, Trans. ASME, 112, 217, 1990.)
FIGURE 22. The dependence of the adhesive force on relative humidity for the polished thin film disk lubricated with 2 nm of Z-15 PFPE lubricant (150 cSt). Adhesive force was measured after a 5 min. contact with Al203-TiC slider at 150 mN load, followed by separation at a normal velocity of 80 µm/s. (From Bhushan, B. and Dugger, M. T., J. Tribol., Trans. ASME, 112, 217, 1990.)
(Tian and Matsudaira40 have reported that static coefficient of friction of lubricated disk increases more rapidly than that of an unlubricated disk, which is contrary to the results by Bhushan and Dugger.37) While the hydrophobic PFPE lubricant repels some of the water condensation, water can replace some PFPE at concentrated asperity contacts either during long exposures or if the lubricant is displaced by high pressure or by sliding of the two Copyright © 1994 CRC Press, LLC
Volume III
353
FIGURE 23. The dependence of adhension on for the same system as Figure 22: 150mN normal load applied for 5 min at 90% RH. (From Bhushan, B. and Dugger, M.T. J. Tribol., Trans ASME, 112, 217, 1990.)
FIGURE 24. Coefficients of static and kinetic friction as a function of lubricant thickness for polar (AM 2001) and nonpolar (Z-25) PFPE on particulate rigid disks sliding against Mn-Zn ferrite-slider. (From Scarati, A. M. and Caporiccio, G., IEEE Trans. Magn., Vol. Mag-23, 106, 1987.)
surfaces. Studies of penetration of lubricant layers by water41 suggest that water may diffuse through the lubricant and condense into droplets around nuclei on the solid surfaces. Spreading of the water droplets will be controlled by the energy difference between water/disk and lubricant/disk interfaces. At this time, water with a surface tension on the order of three to four times that of typical magnetic medium lubricants wets the magnetic-medium surface, creating a meniscus at the asperity contacts. Attempting to separate the surfaces against this water film gives rise to the observed adhesion. Bhushan and Dugger also measured the adhesive force as a function of separation rate (proportional to sliding velocity) (Figure 23). They found that adhesive force increases approximately linearly with the square root of loading rate. This is attributed to viscous effects. Lubricant thickness and its functionality effects on static and kinetic coefficients of friction in particulate disks are shown in Figure 24.42 Static friction increases with an increase in the
Copyright © 1994 CRC Press, LLC
354
CRC Handbook of Lubrication and Tribology
FIGURE 25. Coefficients of kinetic and static friction for thin-film disks with carbon overcoat sliding against Al2O3-TiC slider at 0.12 m/s as a function of lubricant thickness for the four lubricants: (a) disk X, roughness = 2.6 run rms; (b) disk X2 roughness = 6.5 nm rms; and (c) disk B, roughness = 8.2 nm rms.8 Lubricant L1Z03 (viscosity = 30 cSt, molecular weight = 4250, density = 1.824 g/cm3); lubricant L2-Z25 (viscosity = 255 cSt, molecular weight 14550, density = 1.851 g/cc); lubricant L3 = Krytox 143AD (viscosity = 1600 cSt, molecular weight = 8250, density “ 1.91 g/cm3); and lubricant F-Z-Dol (viscosity = 81 cSt, molecular weight = 2000, density - 1.81 g/cm3).38
lubricant film thickness; however, the reverse is true for kinetic friction. Increase in static friction with an increase in the lubricant thickness occurs at a lower thickness for a nonpolar lubricant than that for the functional lubricant. Lubricant and surface roughness effects on static and kinetic coefficients of friction in thin-film disks are shown in Figure 25 for three disks and four perfluoropolyether lubricants.38,43 Lubricants labeled L1, L2, and L3 are nonpolar liquid lubricants, while F is polar with dihydroxyl functional end groups. Carbon atoms form a branched structure in L3, while lubricants L1, L2, and F contain carbon atoms in linear arrangements. The buildup of the friction is believed to be governed by microflow capabilities of the liquid on the disk. Static friction does not show a monotonic increase at higher lubricant thicknesses as seen for Copyright © 1994 CRC Press, LLC
355
Volume III
kinetic friction, and both static and kinetic friction coefficients are essentially independent of lubricant thickness below a “critical” lubricant thickness. While all lubricants show a sharp increase in friction above this thickness, polar lubricant F exhibits the smallest critical thickness for the onset of friction increase. In the case of nonpolar lubricants (L1, L2, and L3) the lower critical thickness for lower viscosity likely reflects more ready flow to develop meniscus bridges with resulting higher friction. Polar lubricant F exhibits high friction compared to the nonpolar lubricants and does not follow the viscosity trend. Increase in friction with increasing lubricant film thickness above the critical thickness can be attributed to strong adhesive forces in the interface.8 Critical thickness increases with disk surface roughness with a larger mean separation of the surface involving a thicker lubricant film when the menisci are formed. The trend for higher friction for smoother disks when above critical film thickness is only applicable to short contact times on the order of seconds or minutes. With longer rest times (hours or days), adhesion usually reaches much higher values, even if the lubricant thickness is well below the critical value. This time effect can be explained by the slow diffusion of the lubricant molecules towards contact points driven by the Laplace pressure and deformation of interacting asperities to increase real area of contact. Streator et al.38 also reported that kinetic coefficient of friction decreases with increasing sliding speed.
INTERFACE TEMPERATURES
During sliding, almost all frictional energy is converted to heat in the material close to the interface and asperity interactions produce numerous high temperature flashes. To predict head-medium interface temperature, the contact can be modeled as a series of sphericallytopped asperities.44,45 The interaction problem at an asperity contact reduces to a sphere against another sphere, assuming the distance to the center of the two spheres is fixed. When one sphere comes in contact with the other, the real area of contact starts to grow; when one sphere is directly above the other, the area is at maximum; as one sphere moves away, that area starts to get smaller. The real area of contact is a source of frictional heat, and the heat intensity is proportional to the real area. The total flash temperature consists of temperature rise of an individual asperity contact supplemented by the related influence of other nearby asperity contacts.46 Relevant equations for the average and maximum asperity temperature rise of the interface are given as:47
where f is coefficient of friction, pa is apparent pressure, pCp is volumetric heat capacity, K is thermal diffusivity, k is thermal conductivity, dmax is the maximum contact diameter, and l is half length of the slider. Average and maximum transient temperatures predicted for a typical particulate headtape interface were 7 and 10°C, respectively.44 These predictions compared reasonably well with infrared measurements.48 Asperity contact temperatures at a head-tape interface are relatively low because of its high area of contact, as compared to that of metal-metal or Copyright © 1994 CRC Press, LLC
356
CRC Handbook of Lubrication and Tribology
FIGURE 26. SEM micrographs of worn Ni-Zn ferrite head with a Cr02 tape A: (a) abrasion marks (direction of tape motion-left to right); (b) surface pull-out. (From Bhushan, B., in Tribology and Mechanics of Magnetic Storage Systems, Vol. 2, Bhushan, B. and Eiss, N. S., Eds., SP-19, ASLE, Park Ridge, IL, 1985, 101.)
ceramic-ceramic contacts. The transient temperature rise of 7 to 10°C rise can lead to high friction in some tapes because transition of some tape mechanical properties occurs within 5°C above ambient temperature. In isolated cases of high speed contact of magnetic particles with the head surface, the average and maximum transient temperature rise could be about 600 and 900°C, respectively. These temperatures potentially cause breakdown of the medium lubricant and degradation of me medium binder to give excessive friction and even seizure. Average and maximum transient temperatures predicted for a typical particulate rigid-diskslider interface are 34 and 44°C, respectively.45 If exposed magnetic or alumina particles contact the slider surface, transient rise could be more than 1000°C. Predicted average and maximum transient temperature rises for a typical thin-film disk-slider interface are 56 and 81°C, respectively, for an Al203-TiC slider and 77 and 110°C for a Mn-Zn ferrite slider, these match reasonably well with the infrared measurements by Bair et al.49 and Suzuki and Kennedy.50 The size of an asperity contact is on the order of 1.5 µm. Since duration of asperity contact at full operating speed is less than 1 ms, the thermal gradients perpendicular to the sliding surfaces are very large (a temperature drop of 90% in a depth typically less than a micron).
WEAR
Head-(Particulate) Tape Interface Wear of oxide magnetic particles and ceramic head materials is different from metallic wear because of the inherent brittleness and relatively low surface energy of ceramics. The first signs of ferrite head wear with a magnetic tape are very fine scratches as small as 25 nm on the head surface (Figure 26(a)).51 Ferrite surface is microscopically removed in a brittle manner as strips or islands, depending on the tape smoothness. Wear generally occurs by microfragmentation of the oxide crystals in the ceramic surface. Fragmentation is the result of cleavage and transgranular fractures, one dominated by intergranular fracture. The worn head surface is work hardened, which reflects a shift from domination by transgranular fracture to intergranular. This commonly results in a “pullout” region where fracture and rupture have occurred (Figure 26(b)). Debris originating from these regions causes additional small scale plastic deformation and grooves (three-body abrasion). The worn ferrite head surface is work hardened with a large compressive stress field which is detrimental to magnetic signal amplitude.52-55 Wear resistance of common head materials against a c-Fe2O3 tape shows a linear relationship with material hardness (Figure 27).56 Head wear also depends on grain size of the head material, magnetic particles, tape surface roughness, isolated asperities on the tape surface, tape tension, and tape sliding speed. Head wear increases with surface roughness of Copyright © 1994 CRC Press, LLC
Volume III
357
FIGURE 27. Wear rate of magnetic materials slid against a diamond cone as a function of Vickers hardness. (From Tanaka, K. and Miyazaki, O., Wear, 66, 289, 1981.)
FIGURE 28. Ni-Zn ferrite head wear as a function of (a) rms surface roughness of a CrO2 tape A in streaming mode;52 (b) asperity counts on CrO2 tape A in streaming mode.57
tapes (Figure 28(a)) and isolated asperities on the tape surface (Figure 28(b)).57 Wear rate also increases above about 40 to 60% relative humidity (Figure 29).58 This increase is believed to be due to moisture-assisted fracture (or static fatigue) of the grains to yield finer particles.8,52 Head surfaces after usage sometimes become coated with thin layers of new organic high molecular weight “friction polymers” or “tribopolymers”. Friction is essential for their formation and one of the surfaces, lubricant, or even a material nearby should be organic.59– 61 It seems clear that all friction polymers are products of chemical reaction, whether they derive initially from solid polymers or from organic liquids or vapors. Corresponding to their discoloration of the head, friction polymer coatings are sometimes called brown or blue stains. During contact of particulate tape with the head in contact start/stops (CSS) or during partial contact in streaming, debris is generated primarily by adhesive wear. The debris can be either loose or adherent.62 Tape debris, loose magnetic particles, worn head material, or foreign contaminants are introduced between the sliding surfaces and abrade material off each. Debris Copyright © 1994 CRC Press, LLC
358
CRC Handbook of Lubrication and Tribology
FIGURE 29. Head wear rate with γ-Fe2O3 tape as a function of relative humidity. (From Kelly, J., in Magnetic Tape Recording for the Eighties, NASA Ref. Publ. 1075, 1982, 7.)
that adheres to drive components leads to polymer-polymer contact whose friction is higher than that of rigid material-polymer contact and can lead to magnetic errors and sometimes catastrophic failures.8 In low-speed sliding experiments of a Ni-Zn ferrite pin against CrO2 tape A, Calabrese et al.63 observed that tape particles consisting of binder resin and magnetic particles were literally thrown out of the interface to land within a radius of 0.8 mm from the pin. If particles landed in the pin path, they would be drawn through the contact to generate more particles (three-body abrasion scenario). Wear particles were generally in the form of about 5-µm blocks and much smaller flakes. In wear studies with an irradiated Ni-Zn ferrite head run against various tapes, Bhushan et al.64 found measurable transfer of ferrite on the CrO2-tape after 5000 passes. A Co-γFe2O3 tape did not show significant transfer, even after 20,000 passes. Wear was represented in four distinct patterns: smeared areas, gray areas, dots or specks, and streaks or lines running in the direction of the tape. About 0.6 ng/cm2 of ferrite deposited on the CrO2-tape after 20,000 passes across the head. About 0.6% of the generated ferrite debris was transferred to the tape, about 0.2% to tape-drive component surfaces, and the rest was believed to be airborne.
Head-(Particulate) Rigid Disk Interface During asperity contacts, disk debris can be generated by adhesive, abrasive, and impact wear. Wear debris generated during disk manufacture (burnishing or buffing) and foreign contaminants can result in three-body abrasive wear and in performance degradation of the slider air bearing. Flash temperatures generated at asperity contacts can render boundary lubricants ineffective and degrade the disk binder to cause high friction and high disk wear. Any of these mechanisms can lead to head crash. Humidity and temperature also significantly affect head-disk friction and wear or debris generation. Disk and head surfaces after head crash in a CSS test show microscopic circumferential wear grooves in either two-body or three-body abrasive wear (Figure 30). Karis et al.65 and Novotny and co-workers66,67 reported the removal of lubricant from portions of the start/ stop track, and degradation of lubricant preceded the final disk failure. Lubrication of the disks increased up to 1000 times the number of cycles until the coating began to wear through. Cycles until frictional failure was proportional to the areal density of lubricant. Scarati and Copyright © 1994 CRC Press, LLC
Volume III
359
FIGURE 30. SEM Micrograph of a worn particulate disk surface against Al2O3-TiC slider after extended use in contact start/stops (CSS). (From Bhushan, B., Tribology and Mechanics of Magnetic Storage Devices, Springer-Verlag, New York, 1990.)
FIGURE 31. Wear life as a function of lubricant film thickness on a particulate rigid disk slid against Mn-Zn ferrite slider. (From Scarati, A. M. and Caporiccio, G., IEEE Trans. Magn., Vol. Mag-23, 106, 1987.)
Caporiccio42 found that relative wear life of particulate disks increases with an increase in the lubricant thickness (Figure 31; see also Reference 65), and that polar PFPE lubricants have longer wear life than nonpolar lubricants. In a study of interface failure or head crash for particulate disks in CSS tests, Kawakubo and co-workers68,69 measured friction force and acoustic emission to monitor head-disk contact, and read-back magnetic signals. At failure, the read-back signal decreased to almost zero and friction force and the AE signal rose significantly. This implies that the head was virtually in contact even at full speed. In videotaping the wear process through a transparent sapphire slider, Kawakubo et al.68 found that disk debris transferred to the rail surfaces preceding the interface failure. Using submicron, fluorescent polystyrene-latex particles, Hiller and Singh70 studied the interaction of contaminant particles with a flying slider. The particles were deposited on the slider mainly in two regions: in the tapered portion of the air-bearing rails and as whiskers along the trailing ends of the rails (Figure 32). The whiskers contained only deformed particles, evidence of strong interaction between the particles and the interface. After flying for some time, large agglomerates of particles were occasionally found on the tapers. Since they Copyright © 1994 CRC Press, LLC
360
CRC Handbook of Lubrication and Tribology
FIGURE 32. (a) Trailing end of one rail of the slider comparison between lubricated and unlubricated disk in a contamination test. Whiskers are only grown on sliders flown on disks with liquid lubricant, (b) All polystyrene particles in the whiskers are deformed, evidence of interaction with the interface. (From Hiller, B. and Singh, G. P., Adv. Info. Storage Syst., 2, 173, 1991.)
contained mainly deformed particles, they could be identified to be whiskers which had detached from the trailing end (Figure 33). Since whiskers were grown on unlubricated disks, liquid lubricant promotes adhesion between particles and surfaces and may have an adverse effect on reliability with large amounts of contaminant particles.
Head-(Thin-Film) Rigid Disk Interface Magnetic films used for thin-film disks are soft and have poorer wear resistance than particulate disks loaded with hard magnetic particles and load-bearing alumina particles. Smoother surfaces and lower flying height of thin-film disks also result in higher friction and increased potential of head-to-disk interactions. During normal drive operation, the isolated asperity contacts of head and disk surfaces result in adhesive and impact wear and generate debris. In addition, any asperity contacts introduce maximum shear stress at the disk subsurface, which may initiate a crack. Repeated contacts would result in crack propagation (subsurface fatigue) leading to delamination of the overcoat and magnetic layer. Isolated contacts in a clean environment generate very fine wear debris which results in rather uniform disk wear from light burnishing. This wear eventually results in high friction and head crash.8 External contamination readily abrades the overcoat and results in localized damage of the disk surface by three-body abrasion. Wear debris invades the spacing between the head slider Copyright © 1994 CRC Press, LLC
Volume III
361
FIGURE 33. Dark-field photograph of the taper showing a well-defined uniform particle deposition pattern and a large agglomerate (top); SEM photographs of spherical particles within the uniform pattern (lower left) and of deformed particles in the agglomerate which is a former whisker from the trailing end (lower right). (From Hiller, B. and Singh, G. P., Adv. Info. Storage Syst., 2, 173, 1991.)
and the disk and/or transfers to the head slider making it unstable. This leads to additional debris and head crash both in the start/stop and fly ability modes. With alumina contaminant particles, Koka and Kumaran71 observed build up in the leading-edge taper of the slider, and abrasive wear on the disk resulted. Engel and Bhushan72 developed a head-disk interface failure model for thin-film disks based on topography of the wearing surfaces. Principal variables include sliding speed, surface topography, mechanical properties, coefficient of friction, and wear rate. Surface asperities and debris particles induce impact and sliding encounters, which represent a damage rate. Failure occurs when a specific damage rate, a characteristic for the system, is reached.
Role of Slider and Overcoat Materials — Figure 34 shows the coefficient of friction as a function of number of passes for a thin-film disk with carbon-overcoat disk B, sliding against various slider materials.73,74 Among ceramics tested, single-crystal diamond had the lowest coefficient of friction (~0.12) followed by partially stabilized zirconia (~0.15); the remaining ceramics all had an initial coefficient of friction of about 0.2. Although coefficient of friction increased with number of passes, this increase was small for single crystal diamond, even after 5500 passes. Rate of increase in coefficient of friction was highest for calcium titanate and Al2O3-TiC sliders; Mn-Zn ferrite and ZrO2-Y2O3 sliders exhibited a smaller increase. Calcium titanate (1200 kg/mm2) showed poor durability because it cracks readily. A12O3TiC is hardest (2300 kg/mm2) and burnished the disk more than Mn-Zn ferrite (600 kg/ mm2) Copyright © 1994 CRC Press, LLC
362
CRC Handbook of Lubrication and Tribology
FIGURE 34. Change of the coefficient of friction of five ceramic sliders while sliding against a thin-disk with a carbon overcoat and perfluoropolyether as the topical lubricant (disk B,) with number of sliding passes. Normal load = 150 mN, sliding velocity = 2.1 m/s. (From Chandrasekar, S. and Bhushan, B., J. Tribol., Trans. ASME, 113, 313, 1991.)
and ZrO2-Y2O3 (1300 kg/mm2). The air-bearing surface of the Mn-Zn ferrite slider was scratched, which suggests that Mn-Zn ferrite is slightly softer than the disk and is gentle to the disk surface. Matching of ceramic slider and disk hardnesses is essential for low wear. Increase in overcoat hardness with baked SiO2 improves the wear resistance of the thinfilm disks.35 Khan et al.75 reported that hard carbon overcoats with better wear performance consist of homogeneous grain size, uniform grain distribution, and higher percentage of sp3 bonded carbon atoms (diamond structure). Yamashita et al.76 reported that wear performance of unlubricated disks with ZrO2-Y2O3 overcoat is superior to that of carbon, while for lubricated disks ZrO2-Y2O3 and carbon overcoats are comparable. In a ceramic-ceramic contact, yttria-stabilized zirconia is known to have excellent friction and wear performance.77 ZrO2 overcoat (30 nm thick) exhibited better corrosion resistance than hard carbon when exposed to 80°C/90% RH for 7 days. Ceramic overcoats with low porosity and high electrical resistivity offer better electrochemical corrosion resistance.8 Calabrese and Bhushan78 conducted in-situ experiments of various head/film disk (95 mm dia.) combinations in the scanning electron microscope (see also Reference 79) to identify the initiation of particle removal. After sliding Al2O3-TiC at 500 mm/min. for a few minutes on a thin-film disk with a zirconia overcoat and perfluoropolyether topical lubricant (disk B2), microscopic particles were removed from the rail edges of the head and deposited on the disk (Figure 35). A little disk debris was also deposited on the rail edges. After 20 min, there was some damage to the slider edges and the disk surface was very lightly burnished with only one scratch. Minute disk debris was found on the rail edges and rail surfaces including the leading taper of the head slider (Figure 36). Continued sliding led to increased surface change of the disk, followed by catastrophic failure. The most significant parameter that influences the initiation of particle removal is the condition of the rail edges which contact the disk. During the start of motion, the head moves with the disk until the spring suspension overcomes adhesion between the head and disk. The head then springs back in an unstable manner, causing the rail edges to contact the disk, which results in transfer of material from the disk or chipping of the rail edge. Calabrese and Bhushan reported head slider and disk wear is strongly dependent on the slider and the disk overcoat materials (Table 5). Zirconia overcoat generally exhibited less wear than the carbon overcoat. Mn-Zn ferrite slider was less aggressive to the disk than the Al2O3-TiC slider. The calcium titanate slider cracked early on in a sliding test; hot processing of this material appears to be a problem. Copyright © 1994 CRC Press, LLC
Volume III
363
FIGURE 35. Appearance of the trailing edge of the Al2O3-TiC slider while sliding against a thin-film disk with a Zirconia overcoat (disk B). Lower micrograph shows a particle which is removed from the head and is sitting on the disk surface. This photograph was taken after 7200 mm of sliding at 50 ram/min. (From Calabrese, S. J. and Bhushan, B., Wear, 139, 367, 1990.)
FIGURE 36. Appearance of leading edge of the Al2O3-TiC slider after sliding against the disk B2. Wear Debris “attached to the ranges and the sides of the slider. (From Calabrese, S. J. and Bhushan, B., Wear, 139, 367, 1990.)
Copyright © 1994 CRC Press, LLC
364
CRC Handbook of Lubrication and Tribology
Role of Lubricant Film — The effect of lubricant viscosity and its thickness, lubricant functionality, and disk surface roughness on the durability were studied by Miyamoto et al.80 and Streator et al.39 Figure 37 shows the friction histories of unlubricated and lubricated disks with disk failure defined by a relatively sharp rise in the friction. The effect of surface roughness is summarized in Figure 38. The disk X2 demonstrates the lowest durability and indicates the effect of texturing as compared to high durability of the untextured disk X1. Textured disk B1 is from a different disk manufacturer, and cannot be compared to the other disks on the basis of roughness alone. Presence of the lubricant improved the disk durability over that of dry sliding in all cases. The polar lubricant has significantly higher durability man the nonpolar lubricant with comparable viscosity. A trend for greater durability with the less viscous lubricants can be attributed to their greater mobility on the disk surface. Figure 39 shows a general increase in durability on disk X2 with increasing lubricant film thickness. Hoshino et al.81 studied the effect of 40-d storage time on static friction for disks with a nonpolar PFPE lubricant film and disks with dual lubricant consisting of polar (aminosilane) and nonpolar (PFPE) fractions (Figure 40). Increase in friction from aging the disks with dual lubricant film was found to be less than that for a disk with only nonpolar lubricant. Lubricant is also spun off with disk rotation during use. Yanagisawa43 and others have shown that polar lubricants spin off less than nonpolar lubricants with disk rotation.8 The dual layer concept with an unbonded layer over a bonded layer is very useful because the unbonded (mobile) top layer would heal any worn areas on the disk surface where lubricant may have been removed, and the bonded layer provides lubricant persistence.
Role of Environment — Strom et al.82 tested unlubricated carbon-coated disks (disks B1) and ZrO2-Y2O3-coated disks (disks B2) in a sliding test against commercial Al2O3-TiC slider (Figures 41(a) and 41(b). The average friction of the carbon-coated disk increased smoothly only in the oxygen environment which indicates poor durability (see also References 83 to 86). At sustained high friction, debris was generated which reduced the real area of contact and the friction dropped. With zirconia-coated disk, no consistent difference was observed between various gas environments. In an additional sliding test on the carbon-coated disk in the presence of humid gases, the coefficient of friction increased smoothly to about 1.4 with all gases during the course of about 50 revolutions (Figure 41(c)). Increase in friction in the oxygen or humid environment for the carbon-coated disk can be explained by oxidation of the Copyright © 1994 CRC Press, LLC
Volume III
365
FIGURE 37. Friction histories during durability tests for unlubricated and lubricated thin-film disks with carbon overcoat sliding against Al203-TiC slider at 1.2 m/s with 3 nm of lubricant: (a) disk X1 roughness = 2.6 nm rms; (b) disk X2 roughness = 4.5 nm rms; and (c) disk B1 roughness = 5.2 nm rms. (From Streator, J. L. et al., J. Tribol., Trans. ASME, 113, 32, 1991.)
carbon film under rubbing, a tribochemical reaction. With the zirconia-coated disk, wear occurs through mechanical means only with no oxidation effects, regardless of the oxygen concentration. Marchon et al.87 also reported that carbon oxidation contributes significantly to increased wear and friction. In sliding experiments with Mn-Zn ferrite or calcium titanate sliders and unlubricated thin-film disks with carbon overcoats, the coefficient of friction gradually increased with repeated sliding contacts in air, in pure nitrogen however, the coefficient of friction remained constant at 0.2. Alternate introduction of oxygen and nitrogen elegantly showed the role of these gases (Figure 42). Contact start/stop tests exhibited the same effect. Wear by the slider in the oxygen environment appeared to involve oxygen chemisorption on the carbon surface and gradual loss of carbon through formation of CO/CO2. The disk wear was believed to be primarily on the asperities, resulting in a smoother surface and higher friction. Copyright © 1994 CRC Press, LLC
366
CRC Handbook of Lubrication and Tribology
FIGURE 38. Disk durability summarized for each of the sliding conditions of Figure 37. (From Streator, J. L. et al., J. Tribol., Trans. ASME, 113, 32, 1991.)
FIGURE 39. Disk durability as a function of lubricant film thickness on disk X2 (roughness = 4.5 nm rms) for selected lubricants. Data point at 0 nm is the durability for dry sliding. (From Streator, J. L. et al., J. Tribol., Trans. ASME, 113,32, 1991.)
In a study of a Mn-Zn ferrite pin with radius of 50 mm sliding against a thin-film disk with carbon overcoat and perfluoropolyether as the topical lubricant (disk B,), Dugger et al.88 found that the contact life, as marked by the total distance slid to the point at which the coefficient of friction increases rapidly over the steady state value, is much larger in air with 50% RH than in dry air or vacuum (Figure 43(a). Characteristics of post failure surfaces in vacuum and dry air are severe damage and roughening of the disk surface, with the pin from the vacuum test also exhibiting extensive damage, including intergranular fracture and grain pull-out. In both cases there is also material transferred from the disk surface to the pin. In humid air, however, the contact area on the pin is covered with very fine debris (about 1µm) particles in a dark film (low atomic number), with fine particles on either side of the worn area. The surface of these wear particles is enriched with cobalt from the magnetic layer. In dry air and vacuum, the Copyright © 1994 CRC Press, LLC
Volume III
367
FIGURE 40. Coefficient of static friction as a function of exposure at 50°C and 10-3 ton for lubricated spincoated SiO2 film on a sputtered oxide magnetic film. (From Hoshino, M. et al., in Tribology and Mechanics of Magnetic Storage Systems, Vol. 5, Bhushan, B. and Eiss, N. J., Eds., SP-25, STLE, Park Ridge, IL, 1988, 37.)
FIGURE 41. Coefficient of friction as a function of number of revolutions during sliding of Al2O3-TiC slider against an unlubricated thin-film disk at a normal load of 150mN and a sliding speed of 60 mm/s: (a) disk with carbon overcoat (B1 with no lubricant) in dry gases; (b) disk with zirconia coating; and (c) disk with carbon overcoat (B1 with no lubricant) in various gases, all at 4% RH. (From Strom, B. D. et al., J. Tribol., Trans. ASME, 113, 689, 1991.)
Copyright © 1994 CRC Press, LLC
368
CRC Handbook of Lubrication and Tribology
FIGURE 42. Coefficient of friction as a function of number of revolutions during sliding of Mn-Zn ferrite against an unlubricated thin-film disk at a normal load of 100 mN and a sliding speed of 60 mm/s in various gases. (From Marchon, B. et al., IEEE Trans. Magn., 26, 168, 1990.)
FIGURE 43. Coefficient of friction as a function of distance slid for hemispherical pins of Mn-Zn ferrite on lubricated thin-film disks in vacuum, dry air, and air with 50% RH, at 1 m/s sliding speed and 98 mN applied load: (a) lubricated thin-film disk with carbon overcoat (disk B1); (b) lubricated thin-film disk with zirconia overcoat (disk B2). (From Dugger, M. T. et al., in Advances in Engineering Tribology, STLE, Park Ridge, IL, 1991, 43.)
debris is substantially larger than or than one micron (Figure 44) and tends to be enriched with nickel (probably from the magnetic layer and Ni-P underlayer) on its surface. Two mechanisms appear to contribute to durability differences in humid air. oxidation of metallic wear debris generated at isolated asperity contacts alteration of the coating surface by interaction with vapor. The rate of metal debris oxidation affects the tendency to agglomerate through sintering or mechanical compaction in into larger particles which are more damaging.89 Significant adhesion in vacuum and less so in dry ait probably increases wear debris generation and, by mechanical compaction or otherwise, more ready agglomeration to large wear fragments that lead to catastrophic failure. Rapid agglomeration to particle sizes of greater than a few microns is believed responsive for the reduction in contact life in vacuum and dry air. However, in 50% RH air, less adhesion the interface and oxidation of metallic debris result in increased wear life. Wahl et al.90 and Copyright © 1994 CRC Press, LLC
Volume III
369
FIGURE 44. SEM micrograph of an isolated metallic wear particle on the pin from a test at 98 mN applied normal load, 1 m/s sliding speed in vacuum, showing evidence of agglomeration. (From Dugger, M. T. et al., J. Tribol, Trans. ASME, 112, 230, 1990.)
FIGURE 45. Summary of the contact lives as a function of environment for lubricated thin-film disks with carbon and zirconia overcoats (disk B1 and B2, respectively) at 1 m/s sliding and 98 mN load in a sliding test with hemispherical pins of Mn-Zn ferrite. Error bars represent the standard deviation from at least four experiments. Cross-hatched bar corresponds to zirconia overcoat.
Dugger et al.89 also found short contact life both for ultra-high-purity nitrogen and helium, in the same range as in vacuum (Figure 45). These data suggest that water vapor and oxygen in the humid air tests and oxygen in the dry air tests are responsible for the greater durabilities in these environments, while neither helium nor nitrogen plays a beneficial role for the durability of the rigid disks. To further explore the effects of water vapor, Wahl et al.90 performed durability testing, using a nitrogen ambient with humidities ranging from 0.2 to 80% (corresponding to partial pressures of water vapor from about 5 Pa to 2600 Pa at room temperature). Introduction of as little as 0.2% RH to the ambient dry nitrogen resulted in over two orders of magnitude increase in the contact life. The coefficient of friction for steady-state sliding in nitrogen was essentially independent of humidity at 0.15 to 0.18 up to about 50% humidity; increase in coefficient of friction to the 0.20 to 0.30 range at very high humidity (>80%) is believed to be due to liquidmediated adhesion as discussed earlier (see also Reference 83). Optimum humidity for maximum durability may depend on stress at the interface; the interface with a small stress Copyright © 1994 CRC Press, LLC
370
CRC Handbook of Lubrication and Tribology
(e.g., smaller slider) may show more sensitivity to the humidity than with high stress (large slider). Dugger et al.91 conducted a wear study on the thin-film disk with ZrO2-Y2O3 overcoat and perfluoropolyether lubricant disk (B2) (Figures 43(b) and 45). The coefficient of friction increased throughout the test to steady-state values between 0.6 and 1.4, depending upon the environment. (In contrast, the rapid increase in coefficient of friction for carbon overcoat occurred after significant sliding and was accompanied by wear track on the disk visible to the naked eye.) Even at high coefficient of friction, no visible wear track could be observed initially on the disk surface. When a wear track did become visible, an abrupt drop encountered in the coefficient of friction was attributed to generation of debris from the zirconia overcoat which reduced the real area of contact and, hence, the friction force. The similarity of contact lives of the zirconia overcoat and the extensive disk damage in vacuum and dry air suggest that oxygen does not significantly affect the wear rate of this material. In vacuum and dry air, the wear damage on the disk surface consisted of complete removal of the zirconia layer in some locations, with transfer of metals to the pin surface. Wear particles were frequently larger than 10 µm. The surface morphology of the damaged area was comparable to that of carbon overcoat. In humid air, the contact life was long; the disk surface appeared polished in the wear track compared to the surrounding regions and showed isolated areas of damage. This study on the zirconia overcoat suggested that the contact life is sensitive to the presence of water vapor. Dugger et al.91 conducted Auger depth profile analyses of the unworn surface, and the wear track formed on the thin-film disk (B1) in humid air stopped at 90% of the anticipated contact life. They found that the average carbon film thickness on the wear track was not very different from that on the untested region of the disk. Therefore, the majority of the film remains intact until very near the point at which the coefficient of friction increases dramatically above the steady-state value. Similar results have been reported by Wahl et al.92 for a vacuum environment. Thus, the precursor to failure is the catastrophic failure of carbon overcoat rather than uniform thinning of the overcoat. It is believed that debris generated at isolated points in the contact zone accumulated until a critical debris size or volume is generated, which results in catastrophic removal of the protective carbon film. It is speculated that wear of lubricated disk against a head slider (low interfacial stresses) rather than a pin (used in studies by Dugger) may be tribochemical rather than mechanical, similar to that found for unlubricated disks.
LUBRICATION
The primary function of the lubricant is to provide wear protection for the life of the magnetic medium and to ensure that friction remains low throughout the operation of the drive, which can be several years in duration. Among many requirements that must be satisfied is optimum lubricant thickness. If the lubricant film is too thick, excessive stiction and mechanical failure of the head/disk is encountered. If the film is too thin, interface protection is compromised and high friction and excessive wear will result in catastrophic failure. An acceptable lubricant must exhibit chemical inertness; low volatility; high thermal, oxidative, and hydrolytic stability; shear stability; and good affinity to the magnetic medium surface. Fatty acid esters are excellent boundary lubricants, and esters such as tridecyl stearate, butyl stearate, butyl palmitate, butyl myristate, stearic acid, and myristic acid are commonly used as internal lubricants, roughly 1 to 3% by weight of the magnetic coating for tapes and flexible disks. The fatty acids involved include those with acid groups with an even number between C12 and C22 with alcohols ranging from C3 to C13. The acids are all solids with Copyright © 1994 CRC Press, LLC
371
Volume III
melting points above the normal surface operating temperature of the magnetic media. This suggests that decomposition products of the ester formed during head-flexible medium contact may be a key to lubrication. Perfluoropolyethers (PFPEs) are the chemically most stable lubricants with some boundary lubrication capability and are most commonly used for topical lubrication to reduce wear of rigid disks.8 PFPEs commonly used include Fomblin Z and Fomblin Y lubricants made by Montiedison (Italy), Krytox 143 AD made by DuPont (U.S.), and Demnum made by Diakin (Japan), and their difunctional polar derivatives containing various reactive end groups, e.g., hydroxyl (Fomblin Z-Dol), piperonyl (Fomblin AM 2001), and isocyanate (Fomblin ZDisoc), all manufactured by Montiedison.93*96 Fomblin Z is a linear PFPE, and Fomblin Y and Krytox 143 AD are branched PFPE where the regularity of the chain is perturbed by -CF3 side groups. Bulk viscosity of Fomblin Y and Krytox 143 AD is almost an order of magnitude higher than the Z type. The molecular coil thickness is about 0.8 nm for these lubricant molecules; monolayer thickness depends on the molecular conformations of the polymer chain on the surface.97 Fomblin Y and Z are most commonly used for particulate and thin-film rigid disks. Usually lower viscosity lubricants (such as Fomblin Z types) are used in thin-film disks to minimize stiction.98,99
1. 2. 3. 4.
5. 6.
7.
8. 9. 10. 11. 12.
13. 14.
15.
16. 17.
18.
REFERENCES
Lowman, C. E., Magnetic Recording, McGraw Hill, New York, 1972. Hoagland, A. S., Digital Magnetic Recording, Wiley, New York, 1963. Camras, M., Magnetic Recording Handbook, Van Nostrand Reinhold, New York, 1988. Jorgensen, F., The Complete Handbook of Magnetic Recording, 3rd ed. Tab Books Inc., Blue Ridge Summit, PA, 1988. Mee, C. D. and Daniel, E. D., Eds., Magnetic Recording Handbook, McGraw Hill, New York, 1990. Van Gestel, W. J., Gorter, F. W., and Kuijk, K. E., Read-out of a magnetic tape by the magnetoresistive effect, Philips Tech. Rev., 37(2/3), 42, 1977. Iwasaki, S. and Nakamora, Y., An analysis for the magnetization mode for high density magnetic recording, IEEE Trans. Magn., Mag-13, 1272, 1977. Bhushan, B., Tribology and Mechanics of Magnetic Storage Devices, Springer-Verlag, New York, 1990. Wallace, R. L., The reproduction of magnetically recorded signal, Bell Syst. Tech. J., 30, 1145, 1951. Bhushan, B., Mechanics and Reliability of Flexible Magnetic Media, Springer-Verlag, New York, 1992. Nishihara, H. S., Dorius, L. K., Bolasna, S. A., and Best, G. L., Performance characteristics of IBM 3380K air bearing design, in Tribology and Mechanics of Magnetic Storage Systems, Bhushan, B. and Eiss, N. S., Eds., SP-25, STLE, Park Ridge, IL, 1988, 117. Bhushan, B., Bradshaw, R. L., and Sharma, B. S., Friction in magnetic tapes II. Role of physical properties, ASLE Trans., 27, 89, 1984. Bhushan, B., Sharma, B. S., and Bradshaw, R. L., Friction in magnetic tapes. I. Assessment of relevant theory, ASLE Trans., 27, 33, 1984. Bradshaw, R. L. and Bhushan, B., Friction in magnetic tapes, m. Role of chemical properties, ASLE Trans., 27, 207, 1984. Bradshaw, R. L., Bhushan, B., Kalthoff, C., and Warne, M., Chemical and mechanical performance of flexible magnetic media containing chromium dioxide, IBM J. Res. Dew, 30, 203, 1986. Bowden, F. P. and Tabor, D., Friction and Lubrication of Solids, Part I, Clarendon Press, Oxford, U.K., 1950. Greenwood, J. A. and Williamson, J. B. P., Contact of nominally flat surfaces, Proc. R. Soc. (Lond.), Ser. A, 295, 300, 1966. Bhushan, B., Analysis of the real area of contact between a polymeric magnetic medium and a rigid surface, J. Lubr. Tech., Trans. ASME. 106, 26, 1984.
Copyright © 1994 CRC Press, LLC
372
CRC Handbook of Lubrication and Tribology
19. Bhushan, B., The real area of contact in polymeric magnetic media. II. Experimental data and analysis, ASLE Trans., 28, 181, 1985. 20. Bhushan, B. and Doerner, M. F., Role of mechanical properties and surface texture in the real area of contact of magnetic rigid disks, J. Tribol., Trans. ASME, 111, 452, 1989. 21. Bhushan, B. and Blackman, G. S., Atomic force microscopy of magnetic rigid disks and sliders and its applications to tribology, J. Tribol., Trans. ASME, 113, 452, 1991. 22. Oden, P. I., Majumdar, A., Bhushan, B., Padmanabhan, A., and Graham, J. J., AFM imaging, roughness analysis and contact mechanics of magnetic tape and head surfaces, J. Tribol., Trans. ASME, 114,666, 1992. 23. Bhushan, B., Wyant, J. C, and Meiling, J., A new three-dimensional digital optical profiler. Wear, 122, 301, 1988. 24. Majumdar, A. and Bhushan, B., Role of fractal geometry in roughness characterization and contact mechanics of surfaces, J. Tribol., Trans. ASME, 112, 205, 1990. 25. Majumdar, A., Bhushan, B., and Tien, C. L., Role of fractal geometry in tribology. Adv. Info. Storage Syst., 1, 231, 1991. 26. Majumdar, A. and Bhushan, B., Fractal model of elastic-plastic contact between rough surfaces, J. Tribol., Trans. ASME, 113, 1, 1991. 27. Bhushan, B. and Majumdar, A., Elastic-plastic contact model of bifractal surfaces. Wear, 153, 53, 1992. 28. Berry, M. V. and Lewis, Z. V., On the Weierstrass-Mandelbrot fractal function, Proc. R. Soc, London Ser. A, 370, 459, 1980. 29. Mandelbrot, B. B., The Fractal Geometry of Nature. W. H. Freeman, New York, 1982. 30. Ling, F. F., The possible role of fractal geometry in tribology, Tribol. Trans., 32(4), 497, 1989. 31. Bhushan, B. and Dugger, M. T., Real contact area measurements on magnetic rigid disks, Wear, 137, 41, 1990. 32. Doan, T. Q. and Mackintosh, N. D., The frictional behavior of rigid-disk carbon overcoats, in Tribology and Mechanics of Magnetic Storage Systems, Vol. 5, Bhushan, B., and Eiss, N. S., Eds., SP-25, STLE, Park Ridge, IL, 1988, 6. 33. Matthewson, M. J. and Mamin, H. J., Liquid-mediated adhesion of ultra-flat solid surfaces, Proc. Mat. Res. Soc. Symp.. 119, 87, 1988. 34. Liu, C. C. and Mee, P. B., Suction at the Winchester head-disk interface, IEEE Trans. Magn.. Vol. Mag-19, 1659, 1983. 35. Yanagisawa, M., Tribological properties of spin-coated SiO2 protective film on plated magnetic recording disks, in Tribology and Mechanics of Magnetic Storage Systems, Vol. 2, Bhushan, B. and Eiss, N. S., Eds., SP-19, ASLE, Park Ridge, IL, 1985, 16. 36. Miyoshi, K., Buckley, D. H., Kusaka, T., Maeda, C, and Bhushan, B., Effect of water vapor on adhesion of ceramic oxide in contact with polymeric magnetic medium and itself, in Tribology and Mechanics of Magnetic Storage Systems, Vol. 5, Bhushan, B. and Eiss, N. S., Eds., SP-25, STLE, Park Ridge, IL, 1988, 12. 37. Bhushan, B. and Dugger, M. T., Liquid-mediated adhesion measurements at the thin-film magnetic disk/ head slider interface, J. Tribol., Trans. ASME, 112, 217, 1990. 38. Streator, J. L., Bhushan, B., and Bogy, D. B., Lubricant performance in magnetic thin film disks with carbon overcoat. I. Dynamic and static friction, J. Tribol., Trans. ASME, 113, 22, 1991. 39. Streator, J. L., Bhushan, B., and Bogy, D. B., Liquid performance in magnetic thin film disks with carbon overcoat. H. Durability, J. Tribol., Trans. ASME, 113, 32, 1991. 40. Tian, H. and Matsudaira, T., Effect of relative humidity on friction behavior of the head/disk interface, IEEE Trans. Magn., 28, 2530, 1992. 41. Baker, H. R., Bascom, W. D., and Singleterry, C. R., The adhesion of ice to lubricated surfaces, J. Coll. Sci., 17, 447, 1962. 42. Scarati, A. M. and Caporiccio, G., Frictional behavior and wear resistance of rigid disks lubricated with neutral and functional perfluoropolyethers, IEEE Trans. Magn., Vol. Mag-23, 106, 1987. 43. Yanagisawa, M., Lubricants on plated magnetic recording disks, in Tribology and Mechanics of Magnetic Storage Systems, Vol. 2, Bhushan, B. and Eiss, N. S., Eds., SP-19, ASLE, Park Ridge, Illinois, 1985, 7. 44. Bhushan, B., Magnetic head-media interface temperatures, II. Application to magnetic tapes, J. Tribol., Trans. ASME, 109, 252, 1987. 45. Bhushan, B., Magnetic head-media interface temperatures. III. Application to rigid disks, J. Tribol., Trans. ASME, 114,420, 1992. 46. Cook, N. H. and Bhushan, B., Sliding surface interface temperatures, J. Lubr. Tech., Trans. ASME, 95, 59, 1973. 47. Bhushan, B., Magnetic head-media interface temperatures. I. Analysis, J. Tribol., ASME, 109, 243,1987. 48. Gulino, R., Bair, S., Winer, W. O., and Bhushan, B., Temperature measurement of microscopic areas within a simulated head/tape interface using infrared radiometric technique, J. Tribol., Trans. ASME, 108, 29, 1986.
Copyright © 1994 CRC Press, LLC
Volume III
373
49. Bair, S., Green, I., and Bhushan, B., Measurements of asperity temperatures of a read/write head slider bearing in hard magnetic recording disks, J. Tribol., Trans. ASME, 113, 547, 1991. 50. Suzuki, S. and Kennedy, F. E., The detection of flash temperatures in a sliding contact by the method of triboinduced thermoluminescence, J. Tribol., Trans. ASME, 113, 120, 1991. 51. Bhushan, B., Assessment of accelerated head-wear test methods and wear mechanisms, in Tribology and Mechanics of Magnetic Storage Systems, Vol. 2, Bhushan, B. and Eiss, N. S., Eds., SP-19, ASLE, Park Ridge, DL, 1985, 101. 52. Chandrasekar, S., Shaw, M. C, and Bhushan, B., Comparison of grinding and lapping of ferrites and metals, J. Eng. Ind., Trans. ASME, 109, 76, 1987. 53. Chandrasekar, S., Shaw, M. C, and Bhushan, B., Morphology of ground and lapped surfaces of ferrite and metal, J. Eng. Ind., Trans. ASME, 109, 83, 1987. 54. Chandrasekar, S. and Bhushan, B., Control of surface finishing residual stresses in magnetic recording head materials, J. Tribol., Trans. ASME, 110, 87, 1988. 55. Chandrasekar, S., Kokini, K., and Bhushan, B., Influence of abrasive properties on residual stresses in lapped ferrite and alumina, J. Am. Ceramic Soc, 73, 1907, 1990. 56. Tanaka, K. and Miyazaki, O., Wear of magnetic materials and audio heads sliding against magnetic tapes, Wear, 66, 289, 1981. 57. Halm, F. W., Head wear as a function of isolated asperities on the surface of magnetic tape, IEEE Trans. On Magn., Mag-20, 918, 1984. 58. Kelly, J., Tape and head wear. Magnetic Tape Recording for the Eighties, NASA Ref. Publ. 1075, 1982, 7. 59. Lauer, J. L. and Jones, W. R., Friction polymers, in Tribology and Mechanics of Magnetic Storage Systems, Vol. 3, Bhushan, B. and Eiss, N. S., Eds., SP-21, ASLE, Park Ridge, IL, 1986, 14. 60. Ota, H., Namura, K., and Ohmae, N., Brown stain on VCR head surface through contact with magnetic tape, Adv. Info. Storage Syst., 2, 85, 1991. 61. Stahle, C. M. and Lee, T. D., Characterization of the deposits on helical scan heads, Adv. Info. Storage Syst., 4, 79, 1992. 62. Bhushan, B. and Phelan, R. M., Frictional properties as a function of physical and chemical changes in magnetic tapes during wear, ASLE Trans., 20, 402, 1986. 63. Calabrese, S. J., Bhushan, B., and Davis, R. E., A study by scanning electron microscopy of magnetic headtape interfacing sliding, Wear, 131, 123, 1989. 64. Bhushan, B., Nelson, G. W., and Wacks, M. E., Head-wear measurements by autoradiography of the worn magnetic tapes, J. Tribol., Trans. ASME, 108, 241, 1986. 65. Karis, T. E., Novotny, V. J., and Crone, R. M., Sliding wear mechanism of particulate magnetic recording media, in Tribology and Mechanics of Magnetic Storage Systems, Vol. 7, Bhushan, B., Ed., SP-29, STLE, Park Ridge, D., 1990, 35. 66. Novotny, V. J. and Karis, T. E., Sensitive tribological studies on magnetic recording disks, Adv. Info. Storage Syst., 2, 137, 1991. 67. Novotny, V. J., Karis, T. E., and Johnson, N. W., Lubricant removal, degradation, and recovery on particulate magnetic recording media, J. Tribol., Trans. ASME, 114, 61, 1992. 68. Kawakubo, Y., Ishihara, K., Seo, H., and Hirano, Y., Head crash process of magnetic coated disk during contact start/stop operations, IEEE Trans. Magn., Vol. Mag-20, 933, 1984. 69. Kawakubo, Y. and Seo, Y., Sliding failure mechanism of coated magnetic recording disks. Adv. Info. Storage Syst., Vol. 2, 73, 1991. 70. Hiller, B. and Singh, G. P., Interaction of contaminant particles with the particular slider/disk interface. Adv. Info. Storage Syst., 2, 173, 1991. 71. Koka, R. and Kumaran, A. R., Visualization and analysis of particulate buildup on the leading edge tapers on sliders, Adv. Info. Storage Syst., 2, 161, 1991. 72. Engel, P. A. and Bhushan, B., Sliding failure model for magnetic head-disk interface, J. Tribol., Trans. ASME, 112,299, 1990. 73. Chandrasekar, S. and Bhushan, B., Friction and wear of ceramics for magnetic recording applications. II. Friction measurement, J. Tribol., Trans. ASME, 113, 313, 1991. 74. Chu, M. Y., Bhushan, B. and De Jonghe, L., Wear behavior of ceramic sliders in sliding contact with rigid magnetic dun-film disks, Tribol. Trans., 35, 603, 1992. 75. Khan, M. R., Heiman, N., Fisher, R. D., Smith, S., Smallen, M., Hughes, G. F., Veirs, K., Marchon, B., Ogletree, D. F., Salmeron, M., and Seikhaus, W., Carbon overcoat and the process dependence on its microstructure and wear characteristics, IEEE Trans. Magn., 24, 2647, 1988. 76. Yamashita, T., Chen, G. L., Shir, J., and Chen, T., Sputtered ZiO2 overcoat with superior corrosion protection and mechanical performance in thin-film rigid disk application, IEEE Trans. Magn., Vol. Mag-24, 2629, 1988. 77. Bhushan, B. and Gupta, B. K., Handbook of Tribology: Materials, Coatings and Surface Treatments, McGrawHill, New York, 1991. 78. Calabrese, S. J. and Bhushan, B., A study of scanning electron microscopy of magnetic head-disk interface sliding, Wear. 139, 367, 1990.
Copyright © 1994 CRC Press, LLC
374
CRC Handbook of Lubrication and Tribology
79. Hedenqvist, P., Olsson, M., Hogmark, S., and Bhushan, B., Tribological studies of various magnetic heads and thin-film rigid disks, Wear, 153, 65, 1992. 80. Miyamoto, T., Sato, I., and Ando, Y., Friction and wear characteristics of thin-film disk media in boundary lubrication, in Tribology and Mechanics of Magnetic Storage Systems, Vol. 5, Bhushan, B. and Eiss, N. S., Eds., SP-25, STLE, Park Ridge, IL, 1988, 55. 81. Hoshino, M., Kimachi, Y., Yoshimura, F., and Terada, A., Lubrication layer using perfluoropolyether and aminosilane for magnetic recording media, in Tribology and Mechanics of Magnetic Storage Systems, Vol. 5, Bhushan, B. and Eiss, N. S., Eds., SP-25, STLE, Park Ridge, IL, 1988, 37. 82. Strom, B. D., Bogy, D. B., Bhatia, C. S., and Bhushan, B., Tribochemical effects of various gases and water vapor on thin film magnetic disks with carbon overcoats, J. Tribol., Trans. ASME, 113, 689,1991. 83. Dimigen, H. and Hubsch, H., Applying low-friction wear-resistant thin solid films by physical vapour deposition, Philips Tech. Rev., 41, 186, 1983/1984. 84. Doan, T. Q. and Mackintosh, N. D., The frictional behavior of rigid-disk carbon overcoats, in Tribology and Mechanics of Magnetic Storage Systems, Vol. 5, Bhushan, B. and Eiss, N. S., Eds., SP-25, STLE, Park Ridge, IL, 1988, 6. 85. Memming, R., Tolle, H. J., and Wierenga, P. E., Properties of polymeric layers of hydrogenated amorphous carbon produced by a plasma-activated chemical vapor deposition process. II. Tribological and mechanical properties, Thin Solid Films, Vol. 143, 31, 1986. 86. Miyoshi, K., Pouch, J. J., and Alterovitz, S. A., Plasma-deposited amorphous hydrogenated carbon films and their tribological properties. Tech. Memo 102379, NASA Lewis Research Center, Cleveland, Ohio, 1989. 87. Marchon, B., Heiman, N., and Khan, M. R., Evidence for tribochemical wear on amorphous carbon thin films, IEEE Trans. Magn., 26, 168, 1990. 88. Dugger, M. T., Chung, Y. W., Bhushan, B., and Rothschild, W., Friction, wear, and interfacial chemistry in thin-film magnetic rigid disk files, J. Tribol., Trans. ASME, 112, 230, 1990. 89. Dugger, M. T., Wahl, K. J., Chung, Y. W., Bhushan, B., and Rothschild, W., An Investigation of environmental effects on the wear and surface composition of dun-film magnetic disks, Advances in Engineering Tribology, Chung, Y. W. and Cheng, H. S., Eds., STLE, Parke Ridge, IL, 1991, 43. 90. Wahl, K. J., Chung, Y. W., Bhushan, B., and Rothschild, W. J., Durability of magnetic thin-film rigid disks in nitrogen and helium environments. Adv. Info. Storage Syst., 1, 327, 1991. 91. Dugger, M. T., Chung, Y. W., Bhushan, B., and Rothschild, W., Wear mechanisms of amorphous carbon and zirconia coatings on rigid disk magnetic recording media, Tribol. Trans., 36, 84, 1993. 92. Wahl, K. J., Chung, Y. W., Bhushan, B., and Rothschild, W. J., In situ-auger measurements of surface chemical changes of magnetic thin-film rigid disks during spherical pin sliding tests. Adv. Info. Storage Syst., 3, 83, 1991. 93. Cantow, M. J. R., Larrabee, R. B., Banall, E. M., Butner, R. S., Cotts, P., Levy, F., and Ting, T. Y., Molecular weights and molecular dimensions of perfluoropolyether fluids, Makromol. Chem., 187, 2475, 1986. 94. Corti, C. and Savelli, P., Perfluoropolyether lubricants: physical and tribological performances and applications, Proc. 5th Int. Congr. Tribology, Holmberg, K. and Nieminen, I., Eds., Vol. 5, Finnish Society for Tribology, Helsinki, Finland, 1989, 155. 95. Israelachvili, J. N., McGuiggan, P. M., and Homola, A. M., Dynamic properties of molecularly thin liquid films. Science, 240, 189, 1988. 96. Homola, A. M., Lin, L. J., and Saperstein, D. D., Process for bonding lubricant to a thin film magnetic recording disk, U.S. Patent 4,960,609, October 2, 1990. 97. Mate, C. M. and Novotny, V. J., Molecular conformation and disjoining pressure of polymeric liquid films, J. Chem. Phys., 94, 8420, 1991. 98. Bhushan, B., Magnetic recording surfaces, Characterization of Tribological Materials, Manning Publications and Butterworth Publishers, Stoneham, MA, 1993. 99. Bhushan, B., Tribology of magnetic head-medium interface, NATO ASI High Density Digital Recording, Series E: Applied Sciences, Kluwer Academic Publishers, Dordrecht, Netherlands, 229, 1993, pp. 281–314.
Copyright © 1994 CRC Press, LLC
COMPUTERS AND OFFICE MACHINES Raymond G. Bayer
A wide range of devices and machines may be included in the general category of computers and office machines. Examples of the diverse equipment involved are typewriters, word processors, copiers, personal computers, central processors or mainframes, credit card readers of various types, check sorters, printers, disk and tape files, keyboards, displays, and workstations. These machines perform a variety of tasks, involve different technologies, experience different amounts of utilization, and cover a wide range of cost, e.g., from a few hundred dollars to several million dollars. Three major subcategories are possible on the basis of their tribological characteristics.1,2 One category is that of magnetic storage and includes such things as disk and tape files and hard and floppy disk drives in personal computers. Another category is electronic packaging, which includes computers and the electronic packaging elements of the other machines. The third category, electromechanical devices, covers all the rest.
MAGNETIC STORAGE
One defining characteristic of devices in the magnetic storage category is concern with the friction and wear behavior of magnetic materials. Another is in terms of the effect of wear on performance in such devices. Detailed discussion of various aspects of tribology of magnetic storage devices can be found in the preceding chapter on magnetic storage systems and in the bibliography at the end of this chapter.1,3 With wear, the geometry and topography of the read/write heads change, as is illustrated in Figure 1. As a result, the ability of the head to read and write data, without error, degrades and ultimately may be lost entirely. However, the head can usually be replaced without the loss of stored information. With wear of the magnetic storage media (the counterface in such application) stored information may be permanently lost, since the information is stored in the magnetic media itself. Each wear fragment carries away with it a portion of the stored data and deformation may distort the shape of the magnetic bit. To increase storage capacity and reduce access time in these devices, it is necessary to decrease the size of the magnetic bit and to bring the head closer to the surface. The magnitude of these dimensions is typically much less than 25 µm (0.001 in.). Both of these aspects lead to concern with dimensions and amounts of wear smaller than those typical for the other types of office machines and computer equipment. This can be illustrated by the sample of high-speed disk drives. In these the heads fly over the surface, and critical separation is controlled by the air bearing that is generated between the head and the disk. In these applications, it is necessary to maintain the air film to avoid continuous sliding contact. At startup and slow-downs, as well as with random collisions, concern involves micron and submicron wear of both the head and the disk. The study of tribology, at this level, has been referred to as micro-tribology. The same level of sensitivity to magnitude of wear is not true for all magnetic storage applications, and several microns or more of wear might be tolerable. As a result, there is greater similarity from a tribological standpoint in these cases to many of the situations encountered in the other two categories. In particular, the problems and solutions associated with these less demanding magnetic storage systems are similar to those encountered with printer ribbons and paper. An example is the use of hard coating on selected areas of the head to control abrasive wear by the magnetic media (Figure 1). Shape and roughness of these Copyright © 1994 CRC Press, LLC
375
CRC Handbook of Lubricants and Tribology
376
FIGURE 1. Examples of magnetic head designs used to resist abrasive wear.
heads are also selected to minimize damage to the magnetic media as well.
ELECTRONIC PACKAGING
In this category, the primary wear situation is associated with electrical contacts.1,4.6 Typically, these are stationary contacts that experience a limited number of actuations over life. Examples of these types of contacts are shown in Figure 2. In applications which require high reliability, these contacts have noble metal surfaces, which are generally plated. A cross section of the metallurgy of such a contact member is also shown in Figure 2. In these applications, concern with wear is the exposure of base metal sublayers, which would allow corrosion to take place. Corrosion of the contact surface results in increased and unstable contact resistance and degraded connector performance. Electroplated gold (Au), palladium (Pd), and palladium-nickel (Pd-Ni) are commonly used as the noble metal in these types of contacts. In most cases, the contact is designed to achieve very high contact stress level, e.g., 1400 MPa (200 kpsi), resulting in plastic deformation, and to provide a desired minimum amount of wipe or sliding during actuation. This level of stress is used to insure a large real area of contact and, in conjunction with the wipe, to break through or displace any surface contamination. In some applications, lead (Pb)-coated contacts are used. In these cases, the high stress and wipe are required to break through and displace the Pb-oxide layer as well. Wear life of contacts using Pb coatings is generally less than those using noble metals, e.g., there are fewer insertions before wear through. In addition to wear, contact friction is often a concern with these types of contacts. In the case of external connections, a connector can involve as few as two or three individual contacts, or it might be in the range of 10 to 20. Internal connections might use a connector system that may involve several hundred contacts (see Figure 3). Particularly in the latter situations, insertion and disengagement of the connector can require significant force as a result of friction between the two halves of the contact. To reduce friction as much as possible, the contacts are frequently lubricated with a thin layer of nonconducting oil. In some cases, the oil may also be required to improve the wear performance. While these contacts generally experience a rather low number of actuation/deactuation cycles, e.g., <100, they can be exposed to other sources of wear, such as relative motion resulting from thermal expansion and contraction in the package and machine-induced or transmitted vibrations. Exposure to these types of motions can result in fretting-type wear. These fretting wear situations can often be significant, particularly when the connector has the dual purpose of providing an electrical path and a mechanical attachment between two packaging elements. Copyright © 1994 CRC Press, LLC
Volume III
377
FIGURE 2. Examples of electrical contact used in electronic packaging applications.
The example in Figure 4 shows a card edge connector which used to attach a card to a printed circuit board and provides a major portion of the mechanical forces holding the card. In this case, fretting can occur at the contact interfaces as a result of induced card vibrations. In addition to me concern with exposure of the base metal, there can be other concerns. When Pd-Ni or Pd is used as the contact material, contact performance can degrade as a result of friction polymer formation.7 The presence of a thin layer of Au on these materials or the use of an appropriate lubricant can inhibit the formation of the friction polymer. Pb-coated contacts generally perform poorly under fretting conditions. The fretting motion, which may break the stable Pb-Pb junctions formed by the insertion, introduces oxides to the contact region. Thermal expansion and contraction and machine vibrations can also lead to other wear problems in electronic packaging. These vary with the particular type of package but often nontypical wear materials and situations.8–9 Two examples of this type of wear are shown in Copyright © 1994 CRC Press, LLC
378
CRC Handbook of Lubricants and Tribology
FIGURE 4. Example of a computer card edge connector system used in an application which may expose it to fretting.
Figure 5 for a T(hermal) C(onduction) M(odule), (TCM). One example is wear of the Pb- and indium (In)-coated C-ring seals with possible development of leak paths from thermal cycles. The other example is wear between the thermal piston and the silicon chip surface with possible degradation of heat transfer across the interface as a result of the generation of wear debris. In both cases, wear was not directly related to the loss of material or change in dimension. Rather it was with the loss of function. As computer technology advances, other types of wear situations may also be encountered. For example, liquid impingement to cool chips and modules may introduce erosion and cavitation wear on the heat transfer surfaces of these devices. Friction and wear are also possible with repeated plugging and unplugging of fiber optic connections. Copyright © 1994 CRC Press, LLC
Volume III
379
FIGURE 5. Schematic of a thermal conduction module (TCM), showing wear points.
ELECTROMECHANICAL APPLICATIONS
In the electromechanical category, two types of wear situations tend to be unique to office machines and computers: (1) in the control and movement of paper, and (2) in printing. A third unique wear situation existed, viz., the punching and reading of punched cards, 20 years ago. In addition to the unique mechanisms and devices used to perform these functions, the machines in the electromechanical category utilize the typical spectrum of mechanical components, e.g., cams, rolling and sliding bearings, gears, latches, belts, pulleys, etc. In these applications, the primary function of these components is generally associated with the positioning and movement of key elements, such as a print hammer or a sheet of paper, not the transmission of power as might be the case with other types of machinery. As a result, the components used in office machines and computer peripherals are relatively small, and they often require higher quality, greater precision, and tighter tolerance than those associated with the transmission of power. In the electromechanical subcategory, primary concern with wear of components or parts is directly associated with the change in dimensions. In most cases, only relatively small changes can be tolerated, i.e., up to 100 microns (0.004 in.), and in many devices, changes as a result of wear must be kept below 50 microns (0.002 in.). Changes beyond these levels can result in the generation of poor quality documents, errors in reading information from a document, or errors in printing. This requirement, along with the desired or expected service life for such equipment, requires appropriate design limits on contact stresses and the use of a lubricant or selflubricating materials. For sliding contacts, contact stress levels are typically a small fraction of the strength of the wearing material, e.g., <0.5 x yield point. For rolling and impact situations, contact stresses can often be significantly higher, but again, within the elastic range of the wearing material. In general, environmental considerations are not a major factor in the wear of computer peripherals and office machines in mat extremely hostile conditions are not present. Typically, the machines are operated in office environments or, in many cases, in air-conditioned facilities. Internal machine temperatures are generally <40°C above room temperature. Because of these conditions, gaseous and particulate contamination is generally not a factor requiring special attention from either a testing or design standpoint. Normal laboratory Copyright © 1994 CRC Press, LLC
380
CRC Handbook of Lubricants and Tribology
conditions are fairly representative of these conditions and generally provide a suitable environment for wear evaluations. However, for some applications, such as workstations used on a manufacturing floor, the hostile environments may require special considerations in the evaluation of materials for use there. In addition to the characteristic requirements of tight tolerances and small wear that are associated with computer peripherals and office equipment, additional aspects affect the selection and use of various tribological components. One of these is noise and vibrations. Since these machines operate in “office environments”, they are expected to operate quietly, without vibration, and reliably over many years with little or no maintenance, much like a home appliance. Design procedures generally associated with selection and implementation of conventional tribological elements are appropriate for computer and office equipment, provided the specific application conditions are addressed. For example, motion of many bearings in this type of equipment is oscillatory or intermittent, not continuous rotation. This type of behavior has to be factored into the design relationships for roller and ball bearings, particularly in terms of the reduced effect that fluid film formation will have on wear performance under these conditions. Another example is selflubricated bearing systems, which are described by P-V (pressure × velocity) limits. In this case, the wear rate that is associated with those limits might be much higher than can be tolerated in devices which are very sensitive to wear. Then, modified P-V limits need to be used. Often special wear or life tests need to be conducted to determine the appropriate values of the various coefficients and limits in conventional P-V relations. A wide range of materials is used in these applications. Actual selection often depends on a number of factors other than wear. When plastics are used, these applications generally require the use of either plastics with good intrinsic wear properties or reinforced or filled grades. High surface hardness is usually required in most situations in which metals are used. This results in the use of stainless steels and steels which can be through or surfaced hardened and the use of hard coatings on materials which cannot be so treated. For ferrous materials, chromium (Cr) and electroless nickel (Ni-Ph) platings are often used for this purpose, as well as sprayed carbide coatings. The thickness of such coatings are typically in the range of 5 to 50 microns (0.0002 to 0.002 in.). In the case of aluminum alloys, hard anodized coatings are frequently used to provide adequate wear resistance. Further details regarding these aspects can be found in References 1,2, and 10 through 18.
PAPER HANDLING
Control and movement of paper in office machines and computer peripherals generally results in an exposure to abrasive wear. The method of addressing these wear situations is to use hard materials or coatings. Examples of some typical situations are shown in Figure 6. For high throughput machines, this generally means the use of materials and coatings with hardnesses greater than 700 kg/mm2 (Rc 60). For low throughput machines, wear rates associated with materials of lower hardnesses may be acceptable. In these applications, there is often the requirement to stop and start the motion of the paper. Elastomer rollers are often used for this purpose because of their high friction. The life of these elastomers is generally quite good and is satisfactory in these applications, provided the amount of sliding that occurs is small. Slip associated with acceleration and deceleration can greatly increase the amount of wear that occurs in such situations.
Copyright © 1994 CRC Press, LLC
Volume III
381
FIGURE 6. Examples of components exposed to abrasive wear by paper in office and computer applications.
PRINTING
At present, the other unique situation associated with computers and office equipment is that associated with printing, particularly impact printing. Several types of printing, including nonimpact and impact methods, are shown in Figure 7. With nonimpact printing, concern with wear is either eliminated or reduced to the abrasive wear of the print element by the paper, such as in electroerosion printing. In the case of impact printing, wear is often a major concern with many elements of the mechanism. In these cases, there are often several wear points involved, as illustrated in Figure 8, and these are associated with both sliding and impact motions. In addition, fatigue can also be a major consideration with some of these elements. Consequently, it is often necessary to develop a component that has good resistance to several types of wear and that also provides good fatigue resistance. The print band in Figures 7 and 8 is an example where fatigue is of concern as a result of its bending around pulleys. The band also has to have good sliding wear resistance against the platen and good abrasive wear resistance against both the ribbon and guiding surfaces. If a friction drive is used to move the band, friction can also be a factor. The velocities in these applications can be quite high, e.g., 90 kmph (103 in./sec), and peak impact forces can exceed 103 N (200 lb). Print elements can be expected to withstand in excess of 108 impacts in some high-speed impact printers. When sliding is combined with impact, hard, tough materials are often used with lubrication to provide adequate life. Because of the combined requirements for different elements of a part, coatings of various types are often used in these applications. Cr and Ni-Ph platings and carbide deposits are often used. For wear points involving only impact, elastomers are frequently used and provide good life. In the units providing the energy for printing, e.g., the magnet assembly, secondary fretting motions result from the vibrations induced by printing impacts. Life of elastomers used in such applications can be greatly reduced by these fretting motions. Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
382
Figure 7. Different methods of printing used in office and computer applications.
Copyright © 1994 CRC Press, LLC
Volume III
383
FIGURE 8. Schematic of a high-speed band printer, showing wear points in the print area.
In printing there is concern not only with wear of the print mechanism but also with wear of the paper and ribbon. Individual fibers of the ribbon can be damaged and the weave can be distorted as a result of printing. Paper can be cut or embossed, as well. The geometry and edges of the print element are major factors in determining life of the ribbon and amount of damage to the paper. High pressures and sharp corners can lead to embossing of the paper and flattening and tearing of ribbon fibers and weave. These aspects become increasingly important as speed of the printers increases. This type of damage is controlled by insuring proper edge conditions on the print element, e.g., engraved or etch type faces or print wires. At the same time, abrasivity of the paper and the ribbon can significantly influence life of the print element. LUBRICATION
Lubrication in both the electromechanical and the packaging categories involves some unique requirements.1,19–21 An overriding requirement is that the lubricant cannot get onto the documents processed by the machine or onto any external surfaces of the machine. In the case of electrical contacts, a small amount of lubricant is initially applied. Usually this is done by dipping, spraying or brushing the lubricant to one member of the contact and removing any excess. These methods of application result in the formation of a thin lubricant layer, e.g., <40 micron (0.0015 in.), which is expected to last the lifetime of the package (5 to 10 years). Thin-film lubrication and the requirement of initial lifetime lubrication are also typical of electromechanical applications. However, in this case, the thin film in the active region is often maintained by flow from a reservoir. Several types of lubrication systems for these applications are shown in Figure 9. The oil migrates along surfaces, as well as through or Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
384
FIGURE 9. Methods of lubrication in office and computer equipment.
from porous materials, to resupply the contact region. With ball and roller bearings and gears, typical initial lubrication techniques are generally used but are required to provide lifetime lubrication. This ability to provide continuous lubrication over the desired lifetime is often the primary selection factor for the lubricant, rather than any unique antiwear or friction characteristics. A range of lubricants are used in these applications, as shown in Table 1. Copyright © 1994 CRC Press, LLC
Volume III
385
REFERENCES 1. Engel, P. A. et al., Review of wear problems in the computer industry, J. Lubr. Tech., 4/78, 100, 189, 1978. 2. Bayer, R. G. and Trivedi, A. K., Wear testing for office and data processing equipment, ASTM STP 615, American Society for Testing and Materials, Philadelphia, 1976, 91. 3. Bayer, R. G., A model for wear in an abrasive environment as applied to a magnetic sensor. Wear, 70, 93, 1981. 4. Bayer, R. G. et al., A note on the application of the stress dependency of wear in the wear analysis of an electrical contact, Wear, 7, 282, 1964. 5. Hsue, E. S. and Bayer, R. G., Metallurgical study and tribological properties of edge card connector spring/tab interface, Proc. 34th IEEE Holm Conf. on Electrical Contacts, September, 1988. 6. Bayer, R. G. et al., A motion induced sub-surface deformation wear mechanism, in Proc, Int. Wear of Materials Conf. 1991, Vol. 1, American Society of Mechanical Engineers, New York, 1991, 489. 7. Bare, J. P. and Graham, A. H., Connector resistance to failure by fretting and frictional polymer formation, in Proc. Holm Conf., IEEE, 147–155, 1985. 8. Bayer, R. G., Influence of oxygen on the wear of silicon, Wear, 69, 235, 1981. 9. Bayer, R. G., Wear of a C-ring seal, Wear, 74, 339, 1981–1982. 10. Bayer, R. G., Wear life prediction in an engineering environment, in Proc. First European Tribology Conf., London, paper #C2 1967/1973. 11. Payne, N. and Bayer, R. G., Friction and wear tests for elastomers. Wear. 150, 67, 1991. 12. Bayer, R. G., Impact wear of elastomers, Wear, 112, 105, 1986. 13. IBM J. Res. Dev., 25(5), 755, 1981. 14. Bayer, R. G., Wear in electroerosion printing, Wear, 92, 197, 1983. 15. Bayer, R. G. and Sirico, J. L., Influence of jet printing inks on wear, IBM J. Res. Dev., 22(1), 90, 1978. 16. Engel, P. A. and Bayer, R. G., Abrasive impact wear of type, J. Lubr. Tech., 98, 330, 1976. 17. [Issue on ink jet printing], IBM J. Res. Dev., Vol. 21(1), 1977. 18. Bayer, R. G. et al., Wear and fatigue analysis in mechanical printing, in Proc. 2nd Int. Computer Conf., August 15 to 19, 1982. 19. Bayer, R. G. et al., Coolant contamination of sintered bronze parts. Wear, 140, 165, 1990. 20. Bayer, R. G. et al., Oil films on electroplated gold contacts, Trans. ASME, J. of Electronic Packaging, 236/111,9, 1989. 21. Bayer, R. G. et al., Evaluation of grease performance in a fluorocarbon atmosphere, Lubr. Eng., 44(10), 866, 1988.
Copyright © 1994 CRC Press, LLC
REFRIGERATION AND AIR CONDITIONING Glenn D. Short
INTRODUCTION
Refrigeration compressor lubrication encompasses several application areas, including industrial refrigeration, air conditioning, heat pumps, home appliances, and food distribution. The lubricant in the many types of compressors in these systems is required to provide many years of service while reducing friction, preventing wear, and removing heat in compressor components. It also acts as a sealing aid for compression. Matching a compressor and its system with a refrigeration oil for optimum performance is difficult and commonly involves special refining or blending of several products and the use of additives.1 Historically highly refined and dewaxed mineral oils, naphthenic or paraffinic, have been used. More recently, various synthetic lubricants have become available which offer a wide range of properties and the opportunity to customize a lubricant for a particular refrigeration system. Lubricant selection and performance is also influenced by the type of refrigerant. Refrigerants include ammonia, hydrocarbons, and halocarbons which are generally characterized by their content as: chlorofluorocarbons (CFC), hydrochlorofluorocarbons (HCFC), and hydrofluorocarbons (HFC).
REFRIGERATION COMPRESSOR LUBRICATION
Theory of compressor lubrication as it applies to refrigeration, air conditioning, and heat pump applications has been well defined.1–5 The primary function of the lubricant is to provide lubrication to the working parts of the compressor. Depending on lubricant pressure, velocity, and viscosity, the mode of lubrication changes but is generally elastohydrodynamic or hydrodynamic. Compressor start-up, shutdown, and overload conditions may also require good boundary lubrication properties.2 A second function is to provide efficient sealing of compression cylinders in positive displacement compressors (except dry types). Another function is to act as a heat transfer medium to help remove heat from friction and compression of the gas. Hermetic drive motors located in a common chamber with the compressor where they are cooled by the refrigerant require special consideration. Motor materials must not be affected by either the refrigerant or the lubricant.
Reciprocating Compressors Major lubrication points include cylinders, valves, pistons, piston rings, crankshafts, connecting rods, main and crank pin bearings, and other associated parts. Double-acting machines (and some single-acting machines) use crosshead and crosshead guides with connecting pins to join the crosshead to connecting rods. Most single-acting one- and twostage compressors are designed with the connecting rods connected directly to the pistons with wristpins or piston pins. Main, crankpin, and crosshead (or wristpin) bearings; crossheads and crosshead guides; and other crankcase components require relatively large amounts of lubricant, which is usually supplied from the crankcase. Several types of systems are used to disperse the oil, depending on compressor design, bearing loads, and application. These include “splash”-utilizing dippers in the oil, “flooded” with oil lifted by devices such as disks, screws, grooves, or oil ring gears, and pressurized forced-feed systems. 0-8493-3903-0/94/$0.00+$.50 © 1994 by CRC Press. Inc.
Copyright © 1994 CRC Press, LLC
387
388
CRC Handbook of Lubrication and Tribology
Cylinder lubrication should be maintained with the minimum amount of lubricant needed to minimize wear and friction; to seal piston rings, valves, and rod-packings; and prevent corrosion. Lubricant that is supplied to the cylinder is carried out to the valves and discharge piping where it eventually enters the refrigeration system.
Rotary Screw Compressors Lubricants in oil-flooded refrigeration screw compressors are generally used for sealing of rotors; cooling; and lubrication of rotors, bearings and shaft seals. In the twin-screw design, refrigerant gas is drawn in and trapped in a cell formed by two intermeshing rotors and the cylinder wall. Lubricant is injected to remove heat and to seal the clearances. Compression results as the main rotor lobe rolls into the secondary rotor groove. The gas is discharged as the end of the female rotor passes over the outlet port. The lubricant is injected directly into refrigerant gas at the beginning of the compression cycle through orifices in the cylinder wall. Relatively large amounts of lubricant mix intimately with the refrigerant gas to adsorb heat of compression. This controls discharge temperatures more effectively than with water jacket cooling. The lubricant is circulated from the discharge side reservoir through a heat exchanger to provide a cool oil supply. An oil/gas separator reduces oil carry over into the system to about 10 to 20 parts per million. Selection of a lubricant depends on many factors, including compressor design (materials, rotor profile, torque, transmission, and bearing type), type of refrigerant, and compressor operating conditions, such as pressure and temperature. Additionally, me location of the lubricant reservoir on the high pressure (compressor discharge) side introduces larger amounts of refrigerant into the lubricant as compared to systems where the oil reservoir is on the low pressure side. This introduces a greater concern for maintaining adequate viscosity and lubrication properties in the presence of the refrigerant to prevent mechanical wear. Higher viscosity lubricants are used with lower rotor tip speeds to improve sealing and compression efficiency for the action between the two intermeshing rotors. Liquid-injected single-screw compressors use the intermeshing of a single worm type screw with one or two flat gate rotors in a casing to provide suction, compression, and discharge of refrigerant gas. The screw and casing combine to act as a cylinder and the gate rotor(s) as a piston. Gas volume is reduced to achieve compression as the gate rotor advances in the groove of the screw rotor. Nearly all the compression (and thrust load) is supplied by the screw. Bearings may be lubricated by grease or fluid, depending upon design.
Rotary Vane Compressors Two common types of small positive displacement vane compressors are used for refrigeration compressors—fixed vane and rotating vane. The fixed vane type uses a ring or roller which rotates around an eccentric shaft. A single vane is mounted in a nonrotating cylinder housing. The rotating vane compressor has a rotor concentric with the shaft and offcenter with respect to the cylinder housing. The rotor is equipped with radially sliding vanes which are forced against the cylinder walls by centrifugal force. Rotating vane design is more common in larger compressors, particularly for booster compressors in refrigeration systems. The lubricant in rotary vane compressors assists in providing a seal between the vanes and the cylinder (or ring) wall as gas is compressed. Larger systems may use oil pumps. Adequate lubrication should be provided to the vanes, vane slots, bearings, and shaft seals. Many newer compressors using composite vanes require only a small amount of lubricant to be mechanically fed to the cylinder during start-up. Additional oil to the cylinders may be supplied later from the bearing lubricant discharge.
Copyright © 1994 CRC Press, LLC
Volume III
389
FIGURE 1. Equipment diagram for basic vapor compression cycle.
Scroll Compressors The scroll compressor, a special type of rotary positive displacement compressor, is largely used for air-conditioning applications. The basic compression unit is a set of two scrolls, one fixed and the other moving in a controlled orbit around a fixed point. Areas of lubrication include a short throw crank mechanism, bearings, and the scroll tip.
Centrifugal Compressors Centrifugal compressors used in refrigeration are generally multistage machines handling high volumes of vapor at relatively low pressures. Major areas requiring lubrication are bearings, gear reducers, and seals. The lubrication system, gears, bearings, and motor for these compressors can be internal or external, depending on whether or not they operate in the refrigerant atmosphere.6 In some units the oil reservoir is vented to the refrigerant system. An internal oil pump, driven by either an internal motor or the compressor shaft, supplies oil to bearings. Auxiliary oil pumps may be used to provide lubricant for start-up. Generally the lubricant from centrifugal compressors is not expected to leave the compressor to enter condensers and evaporators.
THE REFRIGERATION CYCLE
Figure 1 shows the basic principle of a compression refrigeration system.7 Refrigerant initially leaves the evaporator (1) as a low pressure, low temperature, saturated gas. It is drawn into the compressor, where it is compressed reversibly and adiabatically (isentropic). It then leaves the compressor at a higher pressure (2), as superheated vapor. Heat is removed at constant pressure in the condenser (with air or water cooling) and the gas is desuperheated and liquified. The liquid refrigerant collects in a receiver (which may be provided with a drain to remove oil) and subsequently passes at high pressure (3), to an expansion valve. The expansion valve meters refrigerant as low pressure, low temperature liquid (4) and enters the evaporator. Excessive valve flow may pass liquid refrigerant to the compressor and result in lubrication problems. Too little flow of refrigerant will result in ineffective use of the evaporation surface, and may increase the power demand of the compressor. Copyright © 1994 CRC Press, LLC
390
CRC Handbook of Lubrication and Tribology
The evaporator boils the refrigerant to extract heat from the surrounding environment. There are two general types: “flooded”, and “dry” or “direct expansion” (DX). A flooded evaporator maintains a constant level of liquid refrigerant (and oil) within the evaporator. Liquid refrigerant passes through an expansion valve to the dry evaporator in a fixed amount so that all liquid is converted to gas before it leaves the suction end to the compressor.8 Evaporator temperature is determined by the pressure (also called the suction pressure) and type of refrigerant.
LUBRICANT PERFORMANCE PROPERTIES
Most lubricant-related problems occur in a refrigeration compressor when the lubricant viscosity becomes too low or there is absence of oil.10 Improper viscosity can result from any of the following: selection of a lubricant with too low viscosity; over-dilution of the lubricant by refrigerant (solubility); abnormal increase in lubricant temperature, or lubricant breakdown due to chemical or thermal instability. Loss of lubricant can be the result of several mechanisms. Foaming of dissolved refrigerant in the oil reservoir can interfere with the oil pump delivery of oil to the compressor. Surface tension of the oil and refrigerant mixture can also affect both foaming tendency and the ability of the oil to stay between sliding surfaces.9,11 A crankcase heater may be necessary to prevent excessive amounts of refrigerant vapor from dissolving or condensing into the lubricant during shutdown. Crankcase heaters also prevent the formation of two liquid phases in compressors when systems are shut down and the compressor is cooled to low temperatures. Formation of two-phase oil-refrigeration mixtures can cause mechanical failures, through foaming or rapid gas release due to a sudden decrease in pressure and/or increase in temperature during start-up.2,12 Another cause of oil starvation is oil accumulation outside the compressor in the condenser, piping, and the evaporator.
Thermal and Chemical Stability Thermal stress causes unstable oils to form carbonaceous deposits which cause wear and other compressor-related failures. Related problems may also result from breakdown of insulation materials. Cleanliness is essential both for maintenance of heat transfer efficiency and for functioning of mechanical components such as expansion valves. Oil volatility can also create problems. Compression temperatures in refrigeration compressors can reach 160°C (320°F). Any relatively volatile oil fractions can strip away at these higher temperatures to result in lubricant thickening in the compressor. The lighter fractions carry out into the compressor system as a vapor (even in systems with sophisticated oil separation) where they condense and reduce heat transfer efficiency. The lubricant should be chemically resistant to reactions with the refrigerant to avoid sludge, coking, and copper plating. Enhanced oxidative stability may be required for systems with suction pressures below atmospheric pressure or where air is inadequately purged.
Solution Behavior Miscibility characteristics of liquid refrigerant with oil are illustrated in Figure 2.12 Area A represents a single liquid phase. Area D, below the curve, has two phases. As the temperature is lowered, increasing viscosity of the oil-rich phase may introduce a possible problem with collection on evaporator tube walls. Under dynamic conditions, the oil-rich phase is usually carried along with the gas or liquid refrigerant allowing normal operation at temperatures well below the separation of phases. Heat transfer and oil return problems may occur when the oil-rich phase, being of lower density, floats on top of the refrigerant-rich phase. Heat transfer problems are more frequent Copyright © 1994 CRC Press, LLC
Volume III
391
FIGURE 2. Miscibility diagrams.
in systems where the oil is immiscible or partly miscible with the refrigerant. While residual oil is often detrimental, small amounts of oil (1%) may result in improved heat transfer. This is thought to be related to foaming in the evaporator and a resulting improvement in heat transfer coefficient.2 Any wax contained in mineral oils must also be considered with those halocarbon refrigerants which tend to have a dewaxing effect. At low temperatures, wax (and some types of additives) tends to coat out in evaporators, on float valves, in capillary expansion devices, and on other system components causing mechanical and flow problems. Low temperature viscosity and miscibility are considered together in designing evaporators and oil return systems. Direct-expansion type evaporators operate on the principle that a refrigerant-rich liquid phase enters the evaporator and gradually changes to an oil-rich phase before exiting. As the amount of refrigerant dissolved in oil increases, viscosity is lowered and oil return is improved. As oil concentration of the liquid phase increases, two phases may occur, as indicated in area D of Figure 2. The more viscous oil rich-phase can cling to the walls of the evaporator while the less viscous refrigerant-rich phase continues to flow. Under this circumstance, the thermal expansion valve (or other liquid control) responds to the refrigerant-rich phase and eventually causes an accumulation of the oil-rich phase.13 There are different ways to overcome this situation. Most obvious is to use an oil that will not undergo phase separation at the evaporator temperature. Another is to design for a high enough vapor velocity to carry along the oil-rich phase for proper operation of the liquid control device. A third is the use of lubricants with good low temperature fluidity which are independent of miscibility for good oil return. Flooded evaporator systems operate with oil concentration in the evaporator maintained so that the oil is miscible at the evaporator temperature. Some of the oil-refrigerant mixture is sent through a heat exchanger to evaporate the refrigerant and return the oil to the compressor. Otherwise, some arrangement must be made to remove the insoluble layer of oil. In cases where the density of the oil-rich phase is less than that of the refrigerant-rich phase Copyright © 1994 CRC Press, LLC
392
CRC Handbook of Lubrication and Tribology
(as with halocarbons), some type of oil-skimming device may be used. In other cases where the oil is more dense than the refrigerant phase (such as ammonia), the oil may be drained from the bottom of the evaporator. In either case, lower viscosity facilitates oil removal.
Typical Recommended Viscosity Ranges Selection of the viscosity grade for refrigeration oils depends upon refrigerant type, the degree of solubility with the lubricant, and the compressor and system design. Table 1 identifies some typical viscosity ranges.
EVALUATION METHODS
Many evaluation methods for refrigeration oils are common to other types of lubricants (Table 2).14 Additional standards may be applicable which include ISO standards (International Standards Organization) and the German DIN 51 503 minimum requirements for refrigerator oils. Some of the tests are specific to mineral oils or apply for use with specific refrigerants. “Aromatic content” applies only to mineral oils or aromatic type synthetic fluids (alkyl Copyright © 1994 CRC Press, LLC
Volume III
393
benzene). “Floe point” measures the precipitation characteristics (wax) of an oil cooled at a specific rate with a 9:1 mixture of CFC-12 and oil. While “aniline point” is used as an indirect measure of the effect of an oil on elastomers, immersion tests with elastomers are preferred. Low temperature characteristics of the lubricant may affect the refrigeration system efficiency. Generally, pour point is not critical in systems that use flooded evaporators. More important is the oil fluidity (viscosity) in the presence of the refrigerant. Water content is measured for lubricants which are to be used in halocarbon applications, especially for those systems which include hermetic motors. The maximum water content of mineral oils and synthetic hydrocarbon oils is often specified at 30 parts per million. It may not be possible to maintain water content at this low level with other types of synthetic oils due to their hygroscopic nature. Polyglycols often are supplied with a specification of less than 200 ppm, esters at less than 50 ppm. These types of oils often rely on the removal of excess moisture through the use of system filter driers. Lubricity Lubricity may be measured either on the refrigeration oil itself or in combination with the refrigerant. For evaluation in boundary lubrication, standard four-ball, pin and vee block, pin on disk, and ring and block wear tests are most commonly used. These tests have been modified to measure the effects of a refrigerant environment.15,16 Compressor manufacturers utilize bench tests and field tests to confirm laboratory observations. Sealed Tube Stability Test methods are available for CFC refrigerants which expose test oil to the refrigerant at elevated temperatures for a specific test duration. The condition of the oil, catalyst, refrigerant, and presence of breakdown products are used to predict compressor performance. Generally a sealed glass tube containing the lubricant, refrigerant, and metal catalyst(s) is aged at elevated temperatures for a specified time. The most widely used analytical technique is gas analysis.17,18 Coking, copper plating, and corrosion may be visually observed. These tests are also useful in observing effects on elastomers and hermetic motor materials. A variance on this procedure, PHIUPP (DIN 51 593), exposes heated oil (250°C) in one leg of a sealed glass U-tube and cooled refrigerant (-40°C) in the other leg. Oil condition is usually observed throughout the test and refrigerant and oil deterioration are evaluated upon completion. Similar pressurized bomb tests may be used for lubricant and refrigerant combinations with materials used to remove moisture from refrigeration systems (i.e., molecular sieves, activated alumina). The pressurized bomb method has been used in ammonia refrigeration applications to observe carbon formation and corrosion on steel panels.19 Solubility and Miscibility Solubility of a gaseous refrigerant with the lubricant is dependent on temperature, pressure, type of refrigerant, and the lubricant. Calculations and graphical techniques for determining solubility generally follow Henry’s Law. More accurate techniques for estimating solubility have been developed and correlate well with experimental data.20 Solubility of refrigerants in lubricants has been experimentally measured using a device similar to that shown in Figure 3.20 The technique involves exposure of known amounts of lubricant and refrigerant at various temperatures and then calculating the amount of dissolved refrigerant. Miscibility measurements may be made with known amounts of refrigerant and lubricant weighed into sealed glass tubes. This is done at very low temperatures so that the refrigerant Copyright © 1994 CRC Press, LLC
394
CRC Handbook of Lubrication and Tribology
FIGURE 3. Experimental solubility apparatus: sample cylinder.
is in a liquid phase. The amounts are predetermined so that there is limited vapor space. The liquid mixture is observed at various temperatures. When separation is observed, the mixture has exceeded its miscibility limit and the temperature is recorded. Viscosity Viscosity data are obtained for refrigeration lubricants by standard test methods (ASTM D-445) and on viscosity and solubility of oil/refrigerant pairs.21 Recently, new procedures have been developed for a wider range of temperatures and pressures.22 Sampling methods and density-measuring devices may be included to obtain complete information on concentration, viscosity, and density at various temperatures and pressures.
LUBRICANTS
Mineral Oils Highly refined mineral oils, similar to white oils, have been most commonly used for refrigeration applications. These petroleum-base lubricants may vary widely in their physical properties, chemical structure, degree of refining, and performance. They are broadly characterized through their traditional classifications as “paraffinic” or “naphthenic”. Waxes (linear paraffins) are removed from refrigeration oils during the refining process either by solvent dewaxing or catalytic dewaxing. Branched chain paraffins and cycloparaffins have good viscosity retention at higher temperatures and better low temperature fluidity, good chemical stability, high viscosity, and lower volatility. These oils are less miscible with polar refrigerants. They may be identified by their high aniline point, low specific gravity, low refractive index, and higher molecular weight. “Naphthenic refrigeration oils” contain higher levels of unsaturated aromatic molecules. Careful solvent extraction or hydro-finishing removes the more unstable aromatics, unsaturates, and otherwise undesirable components. Copyright © 1994 CRC Press, LLC
Volume III
395
Synthetic Oils Use of synthetic oils for refrigeration was first proposed in 1929 to solve problems with mineral oils such as wax precipitation, low miscibility with some refrigerants, and carbonization of valves in reciprocating compressors.23,24 Additional advantages for some synthetics include improved stability in the presence of refrigerants at high temperatures, better viscosity/ temperature characteristics resulting in improved hydrodynamic lubrication of compressor bearings, and better lubricity in the presence of refrigerants. Table 3 provides a general comparison of some commonly used petroleum and synthetic refrigeration lubricants. Each category of lubricant represents a broad class of base fluids. For example, polyol esters may be derived from various alcohols (pentaerythritol, neopentyl alcohol, etc.) and a wide range of acids. Likewise, not all naphthenic mineral oils are satisfactory for refrigeration applications. Blends Synthetic oils are sometimes blended with mineral oils. Such blends use a synthetic oil that is soluble with the mineral oil (for example, alkyl benzene). Synthetics are sometimes blended together to improve their overall performance. Generally, blended lubricants are supplied as a finished compressor lubricant, but they may also be blended by the compressor operator by adding two types of lubricants to the compressor. Additives Additives are sometimes used in refrigeration lubricants. The most common types include: stability improver, lubricity aid, and antifoam agent. The need for additives depends on the lubricant, refrigerant, and equipment design. Their use should follow only after a thorough review and lubricant qualification procedure.1
LUBRICANT APPLICATIONS AND PRACTICES
Until the 1990s the predominant types of refrigerants included ammonia, hydrochlorofluorocarbon, and chlorofluorocarbon. Ammonia is used in large commercial Copyright © 1994 CRC Press, LLC
396
CRC Handbook of Lubrication and Tribology
FIGURE 4. Miscibility characteristics for various fluids with HCFC-22.
applications such as cold storage warehouses, fishing boats, ice rinks and many specialized applications. The CFCs and HCFCs have been extensively used in commercial plants, air conditioners, heat pumps, and in domestic appliances. Hydrofluorocarbon (HFC) refrigerants with zero ozone depletion potential are replacing CFCs and HCFCs in some applications. Propane and other hydrocarbon refrigerants are used by refineries and chemical plants where such materials are common and thus their flammability control is not an issue. Carbon dioxide may still be found in some older plants. Sulfur dioxide has generally been replaced by HCFCs.
Chlorofluorocarbons and Hydrochlorofluorocarbons Paraffinic mineral oils are most often used in positive displacement compressors in air conditioning applications with CFC-12. Naphthenic oils are more miscible with polar refrigerants such as HCFC-22 than are paraffinic oils. Paraffinic mineral oils may be used when they are not completely miscible with more polar refrigerants, provided they have adequate low temperature fluidity. Precautions taken to insure proper oil return from the system may include the use of high efficiency separators at the compressor discharge, oilskimming devices, and related system design modifications. The most common synthetic oils in use with HCFCs, such as HCFC-22, are the alkyl benzene hydrocarbons. These are similar to wax-free aromatic-naphthenic mineral oil which has been sufficiently refined for good chemical stability in refrigeration systems. Major advantages of these synthetics are improved miscibility (Figure 4) compared to mineral oils. These oils are available in ISO viscosity grades ranging from 22 to 100, and their degree of miscibility decreases with increasing viscosity. Alkylbenzenes are sometimes blended with naphthenic mineral oils or polyalphaolefins. Polyalphaolefins (PAOs) and polyalkylene glycols (PAG) have been extensively used in rotary screw refrigeration compressors. Improved adiabatic efficiency of 3 to 10% may be achieved in these compressors when compared to naphthenic refrigeration oils. This efficiency improvement is largely attributed to higher viscosity under dilution and high temperature conditions. Higher viscosity grades, not available with naphthenic oils, permit higher Copyright © 1994 CRC Press, LLC
Volume III
397
FIGURE 5. Viscosity characteristics with HCFC-22 (20% wt).
efficiencies through improved sealing. These same properties have contributed to their use in industrial heat pumps using CFC-12 or CFC-114.2–4 Lower temperatures in dry expansion evaporators are permissible with PAOs than with mineral oils. Low temperature fluidity is the major reason PAOs have been used in the U.S. for relatively insoluble refrigerants such as CFC-13 and CFC-503.4 Generally, ISO viscosity grades 15 or 32 are selected for operating viscosity in the compressor and fluidity below 73°C (- 100°F) in direct expansion dry-type evaporators. High viscosity modified complex esters (to ISO viscosity grade 320) have shown excellent miscibility with HCFC and efficient compression in rotary screw compressors. High viscosity compensates for the effect of the dissolved refrigerant while maintaining good oil return (Figure 5).25 Silicate esters (secondary butyl polysilicate acid esters)26 and, more recently, neopentyl polyol esters have been used for low temperature applications to provide miscibility with polar refrigerants such as CFC-13 and CFC-503. Oil return has been achieved at temperatures below - 100°C (-148°F) in flooded evaporators. One problem with the silicate ester is that it may form sludge if used after exposure to moisture. Both esters tend to be somewhat hygroscopic and unstable in the presence of moisture. Lubricants for HFC Applications HFC-134a (CH2FCF3), one alternative for CFC-12, is highly insoluble and nonmiscible with conventional mineral oils, alkyl benzenes, polyalphaolefins, and most other common refrigeration lubricants. Miscible types of polyalkylene glycols (Figure 6) have shown promise in small reciprocating compressors with HFC-134a.27,28 While lower molecular weight (lower viscosity) polyglycols have good miscibility, higher viscosity grades, above ISO 100, tend to be less miscible and less soluble at higher temperatures. End groups (hydroxyl) and internal structure (propylene or ethylene) of the lubricant molecule have an influence on miscibility.29 In some cases, these oils are mixed with halogenated lubricants or have fluorine directly added to their
Copyright © 1994 CRC Press, LLC
398
CRC Handbook of Lubrication and Tribology
FIGURE 6. Miscibility of polyalkylene glycols with HFC-134a.
FIGURE 7. Miscibility of esters with HFC-134a. Oil A = ISO 32 polyol ester; Oil B = ISO 22 polyol ester; Oil C = ISO 32 polyol ester; Oil D = ISO 116 proprietary ester; Oil E = ISO 32 diester.
structure to improve their miscibility.30,31 Polyglycols may be blended with other types of synthetic such as esters.32 Higher viscosity grade polyglycols provide an efficient operating viscosity in the presence of the refrigerant for sealing during compression and bearing lubrication. Automotive air conditioning applications with HFC-134a have utilized ISO viscosity grades of 46 to 150;33,34 150 and 220 grades have provided improved efficiency in rotary screw compressors, even when compared to CFC-12.4,35,37 Certain types of esters have good miscibility with HFC-134a (Figure T).28,36,38,39 Copyright © 1994 CRC Press, LLC
Volume III
399
FIGURE 8. Influence of oil film in ammonia brine chiller.
Compatibility of lubricants with hermetic motor insulation components and wire enamels is an issue with polyglycols. Other problems with polyglycols may involve their tendency to absorb water and degradation of the lubricant through use or storage. Polyglycols and esters may be used with other types of HFC refrigerants such as: HFC152a, HFC-32, HFC-125a, HFCs with ether components, and blends of these with more environmentally safe HCFCs. Alkyl benzene lubricants are miscible with certain more environmentally safe chlorofluorocarbons (HCFC), such as HCFC-124 and blends of HCFC with hydrofluorocarbons.40.41
Ammonia Most refrigeration grade mineral oils provide acceptable performance in ammonia systems. Selection of one oil over another is generally based on improved efficiency and reduced maintenance costs. The greatest opportunity for improved efficiency involves oil in the evaporator. Mineral oils have a low degree of miscibility with ammonia, and most applications are with flooded evaporators. The oil is more dense than ammonia and can be drained from the bottom of the evaporator. The problem arises with reduced heat transfer when oil adheres to heat exchange surfaces as illustrated in Figure 8 for a brine chiller.42 There are several ways to improve heat transfer in these systems. A properly sized oil separator can eliminate all liquid oil from the compressor discharge gas, and oil with a low vapor pressure and low foaming tendency will reduce the amount of oil that passes through oil separators and reaches the condenser and evaporator. Lower miscibility with ammonia at a higher condensing temperature will allow more oil to be removed from the bottom of the condenser (Figure 9).36 Good low temperature fluidity will assist oil return (removal) from the evaporator. Since ammonia has a high adiabatic compression temperature,43 poorly refined oils can easily carbonize or produce varnish to cause problems with operation of discharge valves in reciprocating compressors. The constant recirculation in rotary screw compressor also requires a chemically and thermally stable lubricant. Water and oxygen (common contaminants) in ammonia systems can produce nitrogen compounds and acids which deteriorate the oil and generate sludge and deposits. It is common to use additives in lubricants in ammonia applications to improve their stability.44.45 Copyright © 1994 CRC Press, LLC
400
CRC Handbook of Lubrication and Tribology
FIGURE 9. Miscibility of ISO 68 refrigeration oils with ammonia
Extreme low pour point oils such as PAO-type synthetic fluids or catalytically dewaxed paraffinic oils are generally used for evaporator temperatures below - 40°C (-40°F). Caution should be taken before using other types of synthetic lubricants. Esters react with ammonia to produce very viscous liquid or solid materials.
Propane and Other Hydrocarbon Refrigerants Hydrocarbon refrigerants, such as propane, are highly soluble in mineral oils. Dissolved hydrocarbon gas in compressor oil results in lower viscosity and reduced oil film thickness, and washing away or absorption of lubricant into the gas phase in the compressor may result in loss of lubrication. These problems have led to widespread use of polyglycols which are more “resistant to dilution” by hydrocarbons (lower solubility with these gases) than mineral oil. The polar nature of polyglycols helps to wet lubricated surfaces preferentially. Their high viscosity index also helps to seal compressor cylinders at high temperatures while maintaining the lower viscosity needed for good oil return from the low temperature side of the refrigeration system. Rotary screw compressors have shown improvements in volumetric efficiency up to 18%, when polyglycols are compared to mineral oils in propane compression. High pressure reciprocating compressors have benefitted by up to a 20-fold increase in the life of pressure packings, while cylinder oil feed rates have been reduced. Generally, the heavier PAG is returned from the bottom of evaporators to the compressor system. Typical polyglycol miscibility with hydrocarbons is shown in Figure 10. Mineral oils, synthetic hydrocarbons, and esters are highly soluble with hydrocarbon refrigerants. Higher viscosity grades must be used to maintain adequate lubricant films. Lubricants with low volatility, synthetic hydrocarbon and polyol esters, have been used in very low temperature systems, such as -118°C (-180°F) ethylene. Separators can effectively control the amount of lubricant in the low temperature side of the system.
LUBRICANT MAINTENANCE PRACTICES
Equipment manufacturers recommend periodic checking of oil levels and oil pressure and inspection for seal leakage. The following discussion is intended to provide a familiarization with maintenance practices rather than a specific guideline. Copyright © 1994 CRC Press, LLC
Volume III
401
FIGURE 10. Solubility of propane in mineral oil and polyalkylene glycol.
Contaminants in refrigeration systems represent a major cause of lubricant failure, compressor durability problems, and reduced system performance. Several contaminants reported in the literature46 are summarized in Table 4. Moisture should be eliminated prior to system start-up through the use of evacuation. If moisture is found in operating systems, a service consultant may recommend installation or replacement of a filter/drier. Large amounts of water in the system may have to be removed through partial disassembly and/or drainage. Filter driers, or desiccants, remove water through adsorption or react chemically with water contained in the refrigerant. Filter driers are often designed to adsorb or chemically remove other contaminants, such as acids and oil deterioration products, and mechanically remove solids. They may, depending on type, also remove oil additives. These devices are carefully designed for moisture capacity, refrigerant (and flow), and particle size. The adsorption of other contaminants may interfere with the drier capacity to remove water. Compressors are often equipped with suction side strainers and oil line filters to remove particles before they can damage cylinder and bearing surfaces. Compressors can often tolerate small amounts of particles that pass through these filters. Excessive amounts are removed by changing the oil. Compressor or motor failures require that the compressors, and sometimes the system, be cleaned using special measures.46 In the case of hermetic motor burnout, extremely high temperatures severely deteriorate motor insulation, lubricant, and the refrigerant. Excessive amounts of acids, carbonaceous sludge, water, and other deposits must be removed. Selection of a good grade of refrigeration lubricant recommended for the type of refrigerant being used, following good maintenance practices and using a lubricant analysis program for early detection of problems, will result in many years of successful compressor operation.
VISCOSITY-SOLUBILITY CHARTS
Selected viscosity solubility charts appear for reference in Figures 11,47 12,48 13,49 and 14.49 Copyright © 1994 CRC Press, LLC
402
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
Volume III
FIGURE 11. Viscosity-temperature-pressure chart for HFC-134a and 300 SUS polyalkylene glycol.
Copyright © 1994 CRC Press, LLC
403
404
CRC Handbook of Lubrication and Tribology
FIGURE 12. Viscosity-temperature-pressure chart for HFC-134a and 300 SUS polyol ester.
Copyright © 1994 CRC Press, LLC
Volume III
FIGURE 13. Viscosity-temperature-pressure chart for HCFC-22 and 150 SUS alkyl benzene.
Copyright © 1994 CRC Press, LLC
405
406
CRC Handbook of Lubrication and Tribology
FIGURE 14. Viscosity-temperature-pressure chart for HCFC-22 and 150 SUS naphthenic mineral oil.
Copyright © 1994 CRC Press, LLC
Volume III
REFERENCES
1.
2.
3.
4. 5. 6.
7.
8. 9.
10.
11.
12. 13.
14.
15.
16.
17.
18.
19.
20. 21.
22. 23.
24. 25.
26.
27.
28. 29. 30.
407
Spauschus, H. O., Evaluation of lubricants for refrigeration and air conditioning compressors, ASHRAE J., 26(5), 59, 1984. Kruse, H. H. and Schroeder, M., Fundamentals of lubrication in refrigeration systems and heat pumps, ASHRAE J., 26(5), 5, 1984. Daniel, G., Anderson, M. J., Schmid, W., Tokumitsu, M., Performance of selected synthetic lubricants in industrial heat pumps, J. Heat Recovery Syst., 2(4), 359, 1982. Short, G. D., Synthetic lubricants and their refrigeration applications, Lubr. Eng., 46(4), 1990. Refrigeration: systems and applications, in ASHRAE Handbook, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Atlanta, 1990, 1. Equipment, in ASHRAE Handbook, American Society of Heating, Refrigerating and Air-conditioning Engineers, Atlanta, 1988, 35. Fundamentals, in ASHRAE Handbook, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Atlanta, 1989, 8. Reciprocating Refrigeration, The Trane Company, LaCrosse, WI, 1986. Reese, L., Accessible hermetic compressors. Warmepumpen-Grundlagen-Komponenten-Auslegung-Bau und Betrieb, Vulkan-Verlag Essen, Germany, Ed. 1, 1978, 100. Burkhardt, J. and Hahne, E., Surface tension of refrigeration oils, IIR—Commissions B1, B2, E1, E2, Mons, Belgium, 1980, 111. Steinle, H., Oberflachenspannung von Kaltemitteln, Kaltemaschineolen und deren Gemischen, Kaltetechnik, 12(11), 334, 1960. Bosworth, C. M., Predicting the behavior of oils in refrigeration systems, Refrig. Eng., 160(6), 617, 1952. Soling, S. P., Oil recovery from low temperature pump recirculating halocarbon systems, ASHRAE Trans., 1971. ASHRAE Standard 99–1987, American Society of Heating, Refrigerating, and Air-Conditioning Engineers, Atlanta, GA, 1986. Huttenlocker, D. F., A bench scale test procedure for hermetic compressor lubricant, ASHRAE J., 11(6), 85, 1969. Sanvordenker, K. S., Laboratory testing under controlled environment using a falex machine, Proc. 1974 Purdue Compressor Tech. Conf., Lafayette, IN, 1974, 67. Spauschus, H. O. and Doderer, G. C, Reaction of refrigerant-12 with petroleum oils, ASHRAE J., 2, 65, 1961. Sanvordenker, K. S., Mechanism of oil-R12 reactions—The role of iron catalyst in glass sealed tubes, ASHRAE Trans., 91(1), 1985. Short, G. D., Hydrotreated oils for ammonia refrigeration, Tech. Pap., 7th Annu. Meet., Int. Inst. Ammonia Refrigeration, March 1985, 149. Glova, D., High-temperature solubility of refrigerants in lubricating systems, ASHRAE J., 26(5), 59,1984. Spauchus, H. O. and Speaker, L. M., A review of viscosity data for oil-refrigerant solutions, ASHRAE Trans., 93(2), 1987. Van Gaalen, N. A., Pate, M. B., and Zoz, S. C, The measurement of solubility and viscosity of oil/ refrigerant mixtures at high pressures and temperatures. Test facility and initial results for R-22/naphthenic oil mixtures, ASHRAE Trans., 96(2), 1990. Sanvordenker, K. S. and Larime, M. W., A review of synthetic oils for refrigeration use, ASHRAE Trans., 78(2), 1972. Shoemaker, B. H., Symposium, Synthetic lubricating oils, Ind. Eng. Chem., 42(12), 2414, 1959. Sjöholm, L. I. and Short, G. D., Twin screw compressor performance and complex ester lubricants with HCFC-22, Proc. Int. Compressor Engineering Conf., Purdue University, Lafayette, IN, July 1990. Löffler, H. J., The mixing abilities of the synthetic oils Fluisil S55K and Polyran M15 with refrigerants R-22, R-13 or mixtures of R-22 and R-13, Kältetech.-Klim., 5, 135, 1957. Vineyard, E. A., Sand, J. R., and Miller, W. A., Refrigerator-freezer energy testing with alternative refrigerants, Seminar 89–01 Ozone/CFC-CFC Alternative Studies, ASHRAE Annu. Meet., June 24 to 28, 1989, Vancouver, Canada; CFCs Time of Transition, American Society of Heating, Refrigerating and AirConditioning Engineers, Atlanta, 1989, 205. Sanvordenker, K. S., Materials compatibility of R-134a in refrigerant systems. Seminar 89–01 Ozone/ CFCCFC Alternative Studies, ASHRAE Annu. Meet., June 24 to 28, 1989, Vancouver, Canada; CFCs Time of Transition, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Atlanta, 1989,211. Magid, H. et al., Refrigeration Lubricants, U.S. Patent No. 4,755,316, July, 1988. Thomas, R. H. P. et al., Fluorinated Lubricant Compositions, U.S. Patent No. 4,975,212, December 1990
Copyright © 1994 CRC Press, LLC
408 31. 32.
33. 34.
35.
36. 37.
38.
39.
40.
41.
42.
43.
44.
45. 46.
47. 48.
49.
CRC Handbook of Lubrication and Tribology Thomas, R. H. P. et al., Lubricant Enabling Substitution of R12 by R134a—in Compression Refrigeration Comprising Dingle Phase Blend of Poly:oxy:alkylene Glycol and Chloro:tri:fluoroethylene Oil, U.S. Patent No. 4,900,463, February, 1990. McGraw, P. W. and Ward, E. L., Lubricants for Refrigeration Compressors, U.S. Patent No, 4,851,144, July, 1989. El-Bourini, R., Hayahi, K. and Adachi, T., Automotive air conditioning system performance with HFC134a refrigerant, SAE Tech. Pap. No. 900214, Int. Congr. Exposition, Detroit, MI, February/March, 1990. Struss, R. A., Henkes, J. P., and Gabbey, L. W., Performance comparison of HFC-134a and CFC-12 with various heat exchangers in automotive air conditioning systems, Pap. No. 900598, SAE Int. Congr. Exposition, Detroit, MI, February/March, 1990. Sjöhlm, L. I. and Short, G. D., Twin screw compressor performance and suitable lubricants with HFC-134a, Proc. Int. Compressor Engineering Conf., Purdue University, Lafayette, IN, July, 1990. Short, G. D. and Cavestri, R. C, Selection and performance of synthetic and semi-synthetic lubricants for use with alternative refrigerants in refrigeration applications, Proc. ASHRAE-Purdue CFC Conf., Purdue University, Lafayette, IN, July, 1990. Short, G. D., Rotary Displacement Compression Heat Transfer Systems Incorporating Highly Fluorinated Refrigerant-Synthetic Oil Compositions, U.S. Patent No, 4,916,914, April, 1990. Jolly, W. T., New unique lubricants for use on compressors utilizing R-134a refrigerant, Proc. ASHRAEPurdue CFC Conf., Purdue University, Lafayette, IN, July, 1990. Kaimai, T., Refrigeration oils for alternative refrigerants, Proc. ASHRAE-Purdue CFC Conf., Purdue University, Lafayette, IN, July, 1990. Reed, P. R. and Spauchus, H. O., HCFC-124: applications, properties, and comparison with CFC-114, ASHRAE, J., 32(2), 40, 1991. Bateman, D. J. et al., Refrigeration blends for the automotive air conditioning aftermarket, SAC Tech., Pap. 900216, Int. Congr. Exposition, Detroit, MI, February/March, 1990. Briley, G. C, Lubricant (oil) separation, Tech. Pap., 6th Annu. Meet., Int. Inst. Ammonia Refrigeration, San Antonio, TX, 1984, 107. Systems, ASHRAE Handbook, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Atlanta, 1984, chap. 29. Landry, J. F., The effects of contaminants on compressor oils in ammonia systems, Tech. Pap., 9th Annu. Meet., Int. Inst. Ammonia Refrigeration, March 29 to April 1, 1987, San Diego, CA, 213. Short, G. D., Refrigeration lubricants update: synthetic and semi-synthetic oils are solving problems with ammonia and alternative refrigerants, Tech. Pap., 12th Annu. Meet., Int. Inst. Ammonia Refrigeration, March 4 to 7, 1990, Memphis, TN, 19. Refrigeration—Systems and Applications, ASHRAE Handbook, American Society of Heating, Refrigerating, and Air-Conditioning Engineers, Atlanta, 1990, 7.1. Thomas, R. H. P., Wu Wei-Te, and Pham, Hang, The solubility and viscosity of mixtures of R-134a with modified polyglycols, XVIII Congress International Du Froid, XVIIIth International Congress of Refrigeration, Montreal, 1991. Short, G. D. and Cavestri, R. C, 1991 High Viscosity Ester Lubricants for Alternative Refrigerants, ASHRAE Annu. Meet., 98, Part 1, 789–795, January 1992, Anaheim, CA. Van Gaalen, N. A. and Pate, M. B., Method of measuring the solubility and viscosity of lubricating oil/ refrigerant mixtures at high discharge pressures and temperatures, ASHRAE Res. Proj. RP. 580, College of Engineering, Department of Mechanical Engineering, Iowa State University, Ames, 1991.
Copyright © 1994 CRC Press, LLC
OIL MIST LUBRICATION Donald M. Bornarth
INTRODUCTION
Oil mist lubrication was developed in the late 1930s by a European bearing manufacturer. In 1949, a major U.S. manufacturer of lubrication equipment purchased the rights to oil mist and soon began to offer systems to lubricate a greater variety of machines than the original high-speed grinders. Today, few machine elements cannot be oil mist lubricated, and few industries have not used oil mist systems. In this automatic centralized system, compressed gas is used to atomize fresh, clean oil and to continuously convey it, as a “dry” mist, to multiple and often widespread points of lubrication. In addition to the improvements in safety, productivity, housekeeping, and lubrication that are achieved by automatic centralized systems in general, continuous delivery eliminates the over-lubrication that is necessary to insure adequate supply between periodic applications. Lubricant consumption is usually reduced, sometimes by as much as 80%. Continuous delivery often permits elimination of energy-wasting oil sumps. Reductions in power consumption of more than 25% are not rare. Bearing temperatures are often lowered dramatically, not by actual cooling, but because most of that power consumption reduction represents heat that is not generated in churning excess lubricant. Pressurization of housings with continuous outward air flow helps to exclude dirt and corrosive atmospheres. Oil mist is even used to protect equipment on standby and in storage. Oil mist is safe. There are no high pressures involved. Lubricants used are not toxic. Since the ratio of oil to air is far below the lean limit of flammability, oil mist will not burn.
HOW OIL MIST WORKS
The heart of the system is the mist generator (Figure 1), consisting of an oil reservoir and a mist head, with a pressure regulator to control the inlet air supply. In the mist head, air flow, usually from a plant compressed-air system, is directed through a venturi or a vortex and accelerated to around sonic velocity. The resultant low pressure draws oil up from the reservoir into the high velocity air stream to produce very small droplets or particles. Baffles intercept the larger particles and drop them back to the reservoir. What remains is a smoke-like suspension of minute oil particles in air. The particles are liquid oil; oil mist does not involve an evaporation-condensation process. With such small particles, generally having diameters less than 6 or 7 microns (0.00024 or 0.00028 inch), surface tension is high compared to the mass. Particles must impact at relatively high velocities to break this surface tension and wet a surface—hence, the term “dry mist”. The mist is distributed from the generator through pipes, tubes, and hoses, sized to limit flow velocity to the laminar range so particles travel straight through the conduit without striking anything hard enough to stick or “wet out”. At a velocity around 24 feet per second (fps) there is a fairly sharp transition to turbulent flow. Particles then follow erratic paths with undesirable wet-out in distribution lines. Even with laminar flow, some minor wet-out results along conduit walls and from fallout of larger particles that escaped the mist generator baffles. The mist flow is metered to each lubrication point by one of several types of application fittings, the main ones being mist, spray, and condensing. Since mist fittings only meter flow, they depend on turbulence in and around rolling element bearings to cause oil to wet out directly onto the bearing elements. 0-8493-3903-0/94/$0.00 + $.50 © 1994 by CRC Press, Inc.
Copyright © 1994 CRC Press, LLC
409
CRC Handbook of Lubrication and Tribology
410
FIGURE 1. Basic oil mist generators.
Spray and condensing fittings not only control flow but also “reclassify” the “dry” mist into larger droplets. Spray fittings expel oil as fine, wet sprays. Condensing fittings reclassify a higher percentage of the oil from the mist into coarser, wet sprays or larger drops which drip or run down adjacent surfaces. The terms “reclassifiers” or “mist fittings” are sometimes applied to all types of application fittings.
APPLICATION OF OIL MIST—GENERAL
Application procedures are based on standard mist densities, or oil/air ratios, generally around 1/2 in.3 of oil per hour per scfm (standard cubic feet per minute) of mist flow. With standard mist densities, lubricant flow can be expressed in units other than actual amount of lubricant per unit of time. The older bearing inch system relates oil mist requirements and deliveries to those for a single-row rolling element bearing on a one inch diameter shaft. Bearing inches (BI) equals shaft diameter, in inches, multiplied by the number of rows of rolling elements. Although intended for use with rolling element bearings, formulas have been developed to express oil mist requirements of other types of machine elements in bearing inches. The newer and more widely used system expresses lubrication requirements and equipment ratings in scfm (standard cubic feet per minute) of mist flow. Since the scfm method can be related to what happens throughout the mist system, it will be used for the rest of this chapter.
CALCULATING MIST REQUIREMENTS
The first step in designing an oil mist system is identifying and describing the machine elements to be lubricated and calculating the mist flow required by each. Formulas for mist flow requirements vary somewhat from one manufacturer to another, depending on their generator characteristics. They assume that the elements to be lubricated were properly selected for the intended service, properly assembled, and protected from contamination. They also assume use of an oil with the proper misting and lubricating qualities for the application.
APPLICATION FITTING TYPES
The next step is the selection of application fittings types, fittings placement, and venting provisions for housed elements. Copyright © 1994 CRC Press, LLC
Volume III
411
Fittings type selections are primarily based on the types of elements to be lubricated, but are often influenced by other factors such as machine configuration or speed. Typically, spray fittings are used for rolling motion elements: antifriction bearings, gears, chains, etc. Condensing fittings or spray fittings are used for sliding motion elements such as plain bearings, slides, and ways. Mist fittings are used only for rolling element bearings in closed housings, under particular types of loading, and above a minimum speed. At speeds above about 200 linear feet per minute (lfm), at the mean path diameter (d) of the balls or rollers (lfm = rpm X πd [inches]/12), sufficient oil will wet out of the mist to provide adequate lubrication. However, at speeds much below 1000 lfm, enough oil might remain in the vented carrier air to produce an objectionable level of stray mist. As a general rule, to minimize stray mist, use reclassifying fittings wherever possible.
FITTINGS PLACEMENT
Place spray fittings to discharge close to the lubricated elements, preferably within 1 in. Remote positioning is permissible if their outputs are ducted to and flow through the lubricated elements because of relative positions of fittings and vents, and if passages downstream from the fittings are horizontal or slope downward toward the elements. To spray directly on elements moving at speeds up to about 1600 lfm, keep spacing between spray fittings and moving surfaces under 1/20 in./in. water column (w.c.) mist pressure. At higher speeds, install spray fittings 1/8 to 1/4 in. from the moving surfaces and use higher mist pressures—40 in. w.c. above 2000 lfm and up to 80 in. w.c. at much higher speeds. While spray fitting discharge may be in any direction, they work most efficiently when discharging downward. When they discharge upward, the calculated mist cfm used to select fitting sizes should be doubled. Condensing fittings may discharge in any direction between horizontal and downward, with the latter preferred. They should discharge directly into bearing lubrication grooves or as close as possible to lubricated surfaces. Mist fittings may discharge in any direction. Since their output is still in mist form, they may be located some distance from the elements to be lubricated, if flow to those elements can be assured.
VENTING
Venting must be provided for escape of carrier air from closed housings. Wherever possible, relative positions of vents, application fittings, and lubricated elements should produce flow from application fittings to lubricated surfaces. Where mist (nonreclassifying) fittings are used, this is almost always essential. Venting can be by means of appropriately located drilled holes or, frequently, by existing ports in the housing. The minimum recommended vent area is twice the total flow area of application fittings supplying flow to that vent. This vent area will produce housing back pressures equal to about 20% of manifold pressure. Vent ports can often serve as oil overflow or drain ports. In an oil sump application, the vent can be placed just above the normal sump oil level to provide an overflow path for any excess oil delivered by the mist system. Such vents should be located so that liquid oil will not splash out through the port. For a dry-sump application the vent can be placed at the bottom of the housing to also act as a drain. Vents should be protected from outside contaminants. Holes in the sides of housings should slope downward to the outside. Vent ports in the tops of housings should have shielded vent fittings.
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
412
DESIGN MANIFOLD PRESSURE Design manifold pressure (DMP) is the intended mist generator output pressure at which the fittings sizes are selected. The most commonly used manifold (or mist) pressure is 20 in. water column. In general, lower DMPs permit closer oil output control. Higher DMPs produce higher output velocities from spray-type application fittings to penetrate air barriers around high surface-speed elements (over 2000 lfm). Also, spray and condensing fittings reclassify more efficiently with higher pressure drops across them. Lately, this higher reclassification efficiency with higher DMPs has been used to reduce stray mist vented to the atmosphere.
APPLICATION FITTINGS SIZES
After calculating mist requirements and selecting the design manifold pressure and fittings types, refer to a manufacturer’s table or chart for fittings sizes. From Table 1, an edited excerpt from one such chart, select an application fitting to provide the calculated requirement. For example, if the calculated requirement is 0.25 scfm with a spray fitting at a DMP of 20 in. w.c., go down the column headed 20 in the section for spray type fittings. Since flow through a size 4 fitting would be an inadequate 0.159 scfm and that through a size 5 would be 0.300 scfm, the size 5 would be used. A different method for determining fitting type, size, and placement is used for elements in a closed housing that are lubricated from an oil sump by dipping into the oil or by oil rings, flingers, etc. Gear boxes and pump bearing housings are examples of such equipment where oil mist provides purging and makes up oil losses from the sump in a “purge-mist” application. Since the oil mist system does not actually lubricate the machine elements and sump losses cannot be calculated, simply start with a medium-size mist or spray fitting for each purgemisted housing, and then change fittings sizes as indicated by housing oil levels and overflow rates. A 1/8 in. diameter hole at the top of the sump oil level can act as vent and as overflow. Often a constant level oiler is used to control sump oil level, in which case a small hole is drilled in the side of the surge chamber just above the operating oil level as an overflow for any excess delivered by the mist system. To prevent depressing the sump oil level, the housing must be vented to an internal pressure no greater than about 1/10 in. w.c. Copyright © 1994 CRC Press, LLC
Volume III
413
FIGURE 2. Heater recommendations.
SELECTING THE MIST GENERATOR
Oil mist generator selection is based on flow capacity, heater requirements, and, sometimes, on desired reservoir refill interval. The required mist generating capacity is determined by adding together the flow ratings, at design manifold pressure, of all of the application fittings to be supplied. Select a generator for which this total is preferably near the center of its operating range. Flows less than the minimum specified for a generator might not reliably draw an adequate supply of oil from the reservoir to the mist generating head. If necessary to use a mist generator with a higher minimum flow rating man the total of the application fittings, then that total must be increased. This can be accomplished by using the same fittings at a higher design manifold pressure, by using fittings with higher flow ratings, by adding application fittings to lubricate more points, or by discharging reclassifying fittings into a vented receptacle. Heater requirements are determined from the oil mist equipment manufacturer’s design data, usually from a chart similar to Figure 2. Locate the operating point at the intersection of a vertical line at the minimum ambient temperature and a horizontal line at the viscosity of the oil to be used. Relate this operating point to the curves and read the recommendations from the curve labels. In general, whenever air heaters are used, oil heaters are also used. Even though heaters might not be required, they are often used to provide a more stable oil/air ratio under widely varying ambient temperature. If reservoir refill interval is of concern, compare the rate of oil consumption with available reservoir sizes. The oil consumption rate is the mist density (oil volume per scfm of mist flow per hour) multiplied by the total mist flow (scfm) from the generator. Since generators are
Copyright © 1994 CRC Press, LLC
414
CRC Handbook of Lubrication and Tribology
often operated at densities higher than those on which the formulas are based, use a density two or three times the manufacturer’s standard to size a reservoir for a minimum limit on refill interval. If this is an important consideration, consult the equipment manufacturer.
OIL MIST DISTRIBUTION
Oil mist is distributed to the lubricated points by pipe, tubing, and hose. Materials should be compatible with lubricants and resistant to external abuse, chemical attack, and ambient temperatures. Any material meeting these requirements can be used. Black pipe, plain steel tubing, etc. should be protected from external corrosion by paint. Bores should be coated with preservative oil to prevent corrosion prior to the introduction of oil mist. Since the mist system operates at a very low pressure, the use of pipe dope or other threat sealants is not necessary and can contaminate the mist distribution lines if improperly used. Flushing of the manifold is recommended to eliminate scale and dirt which can plug the small bores of application fittings. Mist distribution lines are sized to limit flow velocity. Table 2 gives the flows that various sizes of pipe, tube, and hose carry at several flow velocities. The maximum velocity of mist in the distribution system should be 24 fps. Above this velocity, the transition from laminar to turbulent flow will cause excessive wetting-out of oil from the mist. The general recommendation is to size lines to limit flow velocity to 18 to 20 fps to minimize line wet-out and permit some increase in system flow, if required. For systems supplying rolling mill roll neck bearings, 15 fps should be the maximum flow velocity. Use of oversize mist lines is permissible, but undersize lines should not be used. Copyright © 1994 CRC Press, LLC
Volume III
415
All parts of the mist distribution system must drain by gravity, avoiding traps—low spots in which oil could collect and interfere with proper mist flow. The main manifold should be sloped downward toward the mist generator wherever possible, especially the first part, for a distance equal to 300 times the pipe I.D., where most of the wetting-out occurs. Table 3 gives recommended manifold slopes. Branch lines, if sloped toward the main manifold, should be connected to the tops or sides of the manifold to avoid liquid oil traps. Where drainage provisions allow liquid oil and mist to flow in the same direction, horizontal runs do not require any slope. Horizontal runs which are not sloped should have drainage points not farther apart than 300 times the manifold I.D. The drainage may be to points of lubrication or to a standpipe or sump having provision for periodic dumping of collected oil. Where oil traps are unavoidable, a 3/64-in.-diameter hole can be drilled at the lowest point of the trap to drain oil. If an occasional drip and a little mist released to atmosphere are objectionable, run a drain tube to a suitable receptacle. Mist distribution lines should be routed as directly and with as few changes of direction as possible. While a limiting length of 300 to 400 feet is commonly recommended, numerous systems in service extend over a thousand feet. Valves should be avoided in mist distribution lines. Where valving is unavoidable, use fullflow ball valves. Within a few feet from the mist generator, butterfly valves may also be used. Quick disconnects in mist lines should be full-flow types, without check valves. In addition to the basic oil mist system, accessory components include: •
•
•
A mist manifold pressure gauge, usually calibrated in inches of water column, for visual indication of manifold pressure. Although not absolutely necessary for system operation, a mist pressure gauge is strongly recommended for every oil mist system. It simplifies and improves the accuracy of system adjustment and provides valuable information on system operation, especially during start-up and any trouble-shooting. An air supply on-off valve. A solenoid valve is often used to permit remote control or to interlock mist system operation with that of the lubricated machine. An oil heater to maintain the oil in the generator reservoir at low enough viscosity to insure adequate flow to the mist head.
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
416
• •
•
•
An air heater to stabilize oil/air ratio under conditions of varying ambient temperature or to mist higher viscosity oils which would not properly atomize at the prevailing ambient temperature. The air heater should be interlocked to prevent energizing without air flow through the unit. A mist manifold pressure switch to signal high or low mist pressure. This switch, with a relay, may be used to provide the recommended air heater interlock. An automatic refill system to supply oil to the mist generator reservoir from a drum or bulk tank. An oil mist monitor to signal deviation from the desired density of oil particles in the mist. One type photoelectrically monitors the density of the mist itself. Another, acting at one more lubricated point in the system, collects reclassified oil and periodically checks whether a preset minimum amount has been delivered.
APPLYING OIL MIST TO SOME SPECIFIC TYPES OF MACHINE ELEMENTS
The formulas used to calculate mist flow requirements are mathematically similar from one manufacturer to another. The formulas used here are representative and provide a fairly realistic indication of the mist requirements for most systems.
ROLLING ELEMENT (ANTIFRICTION) BEARINGS cfm Calculations (see Figure 3)
Moderate service cfm = 0.025DR Heavy service cfm = 0.050DR Rolling mill service cfm = 0.070DR Where D = shaft diameter in inches R = number of rows of rolling elements
The heavy service formula is used for bearings that are constantly thrust-loaded or preloaded, or are on shafts transmitting more than 35 to 40 horsepower, or that are subjected to high inertial loads either by frequent hard starting and stopping or by unbalanced shaft designs. The rolling mill service formula is used for work roll and backup roll bearings in both ferrous and nonferrous rolling mills. Moderate service is any not included in the other service definitions. Spray fittings are preferred for rolling element bearings. Mist fittings are used for moderate service in closed housings where it is not practical to place a spray fitting close to each bearing. Because most of the output of a mist fitting will remain airborne until carried into the turbulent region of a bearing, mist fittings can be installed remotely from the bearings and several bearings in a housing can be served by one mist fitting. To utilize mist fittings, bearings must be operating at speeds no lower than 200 lfm and preferably above 1000 lfm, at the mean diameter of the bearing. Vents must be positioned and sized to proportion positive mist flow through each bearing. An exception to the forced flow principle of inlet and vent location is used advantageously on single row, moderate service ball bearings on shafts under 4 in. in diameter, operating over 200 lfm. Bearings in this category, mounted in the wall of a machine housing containing a mist atmosphere, can be lubricated if both sides of the bearing are freely exposed to the mist. A drilled hole or undercut with a minimum area of 0.049 in.2 in the outer race support is used to expose the outboard side of a bearing in a blind wall mounting to the mist in the housing. Windage created by the rotating bearing will create sufficient flow of mist to the rolling Copyright © 1994 CRC Press, LLC
Volume III
417
FIGURE 3. Mist lubrication of rolling element bearings.
elements. The cfm requirement for each such bearing should be included in the calculated mist input for the housing. For radially mounted bearings designed to carry thrust loads (heavy service), such as angular contact ball or tapered roller bearings, spray fittings and vents should be located so the flow through the bearing is opposite to the direction of thrust from the shaft. This is not necessary if an oil sump or bath is maintained.
Copyright © 1994 CRC Press, LLC
418
CRC Handbook of Lubrication and Tribology
Tapered roller bearings, operating with light preloads, are best lubricated from the small end of the rollers. On heavily loaded tapered roller bearings, two spray fittings per row can be used to advantage, especially in critical applications such as precision machine tool spindles. For these installations use the heavy service or even the mill service formula. On the upstream side of the bearing, use a spray fitting sized to deliver about one third of the calculated requirement, and on the vent side about two thirds. On the vent side, place the fitting as close as possible to the bearing and try to direct its output toward the bearing and away from the vent. Vents and application fittings should generally be located so the calculated cfm for each row of rolling elements in heavy service flows through mat row before exhausting through the vent. For bearings in moderate service, such forced flow may be through two consecutive rows, if necessary. With the exception noted previously, forced flow through bearings is necessary when using mist fittings. With spray fittings discharging close to lubricated elements, the vents and fittings need not be on opposite sides of the bearings. Oil sumps are recommended for all heavy service bearings and for all moderate service bearings mounted on shafts 4 in. or larger in diameter. A depth of oil just above the bottom of the inside diameter of the bearing cup is recommended for tapered roller bearings. For other bearing types the depth of oil should be to the midheight of the rolling element at the bottom of the bearing. Vent locations can be used to maintain the proper oil level; for highspeed bearings, these vents should be located so oil is not thrown out of the vent ports. Bearing housings with double-lip seals require spray inlets and vents located to maintain an oil sump in the area between the contacting lips. The seal cfm requirement is equal to a row of elements on the same shaft in moderate service.
PLAIN BEARINGS
cfm Calculations (see Figure 4)
Light service cfm = 0.01LD Moderate service cfm = 0.02LD Heavy service cfm = 0.03LD Where L = bearing axial length in inches D = shaft diameter in inches
Light service bearings are those on horizontal shafts where the load zone is always in the lower half of the bearings, or those which are mounted in any position with contact type seals to retain oil in the bearing, or have porous bushings or synthetic “frictionless” sleeves. Moderate service bearings are those on horizontal, oscillating shafts where the load zone is always in the lower half of the bearing; or are unsealed bearings subjected to shock loading where the load zone constantly shifts, but boundary lubrication is permissible, such as king pins and spring pins on trucks; or are small bearings on rotating nonhorizontal shafts. Heavy service bearings are on rotating or rapidly oscillating shafts where the load zone shifts more than 180 degrees, such as crankshaft and crankarm bearings; or are large bearings without seals that are not mounted on horizontal shafts. Condensing or spray fittings are recommended for plain bearings with 360°-sleeves. Fittings are installed to discharge into bearing grooving or other oil reservoir within the bearing housing. There should be at least one application fitting for each 5 in. of bearing length and spacing between fittings should not exceed 5 in. With half bearings, spray fittings are installed to spray on the shaft near its line of entry into the bearing. Recommended axial spacing between fittings is about 2 in. Groove locations in oil mist-lubricated bearings are consistent with accepted general practice for oil-lubricated plain bearings. Grooves should be located so that 90% of the area Copyright © 1994 CRC Press, LLC
Volume III
419
FIGURE 4. Mist lubrication of plain bearings.
of the ungrooved surface (usually the shaft) passes over one oil groove during each cycle, as with a longitudinal groove extending 90% of the sleeve length in a rotating bearing. Oscillating bearings may require several longitudinal grooves. In small oscillating bearings, these primary grooves can be connected by a circumferential secondary groove with a single application fitting supplying all grooves. On large bearings of this type, a fitting should be used for each longitudinal groove. With a constant source of lubricant input, large volume grooves, acting as reservoirs, are not required and can be a disadvantage on machine startup if oil has
Copyright © 1994 CRC Press, LLC
420
CRC Handbook of Lubrication and Tribology
FIGURE 5. Mist lubrication of gears.
completely drained during a shutdown period. All groove edges or housing parting line edges facing the oncoming sliding surface should be rounded or chamfered to prevent scraping the oil from the shaft. Grooves should be in the unloaded zone of hydro dynamically lubricated bearings. Grooves are also used to vent plain bearings. To accomplish this, longitudinal grooves should extend to within 1/4 in. from the end of the sleeve in horizontal bearings and be provided with one or more vent holes. Circumferential grooves in vertical sleeves should be in the upper third of the sleeve, and a longitudinal groove extending upward from this groove to the end of the sleeve is preferred for venting. This venting groove should be diametrically opposite the application fitting inlet. Bearings with very small clearances or with contact seals require a vent port connected to the top of the internal grooving. The inlet and vent ports may be combined as shown for vertical and horizontal bearings in one of the illustrations of Figure 4. This is practical because the oil output from the condensing fitting is no longer airborne and will flow downward while the carrier air escapes upward.
GEARS
Copyright © 1994 CRC Press, LLC
Volume III
421
FIGURE 6. Mist lubrication of chains.
Spray fittings are used to spray directly on gears with a spray fitting for each 2 in. of face width. For a wide gear face divide the calculated cfm equally among the fittings. For moderately loaded gear trains, sprays directed at every second or third gear in the train will generally suffice. Using the formula, calculate the total cfm requirement for the train. Proportion this by “eyeball estimate” to the lubrication points selected. Use more or larger fittings on larger gears or on those with more mesh points. For heavily loaded gear trains, provide a spray for each mesh point, estimating proportioning to determine sizes of fittings. For unidirectional operation, direct spray at load side of gear teeth. For reversing service, direct spray toward gear axis.
CHAINS
Spray fittings are used for chain lubrication. For roller and conveyor chains, direct spray onto edges of link plates. For single-strand roller and for conveyor chains, apply about one half of the calculated cfm to each row of link plates. For multiple-strand roller chain, divide the calculated cfm by the number of rows of rollers to find the spray fitting size for the inner rows of link plates. Use one half of this value for the two outer rows of link plates For silent chain,
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
422
provide one spray fitting for each 1/2 in. of width, starting 1/4 in. in front outside edges. To size fittings, divide the calculated cfm by twice the chain width. For best results, position fittings to spray slightly against the chain motion on the inside of the slack strand. If spraying the working strand, point the spray slightly in the direction of chain motion.
OILS FOR USE IN MIST SYSTEMS
As a general rule, assistance of oil suppliers should be used in selecting appropriate oil for mist systems. While the primary consideration is the lubricant requirements of the machine elements, the mist system does place some restrictions on oil selection. The oil must, of course, be capable of being misted. Highly foam-suppressed oils and those containing tackiness additives or soap fillers should not be used. Viscosity index improvers might also inhibit mist generation. Motor oils should not be used because their mistability varies widely from one producer to another, from grade to grade, and even from lot to lot. Most oils compounded with EP additives, rust and oxidation inhibitors, detergents, and dispersants can be used in mist systems. Oils with viscosities up to more than 5000 SUS or ISO 1000 can be mist-applied. Because oil mist particles are so small, they might not transport solid particulate additives such as graphite and molybdenum disulfide. Therefore, such additives should be avoided unless it is experimentally determined mat the specific one being considered will not accumulate in the reservoir and eventually plug the intake screen or mist head passages. “Mist oils”, offered by many oil suppliers, are slightly mist-inhibited to reduce production of extremely small oil particles that are the most difficult to reclassify. In most cases, these mist oils greatly reduce problems of stray mist.
PRELUBRICATION AND MACHINE START-UP
All machine elements to be oil mist-lubricated should be prelubricated with the oil used in the mist system. Grease should be flushed out of grease-packed elements and then they should be prelubricated with the mist oil. This flushing must insure that residual grease will neither clog mist inlets or vents nor prevent penetration of the oil to the load-bearing areas. When prelubricated elements are to be stored prior to use, a preservative oil should be used instead of the mist system oil. In applications using oil sumps, these should be prefilled. Heavy service, sliding contact applications which do not have oil sumps, such as large plain bearings and worm gears, should be run in with bath or circulating oil lubrication before converting to mist. Oil flow to the mist head should be set at maximum during initial startup of any mistlubricated equipment.
Copyright © 1994 CRC Press, LLC
TRIBOLOGY IN HIGH-SPEED MACHINING Ranga Komanduri and Donald G. Flom
INTRODUCTION The ability to improve productivity and reduce costs by increasing cutting speed in machining has always attracted the attention of manufacturing engineers. For example, the introduction of tool steel with a new heat treatment method by Taylor1 and White at the mm of the century for machining ferrous materials at 2 to 3 times the then-commonly used cutting speeds led to the naming of that tool material as high-speed tool steel (HSS); that name still persists, even though other tool materials have far surpassed the cutting speed capability of HSS. Potential for further productivity gains (by orders of magnitude) by machining materials at extremely high cutting speeds, i.e., high-speed machining (HSM) has periodically attracted the attention of researchers. Current interest in HSM can be traced to the highly speculative and controversial HSM work of German inventor Carl Salomon in 1931.2 Based on an estimate of chip temperatures, Salomon speculated that temperature and wear of cutting tools increase with increasing speed until they reach maximum values at a speed characteristic of a given material, after which they decrease rapidly with further increase in speed (Figure 1). Such interpretations by Salomon and others3 have led to the false conclusions and high expectations that the higher the cutting speeds, the less wear the tool will suffer. As Schmidt4 has pointed out, this graph is the basis of more speculation than any other claim in metal-cutting literature. Schmidt5 also made an interesting comment that there is no “wonder” cutting fluid, no “miracle” tool material, “no atomic disintegrator”, nor “magic angle” discovered which will automatically result in high, accurate production and thus do away with exacting tooling requirements. At about the same time, Seikmann6 observed that machining titanium and its alloys will always be a problem no matter what techniques are employed. These observations were based on the rapid tool wear experienced in high-speed machining of ferrous alloys7 and titanium alloys.6 Our ability today to machine aluminum alloys and many other materials at higher speeds economically is due to the development of more refractory, stronger, and tougher tool materials as well as the development of high power, rigid machine tool systems and not because of Salomon’s hypothesis.
WHAT IS HIGH-SPEED MACHINING?
HSM involves the use of peripheral cutting speeds significantly higher than those currently used in most machine shops. The actual cutting speed that can be achieved depends on a variety of factors including the materials to be machined (metallurgical and crystal structures, chemical composition, heat treatment, etc.), type of machining operation, machine tool, the cutting tool used, and the requirements of the part itself. Chip disposal, safety, and economic factors must also be considered. For example, the maximum peripheral cutting speeds achieved so far range from less than 2.54 m s-1 (500 surface feet per minute [sfpm]) for titanium alloys, to about 5.08 m s-1 (1000 sfpm) for nickel-based superalloys, to about 7.62 m s-1 (1500 sfpm) for low- to medium-hardness ferrous alloys, and to more than 25.4 m s-1 (5000 sfpm) for gray cast iron. Cutting speeds for conventional aluminum alloys (i.e., nonabrasive low silicon alloys) are practically without limit, while the speeds are about 7.62 m s-1 (1500 sfpm) for high silicon-aluminum alloys with polycrystalline diamond tools. Copyright © 1994 CRC Press, LLC
423
424
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
FIGURE 1. Hypothetical HSM diagram, showing the variation of tool temperature with speed. At a given temperature, t, there are two cutting speeds: the lower speed, Va and the much higher speed Vb, at which tool life is apparently the same. The region beyond b, according to Salomon, is the promised land, where tool temperature apparently decreases with speed resulting in “Unlimited” tool life. (From Salomon, C., German Patent No. 523594, 1931)
Volume III
425
Productivity and overall costs depend on both cutting time and non-cutting time (labor, and overhead).8 HSM can decrease cutting time by increasing cutting speed. Non-cutting time can be decreased by automatic loading and unloading of parts, automatic tool changing, inprocess inspection, in-process sensing, and adaptive control. Labor costs can be decreased in HSM, since fewer operations are needed to work on fewer, more efficient machine tool systems. Similarly, overhead costs can be reduced by operating fewer, more efficient machine tools on two or more shifts, during holidays and at night. Figure 2 shows the percentage decrease in floor-to-floor time with varying cutting speed for different ratios of cutting time to floor-to-floor time, using current speed of whatever the value may be for a given material and process as a base. When the cutting time is a significant fraction of the floor-to-floor time and when tool wear at high speed is not significant (solid lines of Figure 2), cutting speed can be increased considerably to affect a significant reduction in floor-to-floor time. If, however, this ratio is low (bottom curve in Figure 2), an increase in cutting speed by even an order of magnitude or more will result in only a marginal decrease in floor-to-floor time. In this case, unless the non-cutting time could be decreased significantly, HSM would not be advantageous. Similarly, when an increase in cutting speed does not contribute to a significant reduction in floor-to-floor time and/or when tool wear is significant at high speeds, as in the machining of titanium alloys (dotted line in Figure 2), an increase in cutting speed may not be an economic proposition. Several criteria can be used to classify HSM. For example, one somewhat empirical criterion von Turkovich9 proposed for cutting speeds is given as:
A second criterion is based on the type of chip formed (see Section “Mechanisms of Material Removal in HSM” for details). For those materials that do not yield a continuous chip, “high-speed” is only relative for a given work material and its metallurgical condition prior to machining. While some materials exhibit transitions in chip form (e.g., discontinuous chip, continuous chip, segmental chip, shear-localized chip) as cutting speed is increased, others do not exhibit any transitions. For example, 6061-T6 aluminum or soft low carbon steel continue to yield continuous chips from low speed up to very high cutting speeds ( 500 m s1 [ 100,000 sfpm]), whereas titanium 6Al-4V gives a shear-localized chip from an extremely low speed (0.02 mm s-1 [0.050 in/min]) to very high speeds ( 100 m s-1 [ 20,000 sfpm]). A common feature in machining these difficult-to-machine shear-localized chip materials is transition from a continuous chip to a shear-localized chip morphology in a range of critical speeds which depends on the work material. If we now define high-speed machining as that in the speed range where a shear-localized chip forms, the range will be different for different materials and for different metallurgical states. For example, HSM will encompass practically all speeds with titanium alloys.10 It begins at about 1.0 ms-1 (200 sfpm) in the case of nickel-iron base superalloys (Inconel 718)11, and about 4.0 m s-1 (800 sfpm) in the case of hardened AISI 4340 steel (BHN 325).12 Once this type of chip forms, the mechanism of chip formation is found to be invariant with speed, although details may vary slightly depending upon thermomechanical properties of the work material. Copyright © 1994 CRC Press, LLC
426
CRC Handbook of Lubricants and Tribology
FIGURE 2. Variation of percent decrease in floor-to-floor time with cutting speed. Solid lines: high-speed tool wear insignificant. Dashed lines: high tool wear at high speeds. (From Komanduri, R., ASME Mech. Eng., December 1985, 64.)
A third criterion for HSM may be based on momentum effects which become significant at very high speeds. One can also use a criterion based on the percentage ratio of momentum force to the cutting force. Using the law of linear momentum, the following equations for momentum force Fm, and momentum energy per unit volume, Um, in machining can be developed.
where V is cutting velocity, Vc chip velocity, Vs shear velocity, α the rake angle, 6 the shear angle, p the density of work material in pounds per cubic inch, and (Vbt) is the volume of the work material deforming per unit time in cubic inches per minute. This momentum force acts along the shear plane with a velocity (Vc—V) or Vs. Momentum energy per unit volume becomes:
While the magnitudes of Fm and Um are almost negligible in low-speed machining, they increase rapidly (as the square of velocity) with speed in HSM. Table 1 lists representative data in orthogonal low-speed machining after Merchant.13 At a cutting speed of 2.76 m s-1 (542 srpm), Uf ~ 30% U, and both Fm and Um are seen to be very small compared to Fc and U, respectively. Values of Fm and Um calculated from the equations at various cutting speeds show that even at V = 50 m s-1 (10,000 sfpm), Fm and Um are only ~1% of Fc and U, respectively. Thus, momentum effects are still negligible. However, with further increase in Copyright © 1994 CRC Press, LLC
Volume III 427
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
428
speed, both Fm and Um increase rapidly (as square of velocity) and assume significant values. For example, at V = 305 m s-1 (60,000 sfpm), Fm 30% Fc, and Um 35% U; and at V = 500 m s-1 (100,000 sfpm), Fm Fc and Um U. Thus, at very high cutting speeds (>305 m s-1 [>60,000 sfpm]) additional energy, Um, is required to overcome momentum effects in HSM.
TRIBOLOGICAL FACTORS IN HSM OF DIFFERENT MATERIALS
Tribology plays an important role in machining in general and HSM in particular, by involving a real area of contact close to apparent area, high normal stresses, very high interfacial temperatures close to melting, and highly active and freshly generated (nascent) surfaces. In HSM these conditions are further accentuated, leading to rapid tool wear and decreased tool life. In an investigation of the HSM process Recht14 found that higher chip-tool interface temperatures at high speeds can lead to partial melting of the chip and consequent decrease in friction coefficient. In the case of shear-localized chip formation in HSM, the large fluctuating friction force due to start-stop instability and ramp upsetting, and related high temperatures, can produce rapid crater wear. In addition, the wear land is also subjected to ramp climbing pressures and buffeting from the chip segmentation process. This increases temperatures at the flank-workpiece interface; the nearby tool-chip interface also contributes to the wear land temperature.
MECHANISMS OF MATERIAL REMOVAL IN HSM
Two types of chip formation processes have been observed in HSM (Figures 3 [a] and [b]), depending on the type of work material and its metallurgical condition. They are the continuous, ribbon-like chip and the shear-localized (segmental) chip, a term arising from the intense deformation between segments of the chip. Merchant in his classical work on metal cutting13 elucidated the mechanisms of continuous chip formation in considerable detail. Continuous chips are likely to occur in HSM of a metal or alloy of bcc/fcc crystalline structure, a high value of the product of thermal properties, kpc, and by low hardness, such as aluminum alloys or soft, low-carbon steels. Shear-localized chip formation process has been studied and understood in some detail only recently.15,16 Shear localization occurs with such materials as titanium alloys, nickelbased superalloys, and hardened alloy steels which are characterized by low value of thermal properties (kpc), hcp crystalline structure, and high hardness. Figure 4 is a schematic of the observed shear-localized chip formation process. This unstable, cyclic cutting action arises from the thermomechanical instability of the work material under the conditions of cutting. There are basically two stages involved: (1) plastic instability and strain localization in a narrow band in the primary zone ahead of the tool, and (2) upsetting of the inclined wedge of work material by the advancing tool with negligible deformation in forming a chip segment. During the upsetting of the segment ahead of the tool in the primary zone, intense shear takes place at approximately 45° to the direction of cutting. This occurs not in the conventional primary shear plane ahead of the tool, but between the previous segment and the one just forming (i.e., in region 2, 3, 4 in Figure 4), starting at the tip of the tool and gradually moving away from the tool tip. As the width of this intense band is extremely small (a few micrometers) and the extent of shear quite considerable, the shear strain in this band is rather large. The thermomechanical response of these difficult-to-machine work materials tends to localize the intense heat generated due to strain localization, with the result that the material in this band may exhibit superplastic flow and large deformations. In shear-localized chip formation, initial contact between the segments being formed and the tool face is at the apex of the tool, and the contact length is extremely short. The contact Copyright © 1994 CRC Press, LLC
Volume III
429
FIGURE 3. (a) Photomicrograph of a continuous chip obtained in high-speed ballistic machining of AISI 1018 steel at 372 m s-1 (73,170 sfpm). (b) Photomicrograph of a shear localized titanium 6Al–4V chip. Cutting speed 0.76 ms-1 (150 sfpm).
length increases as the flattening progresses. Relative motion between the segment being formed and the tool face is very small, until near the end of the flattening stage. Gradual bulging of the chip segment during the upsetting process slowly pushes the chip segment previously formed. The contact zone between the segment just formed and the one being formed shifts gradually, beginning close to the work surface and shifting towards the tool face as flattening progresses. Also, as the cutting speed increases, the rate of shear between the segments increases, decreasing the contact area between any two segments. Figure 5 (a) and Copyright © 1994 CRC Press, LLC
430
CRC Handbook of Lubricants and Tribology
FIGURE 4. Schematic of the observed shear-localized chip formation process, showing various surfaces that take part in the process.
(b) are optical photomicrographs of shear-localized chips of AISI 4340 steel at two speeds, showing the segments intact at the lower speed and completely separated at higher speed. These observations in HSM are extremely important from the point of tool wear, tool life, chip-tool interference temperature, and friction. In the case of titanium alloys, shear-localization occurs at all speeds from extremely low (.021 m/s [0.050 in./min]) to very high ( 100 m s-1 [ 20,000 sfpm]). With other difficultto-machine materials, such as nickel-base superalloys and hardened alloy steels, it occurs within and above a transition speed range that depends on the material and its metallurgical condition. Figure 6 shows such a transition in chip morphology with cutting sped for AISI 4340 steel (325 BHN). Basically, continuous chips at low speed (Figure 6(a)) transform to segmented chips composed of distinct trapezoidal segments. The deformation of the chip is inhomogeneous on a gross level with two wide regions: one where deformation is very high (i.e., between the segments) and the other where deformation is relatively low (i.e., within the segments). With further increase in speed, the shear-localized instability is established completely in the primary shear zone, which results in concentrated intense shear between the segments separated by large areas of relatively less deformed material within the segments (Figure 6(b)). At still higher speeds, the extent of contact between segments decreases, which results eventually in completely isolated segments.
TEMPERATURES GENERATED, TOOL WEAR, AND TOOL LIFE IN HSM
Tool-Chip Interface Temperature Salomon in 1931,2 based on thermocouple measurements in high-speed milling tests, made an intriguing proposition that chip-tool interface temperature (and consequently tool Copyright © 1994 CRC Press, LLC
Volume III
431
FIGURE 5. Photographs of segmented AISI 4340 steel (325 BHN) chips, (a) Individual segments intact at 2.0 ms-1 (400 SFPM). (b) Segments completely separated at 16 ms-1 (3200 SFPM).
wear) increases with cutting speed, reaching a maximum value for a given work material followed by a sharp decrease with further increase in speed (Figure 7). The apex of each curve coincidentally occurs near the melting point of each work material under consideration. While it is understandable that tool-chip interface temperatures increase with cutting speed, no explanation was given for the rapid decrease in temperature at higher speeds. Opitz and Kob17 and Schmidt5 subsequently clarified this anomaly by pointing out misinterpretation of thermocouple data by Salomon at or near the melting temperature. McGee18,19 conducted high-speed turning tests on a 2024-T652 aluminum alloy using a C2 carbide cutting tool at various speeds (up to 25.4 ms-1 [ 5000 sfpm]). Cutting edge temperatures continued to increase with speed, reaching close to the melting point of the aluminum alloy (Figure 8). Based on this work, McGee pointed out that successful HSM is dependent upon the cutting tool withstanding the melting temperature of the work material and the impact load to which it is subjected. Recht and Kottenstette developed a two-color pyrometer technique to measure the chiptool interface temperature in HSM at speeds up to 254 m s-1 (50,000 sfpm).20 Recht also proposed an equation for the chip-tool interface temperature.21 Figure 9 shows the variation of the experimental and computed values of chip-tool interface temperature with cutting speed for several combinations of friction coefficient and work material characteristics. These results are similar to the results McGee reported. The possibility of partial melting was verified by SEM examination of the chips generated wherein particles worn from the tool insert were found embedded in a “molten” matrix of steel frozen on each segmented chip. Figure 10 shows evidence of melting in ultra-high-speed machining of AISI 4340 steel (RC39).20 The micrograph at the upper left shows individual segments of chips at 249 m s-1 (49,000 sfpm). Copyright © 1994 CRC Press, LLC
432
CRC Handbook of Lubricants and Tribology
FIGURE 6. Photomicrographs showing the variation in chip morphology with cutting speed for AISI-4340 steel (325 BHN). (a) Continuous chip at 0.64 ms-1 (125 SFPM). (b) Fully developed shear-localized chip at 4.1 ms1 (800 SFPM).
At upper right is a micrograph at higher magnification of another AISI 4340 steel chip generated at 76 ms-1 (15,000 sfpm), showing the boundary between the shear surface (right portion of picture; surface 3 in Figure 4) and the rake face surface of the chip (left portion of picture, surface 4 in Figure 4). Note the melt from the chip-tool interface has flowed out onto the shear surface and frozen. The lower left micrograph shows the two quenched melts (white specks) at higher magnification. The large central globule was examined by X-ray energy dispersive analysis to give the spectrum at lower right. In the X-ray spectrum, iron is predominant. In addition, tungsten and titanium are present, indicating the transfer of material from wear of the tool (titanium carbide coated cemented tungsten carbide) on to the chip.
Copyright © 1994 CRC Press, LLC
Volume III
FIGURE 7. Illustration of Solomon’s hypothesis for the effect of cutting speed on cutting temperature for different work materials.
433
Copyright © 1994 CRC Press, LLC
434
CRC Handbook of Lubricants and Tribology
FIGURE 8. Variation of tool temperature with cutting speed. Temperature increases with cutting speed and reaches the melting temperature of the work material asymptotically at higher speeds. (From McGee, F. J., in High-Speed Machining, Komanduri, R. R., et al., Eds., PED Vol. 12, American Society of Mechanical Engineers, New York, 1984, 205.)
FIGURE 9. Variation of chip-tool interface temperature with cutting speed. Dashed lines are computed values of the tool-chip interface temperature; plotted values are measurements of interface temperature for three alloy steels and Inconel 718. (From Kottenstette, J. B., in High-Speed Machining, Komanduri, R. et al., PED Vol. 12, American Society of Mechanical Engineers, New York, 1984, 372.)
TOOL WEAR
Kramer and Suh22–24 have developed a methodology for estimation of wear rates of different tool materials based on different wear mechanisms experienced in machining, such as abrasive wear, delamination wear, solution wear, chemical wear, and diffusion wear. General guidelines were proposed for selecting the tool material on the basis of properties Copyright © 1994 CRC Press, LLC
Volume III
435
FIGURE 10. Evidence of melting in the tool-chip interface during high-speed machining of AISI 4340 steel (Rc 39). Chip segments (top left). Quenched melt (top right and bottom left). X-ray energy spectrum showing Ti and W is additional iron (bottom right). (From Recht, R. F., in High-Speed Machining, Komanduri, R. et al., Eds., PED Vol. 12, American Society of Mechanical Engineers, New York, 1984, 91.)
such as transverse rupture strength (TRS), fracture toughness, hot hardness, strength, deformation resistance, abrasion wear resistance, softening and melting temperatures, solubility in the work material at different operating temperatures, and chemical stability. Although more than one mode of wear can be operative at any given cutting condition, one mode generally predominates. Once identified, factors that affect this mode of wear can guide selection of a suitable tool material. Alternately, once the best tool material is selected, cutting conditions can be optimized to minimize wear leading to economic tool life consistent with other requirements of the part. While somewhat difficult to quantify the level of fracture resistance required of a tool material for a given application, those materials with higher transverse rupture strength (TRS) and fracture toughness are notably less prone to fracture. These materials may be used in more complex geometries, at higher force level, or under interrupted cutting conditions. Factors other than TRS or fracture toughness should, of course, also be considered in overall selection of a tool material for HSM. Physical mechanisms responsible for producing gradual wear depend on cutting temperature and the corresponding cutting speed. Kramer23 proposed the following three cutting speeds in terms of the predominant wear mechanisms: 1. 2. 3.
Low speed (mechanically activated wear) High speed (chemical dissolution wear) Very high speed (diffusion limited wear)
In the low speed range, chemical interactions between the tool and work material are insignificant and wear occurs predominantly as discrete particles caused by microfracture, Copyright © 1994 CRC Press, LLC
436
CRC Handbook of Lubricants and Tribology
FIGURE 11. Tool life vs. cutting speed for machining AISI 1045 steel (170 BHN) with a carbide and oxide tool. (From Seikmann, H. J., Tool Eng., April 1958, 85.)
thermal and mechanical fatigue, and abrasion.23 In the high speed range, chemical dissolution of the tool material into the work material is considered the most important contributor to wear. When the sliding chip is in contact with the tool face, materials dissolve in each other if the free energy of the material pair decreases by the formation of solution. The rate of dissolution increases with increase in tool-chip interface temperature, and the tool material more resistant to dissolution wears the least. In considering thermodynamic properties of the tool-workpiece system, Kramer found the theoretical prediction for carbides and nitrides to agree closely with experimental values. Although wear rates predicted for oxides were significantly lower than the measured values, this methodology provides a reasonable basis for selection of tool materials. At very high speeds, according to Kramer, wear is limited by the rate of mass diffusion. Both in the case of diffusion and solution. Wear is caused by the lack of sufficient chemical stability in the tool material to resist dissolution or decomposition. In high-speed machining, chemical interactions become more predominant than mechanical factors.
Tool Life Seikmann7 conducted turning tests on a specially built, high-speed (up to 127 m s-1 [25,000 sfpm]), high-power (150 hp) lathe. Figure 11 shows variation of tool life with cutting speed for a AISI 1045 steel (170 BHN) with a carbide and a ceramic cutting tool. At cutting speeds over 50 ms-1 (10,000 sfpm), tool life is less than 1 minute with the oxide ceramic tool. The tool life, however, is further affected by the chemistry, metallurgical and crystal structure, hardness, and thermomechanical properties of the work material. McGee19 conducted high-speed milling tests on a 6010-T6 aluminum over a range of cutting speeds from 40 m s-1 to 50 m s-l (8,000 to 10,000 sfpm) with a two flute, 2 in. diameter HSS end mill. Tool life was somewhat less than 1 hour at 40 m s-1 (8,000 sfpm) and dropped to about 4 min at the maximum speed. If, however, a solid carbide end mill is used, tool life can be extended an order of magnitude or more. If smaller size end mills have to be used, Copyright © 1994 CRC Press, LLC
Volume III
437
then use of solid carbide end mill will provide higher stiffness over HSS end mill. Extra cost of a solid carbide end mill will be more than offset by increased tool life, higher stiffness of the tool, better finish, and accuracy of the part.
TOOL MATERIAL SELECTION FOR HSM OF SPECIFIC MATERIALS
A wide range of cutting tool materials is available for HSM including HSS, cemented carbide (coated and uncoated), ceramics (oxides such as hot pressed or HIP’ed Al2O3 or Al2,O3 + TiC, SiC whisker-reinforced Al2O3, nitrogen ceramics such as SiAlON, Si3N4), sintered polycrystalline diamond or cubic boron nitride.25 Selection of an appropriate tool material will depend on the following: range of properties; performance capabilities; cost; work material to be machined (chemistry, metallurgical structure, and thermomechanical properties); type of machining operation; cutting conditions; part geometry and size; lot size; condition and capabilities of available machine tools; finish, accuracy, and surface integrity requirements of the part; machining time vs. nonmachining time, etc. As a candidate for a given HSM application, a tool material must meet one or more of the following requirements: high elevated temperature hardness and wear resistance; high transverse rupture strength and toughness; high deformation resistance; high softening and melting temperature; chemical stability; low solubility in the work material at different operating temperatures; adequate thermal properties; high stiffness; ease of fabricability; and reasonable cost. Table 2 gives suggested tool materials for HSM of different work materials, and discussion follows for specific alloy types. Aluminum alloys —Tool temperatures are characteristically low in machining most aluminum alloys because of their low melting temperatures and high thermal conductivity. Chemical interactions are minimal and wear is primarily mechanical in nature. Wear resistance increases from HSS to cemented carbide to polycrystalline diamond tools. While HSS can be used to machine most conventional aluminum alloys, for HSM cemented carbide is preferred because of longer tool life and higher stiffness of the tool. The latter is very critical for endmilling cutters and drills where deflection of the tool may affect the finish and accuracy. For the highly abrasive silicon-aluminum alloys, polycrystalline diamond is preferred. All in all, existing tool materials are adequate for machining aluminum alloys at any conceivable speed provided adequate machine tool systems (rigidity, power, etc.) are available. Ferrous alloys—Oxide ceramics are the only potential tool materials for HSM of soft to medium steels, in the 5 to 12.7 m s-1 (1000 to 2500 sfpm) speed range. Improving the Copyright © 1994 CRC Press, LLC
438
CRC Handbook of Lubricants and Tribology
toughness of ceramic tools with reinforcement similar to SiC whisker-reinforced alumina may provide the thermomechanical shock resistance required for the tool in HSM. Further increase in cutting speed capability depends on the development of more refractory and thermally stable tool materials. While cubic boron nitride is perhaps the only tool material for machining hardened steels (>Rc50), cutting speeds are still limited to a few hundred sfpm. Cast iron—Polycrystalline cubic boron nitride can be used to machine gray pearlitic cast iron at speeds to 66 m s-1 (13,000 sfpm) with acceptable tool life.26 Similarly, silicon nitride tools (e.g., SiAlON) can be used successfully up to 25.4 m s-1 (5000 sfpm). Alloyed cast iron and white cast iron, however, can be machined with oxide ceramic tools (Al2O3, Al2O3 + TiC) at much lower speed (~7.6 m s-1 [~ 1,500 ft/min]). Tool cost would be a factor in the most costeffective tool selection. Nickel-iron-based superalloys—SiAlON, SiC whisker-reinforced alumina, alumina, and cubic boron nitride can be used to machine most nickel-iron based superalloys. With its high fracture toughness, SiC whisker-reinforced alumina is preferred in rough machining and in interrupted cutting. Alumina ceramics are used for finish machining at higher speeds. Titanium alloys—Titanium alloys are extremely difficult to machine except at low speeds (0.76 ms-1 [150 sfpm]) because of rapid flank wear and crater wear at the apex of the tool on the rake face. Almost all tool materials developed thus far, including oxides, carbides, borides, nitrides, diamond, and cubic boron nitride are highly reactive with titanium alloys to cause rapid wear at high speeds. HSS at low speeds and cemented tungsten carbide at moderate speeds are recommended for machining titanium alloys.
CUTTING FLUIDS IN HSM
Detailed discussions of cutting fluids and coolants in conventional machining are given in Volume 2 of this handbook27,28 and in the earlier ASLE Standard Handbook of Lubrication.29 Summarizing briefly, cutting fluids serve one or more of several basic functions: (1) lubrication of tool/chip/workpiece interfaces, (2) cooling the workpiece and the tool, (3) prevention or delaying formation of a built-up edge (BUE) on the tool, (4) inhibiting corrosion of the workpiece and/or the machine tool, and (5) clearing chips out of the cutting area. The kinds of cutting fluids used are straight petroleum-based oils, emulsifiable oils, synthetic fluids, and semisynthetic fluids. The straight oils generally contain additives such as oleic acid, neatsfoot oil, lard oil, and butyl stearate as well as chlorinated paraffins and sulfurized compounds, all for the purpose of aiding lubrication. Thus, such fluids are designed primarily for severe operations such as broaching, tapping, and threading which require excellent lubrication. The emulsifiable oils (mixtures of oil and water) generally contain rust preventives, antibacterial agents, and extreme pressure (EP) additives. Such fluids are designed for moderately severe operations such as reaming and drilling. The synthetic fluids are free of oils but may contain water-soluble materials such as fatty esters, inorganic salts, polyglycols, and alkanolamines. Since the synthetic fluids contain the optimum heat transfer agent—water—they are used in operations such as turning and milling which require maximum cooling. The semisynthetic fluids are combinations of emulsion-type fluids and water-soluble additives, aimed at providing the mixed functions of lubrication and cooling. There are many exceptions to these “rules of thumb” regarding the type of cutting fluid for a given machining operation. HSM falls into this category; the workpiece material, cutting tool, type of machining, speed regime, and other operating parameters all play a role in proper selection of a cutting fluid. Indeed, dry cutting with no fluid is sometimes quite adequate, one example being the machining of cast iron with cubic boron nitride (CBN) compacts. CBN can also be used to machine superalloys such as Inconel 718 without use of a cutting fluid; it Copyright © 1994 CRC Press, LLC
Volume III
439
has been speculated that CBN can form its own lubricant (boric oxide) at elevated temperature. There is also the possibility that in interrupted cutting, such as milling, the thermal shock provided by the cutting fluid can lead to premature tool failure. Thus, “control of temperature” is sometimes more pertinent than cooling or “lowering of temperature” per se. As already shown, cutting temperature increases with increasing speed, approaching asymptotically the melting point of the workpiece material. However, in HSM most of the heat generated in the shear zone goes into the chip and there is less time for heat to flow into the bulk of the workpiece. In the case of aluminum HSM, this temperature-limiting effect must be balanced against two other competing factors, namely, the greater coefficient of thermal expansion of aluminum relative to other metals (less tolerance for HSM), and reduced cutting force in HSM (improved tolerance). The same applies to the cutting tool in aluminum HSM. Built up edge (BUE) formation, a problem with aluminum at low speeds, is reduced in HSM; thus, the need for a cutting fluid to combat this effect is reduced. Also, since the cutting tool can withstand the temperatures generated in aluminum HSM, the tool wear rate is not greatly effected by the cutting fluid. On the other hand, aluminum machining tends to form a “bird’s nest of chips”. A high flow rate of fluid helps flush these chips out of the cutting area. Semisynthetic fluids are commonly a good choice for HSM of the more readily machinable aluminum alloys. These fluids are relatively inexpensive, aid chip removal, are good coolants, and provide adequate lubrication. They afford the added advantage of being smoke-free and, for those fluids which are transparent, the machine operator can easily observe the cutting operation. Recommendations of specific cutting fluids and manufacturers may be found in trade journals.30,31 For silicon-containing aluminum alloys, the need for cutting fluids is even more mandatory than for the more conventional aluminum alloys. Since lubrication requirements are more severe, oils or emulsions should be given strong consideration relative to synthetics and semisynthetics. For these workpiece materials, the choice of cutting tool is very important; as stated earlier, polycrystalline diamond compacts (PCD) are preferred. None of the cutting fluids can be expected to function if they cannot be injected into the cutting interface at the proper time. For this reason, mist cooling, high pressure jets, and even shop air are techniques under investigation. These methods are especially effective in aiding chip removal. This discussion of cutting fluids has centered primarily on HSM of aluminum, because it can be machined most productively at very high speeds. Similar considerations apply to steels, cast irons, superalloys, and nonferrous alloys. At the other end of the speed spectrum, any speed over 0.5 m s-1 (100 sfpm) may be regarded as HSM for titanium and its alloys. Cutting fluids are needed not only to cool both the titanium workpiece and the cutting tool, but also to prevent titanium chips from welding to the cutter and being recut with detrimental results. The best fluids for machining titanium contain chlorine; but fluids containing chlorine and other halogens are frequently banned from use on titanium, despite much evidence that such fluids do not harm the metal. Viscous chlorine-based oils have been used successfully, but they have shortcomings (they must be brushed on) and result in excessive smoke. Clearly, a good water-soluble cutting fluid is needed which can also minimize the burning of titanium chips.
ACKNOWLEDGMENTS
Much of the work reported in this article was conducted while the authors were at the General Electric Corporate Research Center, Schenectady, NY. The authors gratefully Copyright © 1994 CRC Press, LLC
440
CRC Handbook of Lubricants and Tribology
acknowledge the support of the Defense Advanced Projects Agency (DARPA through Contract F33615–79-C-5119 and the Air Force Wright Aeronautical Laboratories through Contract F33615–80-C-5057. Thanks are also due to Mr. Rodney Recht and Mr. James Kottenstette for the use of some figures in this article.
REFERENCES 1. Taylor, F. W., On the art of cutting metals, Trans. ASME, 28, 1907. 2. Salomon, C, Process for the Machining of Metals or Similarly Acting Materials When Being Worked by Cutting Tools, German Patent No. 523594, 1931. 3. Vaughn, R. L., Recent developments in ultra-high speed machining, ASTME Tech. Pap., No- 255, 1960. 4. Schmidt, A. O., Machine tool engineering: today’s facts and fiction, in High Speed Machining, Komanduri, R., Subramanian, K., and von Turkovich, B. E, Eds., Production Engineering Division (PED), Vol. 12, American Society of Mechanical Engineers, New York, 1984, 69. 5. Schmidt, A. O., Ultra high-speed machining—panacea or pipe dream?, Tool Eng., 105, 1958. 6. Seikmann, H. J., How to machine titanium, Tool Eng., Jan. 1955, 78. 7. Seikmann, H. J., High-speed cutting with ceramic tools, Tool Eng., April 1958, 85. 8. Komanduri, R., High-speed machining, ASME Mech. Eng., December 1985, 64. 9. von Turkovich, B. F., Influence of very high cutting speed on chip formation mechanics, Proc. 7th North Am. Manufacturing Res. Conf., 1979, 241. 10. Komanduri, R., Schroeder, T. A., Bandhopadhye, D. K., and Hazra, J., Titanium: a model material for analyses of the high-speed machining process, Advanced Process Methods for Titanium, Metallurgical Society of AJME, 1981, 241. 11. Komanduri, R. and Schroeder, T. A., On shear instability in machining of a nickel-iron base superalloy, Trans. ASME, J. Eng. Ind., 108, 93, 1986. 12. Komanduri, R., Schroeder, T. A., Hazra, J., von Turkovich, B. F., and Flom, D. G., On the catastrophic shear instability in high-speed machining of an AISI 4340 steel, Trans. ASME, J. Eng. Ind., 104, 181, 1982. 13. Merchant, M. E., Mechanics of the metal cutting process—orthogonal machining and a type-2 chip, J. Appl. Phys., 16(5), 267, 1945. 14. Recht, R. F., A dynamic analysis of high-speed machining, in High-Speed Machining, Komanduri, R., Subramanian, K., and von Turkovich, B. F., Eds., PED, Vol. 12, American Society of Mechanical Engineers, New York, 1984, 93. 15. Komanduri, R., Subramanian, K., and von Turkovich, B. F., High-speed machining, PED Vol. 12, American Society of Mechanical Engineers, New York, 1984. 16. Flora, D. G., Komanduri, R., and Lee, M., High-speed machining of metals, Annu. Rev. Mat. Sci., 1984, 231. 17.Opitz, H. and Kob, J., Richtwerter, Schnittkraffo, and Schnittemperaturen beiwn Frazen mit Hartmetalwerkzengen, Werkstatt Betr., 85, 81, 1952. 18. McGee, F. J., An assessment of high-speed machining, SME MR 78–648, 1978. 19. McGee, F. J., High-speed machining of aluminum alloys, in High-Speed Machining, Komanduri, R., Subramanian, K., and von Turkovich, B. F., Eds., PED Vol. 12, American Society of Mechanical Engineers, New York, 1984, 205. 20 Kottenstette, J. P., Measuring tool-chip interface temperatures, in High-Speed Machining, Komanduri, R., Subramanian, K., and von Turkovich, B. F., Eds., PED Vol. 12, American Society of Mechanical Engineers, New York, 1984, 371. 21. Recht, R. F., The feasibility of ultra-high speed machining, M.S. thesis, University of Denver, Denver, CO, 1960. 22. Kramer, B. M. and Suh, N. P., Tool wear by solution: a quantitative understanding, Trans. ASME J. Eng. Ind., 102, 303, 1979. 23. Kramer, B. M., On tool materials in high-speed machining, in High-Speed Machining, Komanduri, R., Subramanian, K., and von Turkovich, B. F., Eds., PED Vol. 12, American Society of Mechanical Engineers, New York, 1984, 127.
Copyright © 1994 CRC Press, LLC
Volume III
441
24. Suh, N. P., Tribophysics, Prentice-Hall, New York, 1986. 25. Komanduri, R. and Desal, J., Tool materials, in Encyclopedia of Chemical Technology, Vol. 23, John Wiley & Co., NY, 1983,273. 26. Flom, D. G., Reed, W. R., Jr., Hibbs, L. E., Jr., and Broskea, T. J., High-speed machining of cast iron with BZN compacts, Wear, 147, 253, 1991. 27. Kelly, R. and Foltz, G., Cutting fluids, in CRC Handbook of Lubrication, Vol. 2, Theory and Design, Booser, E. R., Ed., CRC Press, Boca Raton, FL, 1984, 357. 28. Bennett, E. O., Cutting fluids—Microbial Action, in CRC Handbook of Lubrication, Vol. 2, Theory and Design, Booser, E. R., Ed., CRC Press, Boca Raton, FL, 1984, 371. 29. Bastian, E. L. H., Metalworking, in ASLE Standard Handbook of Lubrication Engineering, McGraw-Hill Book Company, O’Connor, J. J. and Boyd, J., Eds., 1968, 23–1. 30. Bennett, K. W., Iron age’s guide to metal cutting fluids, Iron Age, November 5, 1984, 18. 31. Cole, T., Know your coolants, Cutting Tool Eng., October 1990, 59.
Copyright © 1994 CRC Press, LLC
HYDRODYNAMIC AND HYDROSTATIC SEALS Wilbur Shapiro
INTRODUCTION
Hydrodynamic and hydrostatic fluid-film seals incorporate interfacial geometric features that promote a positive-stiffness fluid-film between the opposed surfaces. Their principal advantage is negligible wear regardless of the pressure gradient and speed of operation. The disadvantage is that some leakage must occur to maintain the fluid-film. Operation of fluidfilm seals is very much related to fluid-film bearings except a pressure gradient exists across the boundaries and they operate at very small film thickness to inhibit leakage. “Hybrid” combinations also exist in which both hydrodynamic and hydrostatic action take place simultaneously. Figure 1 demonstrates the principal of operation for a hydrostatic face seal. By introducing a restrictor, or compensating element, between the high-pressure fluid and interface, a selfenergized seal is produced. A properly designed orifice-compensated seal face will have a pressure distribution lying somewhere between the distribution of a flat plate as shown by the linear drop in Figure 1 and the high-pressure fluid. By proper sizing of the orifice, the pressure at the recess can approach the high pressure when the clearance closes and the low pressure of the flat surface when the clearance opens. This arrangement provides stiffness and ensures the existence of a fluid-film at the interface so that wear is minimized. The interface clearance can be designed to be small (e.g., 5 microns) to inhibit leakage. A hydrodynamic seal uses relative velocity to generate a positive stiffness fluid film at the interface. Superimposed upon the usual pressure drop across the interface is the hydrodynamic contribution, produced by a geometric feature, such as a Rayleigh step, which causes pressure to rise in the circumferential direction. Clearance closure will cause the peak pressure to increase, while a clearance opening will cause a reduction in pressure. The phenomenon again produces positive stiffness.
Seal Types Figure 2 shows a variety of cylindrical hydrodynamic and hydrostatic seal configurations. Converging films in the flow direction are hydrostatic because stiffness and load is a function of the pressure gradient and not the rotation of the seal runner. Thus, the tapered and Rayleigh step seals in the flow direction are hydrostatic. The geometries machined around the circumference are in the hydrodynamic category. Figure 3 shows a variety of hydrostatic and hydrodynamic face seal geometries. The principal of the spiral groove designs is that relative motion causes viscous pumping of the fluid in the grooves. The fluid is pumped against a dam region so that a pressure buildup occurs that increases with decreasing clearance. If the surfaces do not distort, the spiral groove offers greater load capacity and stiffness than any other geometry. The grooves can be oriented in several different ways. A herringbone pattern is shown on Figure 2 in which each side pumps against each other towards the center. The grooves can be separated by a land or dam region. A more usual configuration for a seal application would be to have only one row of grooves and a land region. The end of the land would be exposed to the high pressure and the grooves would be oriented to pump against the pressure gradient. Figure 3 shows a spiral groove face seal that pumps the fluid from the OD to an interior dam region. Spiral-groove gaslubricated face seals are extensively applied in pipeline compressors between the high-pressure gas being pumped and the atmosphere. 0-8493-3903-0/94/$0.00+$.50 © 1944 by CRC Press. Inc.
Copyright © 1994 CRC Press, LLC
445
446
CRC Handbook of Lubrication and Tribology
FIGURE 1. Hydrostatic seal pressure distribution
FIGURE 2. Cylindrical seal configurations.
In all of the above configurations, the hydrodynamic and hydrostatic geometric thicknesses are of the same order of magnitude as the film thickness or seal clearance. If a seal operates at 5 µm, the depth of the step or taper or spiral groove is of the same order of magnitude, with a 1:1 ratio being quite common.
INTERFACIAL PERFORMANCE CHARACTERISTICS
General Theory To design a seal, it is necessary to know the load-carrying capability of the film, the stiffness of the film, and the leakage as a function of the film thickness. Viscous dissipation Copyright © 1994 CRC Press, LLC
Volume III
447
FIGURE 3. Face seal configurations.
is also an important input for determination of thermoelastic distortions. In most cases, the governing relation is the laminar Reynolds equation, because the film thicknesses are small. The general assumptions are: (1) laminar flow prevails and the shear stress is proportional to the velocity gradient; (2) inertia is small and neglected compared to viscous shear; (3) pressure across the film is constant; (4) height of the film is small compared to other geometric dimensions: curvature is ignored; and (5) viscosity of the fluid is constant. For cylindrical coordinates, a nondimensional form of the Reynolds equation is as follows:
Where X = x/L; Y = y/L; P = p/P0; p = pressure; A = 6m UL/(p0C20; p0 - reference pressure, y = direction normal to sliding; QC0, = reference clearance; L = length; U = sliding velocity; h = film thickness; t = time. Boundary conditions include: constant but not necessarily equal pressures on each boundary, periodic boundaries in the circumferential direction, and symmetric boundaries in the axial direction. The literature contains numerous papers describing various techniques for obtaining computer solutions to the equations.1,2
Hydrostatic Rayleigh Step Seal Figure 4 shows parameters for a hydrostatic Rayleigh step seal, which is a good candidate when the pressure differential is significant. The pressure differential is from the O.D. to I.D. Dimensionless performance curves are developed as a function of the radius ratio Ri/R0 and the step depth to clearance ratio δ/C (Figures 5 and 6). The load curves on Figure 5 are important for establishing the balance diameter that would supply an equilibrating closing Copyright © 1994 CRC Press, LLC
448
CRC Handbook of Lubrication and Tribology
FIGURE 4. Hydrostatic Rayleigh step seal.
FIGURE 5. Hydrostatic step seal. Load and leakage as a function of step depth/clearance ratio
load. Significant increases in leakage occur as the radius ratio is increased; as the step depth increases, the leakage increase is more moderate. The axial and moment stiffness of the film are shown on Figure 6. Maximum stiffness occurs when δ/C equals 0.5. Stiffness is an extremely important parameter because high values ensure against contact. Design values of 8/C should lie between 0.5 and 1. Negative moment stiffness values for δ/C = 0.25 and Ri/Ro = 0.7 imply that any seal ring tilt will result in further tilting and ultimate contact and failure; i.e., seal operation would be unstable. Damping values are greater for smaller step depths because more low clearance land area is exposed to normal velocities.
Sample Problem 1: Produce the interface performance for a Rayleigh-step hydrostatic seal with a radius ratio Ri/Ro = 0.85, an outside radius Ro = 1 inch, a film thickness, C = 0.0002 in. (5.08 x 10-6 m), a fluid viscosity µ, = 1 x 10-7 lb s/in2 (689 x 10-6 Pa s), δ/C = 1, and ∆p = 400 psig (2.76 x 106 Pa). Also, determine performance if the clearance closes to 0.0001 in (2.54 x 10-6 m). Table 1 lists values of the dimensionless parameters extracted from the curves for the two cases. Using definitions in the nomenclature, dimensional performance parameters can be produced. For example, the load capacity and flow are computed as follows:
Copyright © 1994 CRC Press, LLC
Volume III
449
FIGURE 6. Hydrostatic step seal, axial and moment stiffness and damping.
The results of the dimensionalizations are indicated on Table 2. The numbers indicate a reduction in stiffness as the clearance closes, a significant reduction in leakage, and an increase in damping. By considering two opposed flat surfaces, the following formula approximates the moment about the rotation axis due to viscous friction
For a nonuniform clearance, as with a step or tapered land, an average clearance can be utilized. With seals, it is not correct to determine fluid temperature rise by equating the viscous heat generation to the heat carried away by the fluid. Because of the low leakage of seals, this procedure would yield very high temperatures. The proper procedure is to conduct a complete heat transfer analysis, including the heat conducted to the opposed members and to the Copyright © 1994 CRC Press, LLC
450
CRC Handbook of Lubrication and Tribology
environment. Generally, an axisymmetric finite element analysis is sufficient. If distortions approach the same order of magnitude as the film thickness, corrective action is required.
Hydrostatic Tapered Land Seal Similar curves have been generated for a tapered land hydrostatic seal of a configuration s own on Figure 7. The curves are included as Figures 8 and 9. A comparison of the dimensionless parameters for the Rayleigh step and tapered land seals in Table 3 indicates at the Rayleigh step has greater load capacity and stiffness, but the tapered land seal has greater damping characteristics. The Rayleigh step seal is also more stable in that the range of negative moment stiffness is less, and that would make it more attractive than the tapered and. However, if properly designed and applied, both seals should perform well.
Hydrodynamic Rayleigh Step Seal Figure 10 shows a shrouded hydrodynamic Rayleigh step geometry with high pressure at the outer radius. This seal derives film stiffness from hydrodynamic action as opposed to hydrostatic action, and thus performance is dependent upon surface speed. The pressure effects do contribute to load and flow and to the other parameters to a lesser degree. To establish generality of the performance curves, the hydrodynamic or rotation effects and the hydrostatic or pressure effects can be treated separately and total performance obtained by superposition. The performance curves for the configuration shown in Figure 10 are indicated on Figures 11 and 12.
Sample Problem 2: Determine the interface performance of a hydrodynamic Rayleigh step seal as shown on Figure 10, with a rotational speed of 314 rad/s (3000 rpm), a radius x RiR/o= 0.85, an outside radius Ro=1, inch, a film thickness, C = 0.0002 in. (5.08 x10–6m), a fluid viscosity µ = 1 x 10-7 lb s/in.2 (689 x 10-6 Pa s), δ/C = 1, and ∆p = 400 psig (2.76 x 106 pa)
Copyright © 1994 CRC Press, LLC
Volume III
451
FIGURE 7. Hydrostatic tapered land seal
FIGURE 8. Hydrostatic tapered land seal. load and leakage.
The dimensionless parameters extracted from the performance curves are shown on Table 4. The hydrodynamic contributions are identified with subscript or superscript h, while hydrostatic values are primed. Using the definitions in the nomenclature and superposition for all performance except damping, the dimensional parameters in Table 5 were computed. The damping values are not dependent upon external pressure or rotational velocity, but only upon normal and angular velocities about orthogonal axes. Therefore, the values do not superpose. Table 5 indicates the performance parameters in English and SI units. Examples of the computation are as follows:
Hydrodynamic Tapered Land Seal Dimensionless performance curves have been generated for a tapered land hydrodynamic seal of a configuration shown in Figure 13. The curves are included as Figures 14 and 15. Copyright © 1994 CRC Press, LLC
452
CRC Handbook of Lubrication and Tribology
FIGURE 9. Hydrostatic tapered land seal. Axial and moment stiffness and damping.
FIGURE 10. Hydrodynamic Rayleigh step seal.
Copyright © 1994 CRC Press, LLC
Volume III
453
FIGURE 11. Hydrodynamic Rayleigh step seal. Load and leakage with rotation and pressure effects.
Hydrostatic Recess Seal The face configuration for the hydrostatic recess seal is shown on Figure 16. This seal is self energized, which means that the supply pressure to the recess comes from the fluid being sealed (see Figure 1). The recess in a hydrostatic seal is located significantly downstream from the diameter at which the high pressure resides. The recess is located a distance that is 60% of the face width from the high pressure on the outside diameter. Performance curves for the hydrostatic recess seal are shown on Figures 17 and 18, for a radius ratio of 0.8. The parameter ζ is an orifice variable. For the recess location shown on Figure 16, the minimum pressure, assuming a straight line pressure drop, would be 40% of the supply pressure, and the optimum recess pressure would then be 70% of the supply pressure to be half way between the minimum and maximum pressure that could occur at that location. This would allow the recess pressure the greatest latitude in providing stiffness as variations in clearance occur. Examination of the curves reveals that a value of ζ = 200 would meet this criterion. Another limitation for hydrostatic seals is the orifice diameter, which should not be small enough to allow clogging. The normally acceptable minimum value is 0.020 in. (508 µm). Sample Problem 3: Determine interfacial performance of a hydrostatic recess seal, with the general configuration of Figure 16, a radius ratio Ri/R0 = 0.8, an outside radius Ro = 1 in. (0.0254 m), a sealed pressure ps = 400 psig (2.76 x 106 Pa), a fluid viscosity µ = 1 x 10-7 lb s/in.2 (6.89 x 10-4 Pa s), a fluid mass density of 93.46 x 10-6 lb sec2/ in.4 (1000 kg/m3), an orifice diameter d” = 0.020 in. (5.08 x 10-4 m), a discharge coefficient CD = 0.6, and a value of the orifice coefficient ζ = 200. The reference pressure Pr is to be taken as unity. The first step in the analysis is to determine the operating film thickness from the equation for the orifice coefficient ζ. Copyright © 1994 CRC Press, LLC
454
CRC Handbook of Lubrication and Tribology
FIGURE 12. Hydrodynamic Rayleigh step seal. Stiffness and damping with rotation and pressure effects.
The dimensionless values, obtained from Figures 17 and 18, for a pressure gradient 400 psig and an orifice coefficient of 200, are as follows: The information is now available to calculate the dimensional performance numbers: Copyright © 1994 CRC Press, LLC
Volume III
455
FIGURE 13. Hydrodynamic tapered land seal
The results indicate high leakage flows. Unfortunately leakage is necessary for hydrostatic recess seals to develop stiffness. While hydrodynamic seals can gain stiffness as clearance loses and thus can operate with low leakage, the hydrostatic seal loses stiffness as the clearance closes because the recess to supply pressure ratio approaches 1 and stiffness approaches zero. Copyright © 1994 CRC Press, LLC
456
CRC Handbook of Lubrication and Tribology
FIGURE 14. Hydrodynamic tapered land seal. Load and leakage as a function of taper height/clearance ratio.
A reduction in recess size could reduce leakage, but at the expense of load capacity and stiffness. A hydrostatic seal can sometimes act as a pressure breakdown device for a downstream contact seal that prevents fluid egress.
Spiral Groove Seal The spiral groove seal, schematically shown on Figure 3, is being used extensively for gas seal applications, such as the principal seals for pipeline compressors. Data has been developed for a 6 in. (.1524 m) diameter seal with a radius ratio of 0.7, operating with air as the gas (viscosity = 2 x 10-9 16 - S/ln2 (1.379 x 105 Pa - 5)). The film thickness was maintained at a constant 0.0002 in. (5.08 x 10-6 m). Parameters to be optimized include groove angle β, circumferential land width/groove width ratio γ, groove depth δ, and dam radius Rm. Optimization was done on the basis of stiffness. Figure 19 shows the optimum geometric parameters as a function of the pressure gradient and the operating speed. The optimum groove angle p increases with pressure gradient, but decreases with operating speed. The greater the groove angle, the greater the mass flow that can be pumped through the grooves because the groove volume is larger. That is why greater angles are necessary at lower speeds. A higher pressure gradient would allow for an angle more oblique to the pressure flow than would a lower gradient and is an explanation for the increase in angle with pressure gradient. Higher groove depth 5 is necessary at lower speeds to allow more mass in the grooves and to counter the reduced hydrodynamic action. The groove depth also reduces with pressure gradient. Note however that the total variation in groove depth is small (the scale of the curves exaggerate their differences). The groove depth also is reasonably close to the operating film thickness with an average increase of about 30%. Copyright © 1994 CRC Press, LLC
Volume III
457
FIGURE 15. Hydrodynamic tapered land seal. Axial and moment stiffness and damping
The land width/groove ratio γ increases with pressure gradient and decreases with speed. Less groove is required as the pressure gradient increases. Because the groove angle is smaller at higher speeds, more groove width is required than at the lower speeds. The final parametric curve is the dam radius, Rm, which reduces slightly with pressure gradient and increases with speed. The greater the dam radius, the less radial width is available for the groove region. Less groove area is required at higher speeds because of the increased hydrodynamic action. The reduction in Rm with pressure gradient is explainable by the increase in groove angle. Performance parameters for optimized configurations are shown on Figure 20. They include interface load, leakage, axial stiffness, and power loss. Copyright © 1994 CRC Press, LLC
458
CRC Handbook of Lubrication and Tribology
FIGURE 15. Hydrodynamic tapered land seal. Axial and moment stiffness and damping
FIGURE 17. Hydrostatic recess seal. Load, leakage, and recess pressure.
Sample Problem 4: Determine geometric parameters and performance of a 6 in. (0.1524 m) diameter spiral groove seal operating at 20,000 rpm, with a pressure gradient of 725 psig (5 MPa). Air is the lubricant. The following values were obtained from the performance curves: 䊉 䊉
Groove angle, β = 37° Groove depth, δ = 6.5 x 10-6 m (0.000256 in.)
Copyright © 1994 CRC Press, LLC
Volume III
459
FIGURE 18. Hydrostatic recess seal. Stiffness and damping 䊉 䊉 䊉 䊉 䊉 䊉
Land/groove ratio, γ = 0.83 Dam radius, Rm = 5.81 x 10-2 m (2.287 in.) Interface load = 40,000 N (8993 lb) Leakage = 590 x 10-5 kg/s (0.013 lb/s) Axial stiffness = 1750 x 106 N/m (9.999 x 106 lb/in.) Power loss = 340 w (0.456 HP)
DYNAMIC RESPONSE
Shaft excursions are generally greater than the film thickness of fluid-film seals. Therefore the film must be stiff enough to maintain separation and avoid seal contact. Dynamic analysis is necessary to determine if the seal will operate without contact. Figure 21 shows a face-type seal that can be exposed to five degrees of freedom of the shaft; x, y, and z translations; and two rotations βX and αy. Considering small displacements, the following equations apply:
Copyright © 1994 CRC Press, LLC
460
CRC Handbook of Lubrication and Tribology
FIGURE 19. Spiral groove seal. Optimum geometric parameters.
The external forces and the moments are as follows: Forces Fluid-film forces Secondary seal friction force Secondary seal spring force Secondary seal damping force Spring force Hydraulic closing force
Moments Secondary seal friction moment Secondary seal spring moment Secondary seal damping moment Fluid-film moments Spring moments
Reference 10 describes how these forces and moments are represented. One form of dynamic analysis is to employ a computerized forward integration in time scheme such as Newmark’s method or the average acceleration method using procedures similar to those described by Shapiro.10 Applying Newmark’s procedures produces the following equation: Subscript i + 1 refers to the updated time step, while i references to the previous time step. The expression relates the displacement at the new time step to displacements, velocities, and accelerations at the prior time step. Once Ui+1 is obtained,Ui+1 and Üi+1 are obtained from
Copyright © 1994 CRC Press, LLC
Volume III
FIGURE 20. Spiral groove seal. Performance parameters.
FIGURE 21. Face seal dynamic parameters.
Copyright © 1994 CRC Press, LLC
461
462
CRC Handbook of Lubrication and Tribology
Thus, displacements, velocities, and accelerations are determined from the results of previous time steps. Initially these quantities equal zero. The mass and inertia properties of the seal ring and the location of the center of gravity are first computed. After computing all constants and matrix elements that are independent of time, the time step loop is initiated. Shaft motions are incremented first. Using updated shaft motions, the secondary seal friction is determined. Included are friction magnitudes and direction in the x, y, z, βx, and αy directions (see Figure 21), as well as the friction components that go into the stiffness and damping matrices and force vector. After the force vector is updated, Newmark’s method is applied, and the new seal displacements, velocities and accelerations are determined. Subsequent to these calculations, adjustments are made to these variables because of friction resistance. 䊉
䊉
If the velocity is zero and the absolute value of the applied forces exceed those of the friction forces, the motion continues. Otherwise it remains stopped. If the velocity is not zero, and the absolute value of the friction forces (or moments) exceeds those of the applied forces, and the applied force is in a direction opposite to the friction force, displacement of the seal ring is stopped and velocities and accelerations are nulled in that degree of freedom, i.e.;
Finally, seal motions and displacements are updated for the following time step. As an example of dynamics, an analysis was conducted of the fluid-film face seal described by Figure 21. The fluid-film stiffnesses used in the analysis were as follows: Axial stiffness = 73.5 x 106 N/m; angular stiffness = 48.5 x 103 N m/rad The principal variables are the shaft displacements and frequencies and the coefficient of friction of the secondary seal. Five dynamic cases were considered where displacements and friction were varied as summarized in Table 6. Other significant parameters are identified in Table 7. For case 1, the shaft was given vibration displacements of 0.0025 mm and angular vibrations of 0.0001 radians about orthogonal axes normal to the axes of shaft rotation. The friction coefficient of the secondary seal was 0.2. The seal ring tracked the runner without difficulty. The minimum film thickness for this case was 0.009 mm. Excitation amplitudes were doubled for case 2; the seal ring could not track, and contact occurred. The coefficient of friction of the secondary seal was reduced to 0.1 for case 3, with Copyright © 1994 CRC Press, LLC
Volume III
463
FIGURE 22. Film thickness vs. shaft revolutions.
amplitudes the same as for case 2. A Teflon-coated piston ring may have friction coefficients as low as 0.1. The seal ring tracked without difficulty, and displacements were in phase with nearly identical amplitudes as the forcing function from shaft vibrations. Similar results were noted for case 4, when amplitudes of translatory vibrations were increased to 0.0076 mm (semi peak-to-peak values) and misalignments increased to 0.0003 radians. This combination of axial and angular movements causes a total seal ring half peakto-peak displacement at the outer diameter of the rotating collar of 0.0238 mm, which is significantly greater than the axial equilibrium clearance of 0.012 mm. For case 5, the high amplitudes were retained, but the friction coefficient of the secondary seal was increased to a value of 0.15. Again, the seal ring tracked without difficulty. Figure 22 shows minimum film thickness as a function of shaft revolutions for case 5. The minimum film thickness is approximately 0.007 mm, which is a respectable value for a seal of this size and speed. The dynamic studies clearly point out the benefits of minimizing the friction characteristics of the secondary seal. It is concluded that, if the secondary seal friction coefficient could be maintained at 0.15 or below, excellent tracking characteristics would result. In addition to the friction characteristics of the secondary seal, other parameters that can be varied to improve dynamic performance of fluid-film seals are the following:
䊉 Mass and inertia of the seal ring. In general, the lower these values, the better will be the dynamic response.
Copyright © 1994 CRC Press, LLC
464
CRC Handbook of Lubrication and Tribology
FIGURE 23. Helium-buffered face seal.
䊉 Stiffness and damping characteristics of the fluid film. The greater the direct stiffness and damping, the better the dynamic response.
THERMOELASTIC DISTORTIONS
Pressures, temperatures, and heat sources can impose deformations deleterious to seal operation. It is important to prevent deformations that would significantly increase or decrease the film thickness. Most deformation analysis is accomplished by finite element computer codes, and for fluid-film seals axisymmetric models are adequate. As an example, consider the arrangement shown on Figure 23. It shows a buffer purge seal that was investigated for the oxidizer pump of the space shuttle main engine (SSME). The buffer purge seal consists of two opposed fluid-film face seals that mate against a single collar. The helium buffer fluid flows into the outer periphery of the rotating collar and through the clearance spaces between the seal rings and collar. A positive stiffness fluid-film interface maintains the clearance at design levels irrespective of the axial position of the collar. Important geometric and operating parameters are identified in Table 8. The seals are exposed to varying ambient pressures and large temperature gradients. Two separate finite element analyses were conducted: (1) a thermal analysis to establish the temperature distribution and thermal distortion; and (2) a pressure analysis to determine distortions and stresses due to the pressure boundary conditions. The two analyses were then superimposed to determine total distortions. Figure 24 shows the large temperature gradients Copyright © 1994 CRC Press, LLC
Volume III
FIGURE 24. Helium-buffered face seal. Temperature distribution and thermal distortions
465
Copyright © 1994 CRC Press, LLC
466
CRC Handbook of Lubrication and Tribology
that occur across the turbine side seal ring (approximately 200°C), while the oxygen side seal ring has a very low gradient. The high-temperature (turbine side) ring is subject to considerable thermal distortion, which produces a divergent clearance distribution in the direction of flow. This thermal distortion is further aggravated by distortion due to pressure. Summary of the deformations in Table 9 shows the divergent clearance distribution in the direction of flow, which is detrimental to performance. Since the design operating clearance is 0.012 mm, deformation of the hydrogen side ring of 0.0495 mm would exclude this seal geometry for the application.
MATERIALS
The important parameters in selecting materials for fluid-film seals are rubbing compatibility to minimize wear during start-stop operation, thermal conductivity, coefficient of thermal expansion to minimize distortions, and Young’s modulus to assure adequate strength for high-speed and high- or low-temperature applications. Carbon is a favorite for mechanical seal applications and works well in combination with a hard face. Some disadvantages of carbon are brittleness and low tensile strength. For fluid-film applications, a hydrodynamic geometry could be worn away relatively easily in the carbon. The geometry should be installed in the hard face. For water applications, a seal nose material of carbon graphite against a shoulder material of bronze, NI-resist, tungsten carbide, or chrome-plated metals have been found to be effective. In addition, tungsten carbide against itself works well. Silicon nitride has also been utilized for its excellent wearing qualities. For oil applications, the shoulder materials can be extended to include 400 series stainless steel hardened to Rockwell C-50. For gas seals, the primary sealing rings are often carbon mating against a chrome-plated 316 stainless steel, chrome-plated Invar, tungsten carbide, and solid titanium carbide.
NOMENCLATURE
C = film thickness CD = orifice discharge coefficient Do = orifice diameter dxx, d’xx, dhxx = moment damping, total, hydrostatic, hydrodynamic D = damping matrix D’xx = dimensionless hydrostatic moment damping = d’xxC3/(12µRo6) Dhxx= dimensionless hydrodynamic moment damping = dhxx C2/(12µRo5) dzz, d’zz, dhzz = axial damping; total, hydrostatic, hydrodynamic D’zz = dimensionless hydrostatic axial damping = d’zzC3/(12µ,Ro4) Dhzz = dimensionless hydrodynamic axial damping = dhzzC3/(12µRo4) Fi = force on seal ring in i direction, i = x, y, z It = transverse moment of inertia of seal ring kxx, k’xx, khxx, krxx, = moment stiffness, total, hydrostatic, hydrodynamic, recess K’xx = dimensionless hydrostatic moment stiffness = k’xxC/(∆pRo4) Copyright © 1994 CRC Press, LLC
Volume III
Khxx = dimensionless hydrodynamic moment stiffness = khxxC2/(6µωR5o) Krxx = dimensionless hydrostatic recess moment stiffness = k’xxC/(poR4o) kzz k’zz, khzz, krxx - axial stiffness, total, hydrostatic, hydrodynamic, recess K = stiffness matrix K’zz = dimensionless hydrostatic axial stiffness = k’zz C/(∆pR2o) Khzz. = dimensionless hydrodynamic axial stiffness = khzzC3/(6µωR4o) Krzz = dimensionless hydrostatic recess axial stiffness = k’zzC/(poR2o) m = mass of seal ring M = mass matrix Mxx = Myy = moment on seal ring about x and y axis po = reference pressure pr = recess pressure pr = dimensionless recess pressure = pr/po q, q’, qh,qr = leakage, total, hydrostatic, hydrodynamic, recess Q’ = dimensionless hydrostatic leakage = 12µq’/(∆pC3) Qh = dimensionless hydrodynamic leakage = 2qh/(R2oωC) Qr = dimensionless hydrostatic recess seal leakage = 12µqr/(poC3) Ro, Ri = outside, inside radius Rm = Spiral groove dam radius Ri/Ro = radius ratio Rs = secondary seal radius Rsp = radius to spring Rso = outer radius of secondary seal U = displacement vector w, w’, wh = interface load: total, hydrostatic, hydrodynamic W’ = dimensionless hydrostatic load = w’/(∆pR2o) Wh = dimensionless hydrodynamic load = whC2/(6µωR4o) Wr = dimensionless hydrostatic recess load =wr/(poR2o) x, y, z = displacement along x, y, and z axes αy = rotation angle about y axis β = groove angle βx = rotation angle about x axis δ = step, taper or groove depth ∆p = pressure gradient ∆t = time increment γ = land/groove ratio ρ = fluid mass density µ = absolute viscosity ω = rotational speed ζ = orifice coefficient for hydrostatic recess seal=
Copyright © 1994 CRC Press, LLC
467
468
REFERENCES 1.
2.
3. 4.
5.
6.
7.
8.
9.
10. 11.
12.
13. 14.
CRC Handbook of Lubrication and Tribology
Castelli, V. and Pirvics, J., Review of methods in gas bearing film analysis, J. Lubr. Technol., 90, Oct. 1968, 777. Castelli, V. and Shapiro, W., Improved method for numerical solutions of the general incompressible fluid film lubrication problems, J. Lubr. Technol., 89, April 1967, 211. Shapiro, W., Water lubricated thrust bearings for a helium circulator, Bearing and Seal Design in Nuclear Power Machinery, American Society of Mechanical Engineers, New York, June 1967, 140. Shapiro, W., Concepts for deep submergence hydrostatic shaft seals, Lubr. Eng., 24(7), July 1968. Shapiro, W. and Colsher, R., Selection, analysis and preliminary design of a steamlubricated, steam turbine, shaft seal, ASLE Trans., 14(3), 226, 1971. Shapiro, W. and Colsher, R., Steady-state and dynamic analysis of a jet engine, gaslubricated shaft seal, ASLE Trans., 17(3), 190, 1974. Shapiro, W., Walowit, J., and Jones, H. F., Analysis of spiral-groove face seals for liquid oxygen, ASLE Trans., 27(3), 177, 1984. Artiles, A., Shapiro, W., and Jones, H. F., Design analysis of Rayleigh-step, floatingring seals, ASLE Trans., 27(4), 321, 1984. Hamm, R. and Shapiro, W., Testing of helium buffered, Rayleigh-step, floating-ring seals, Lubr. Eng., 43(5), 376, 1987. Shapiro, W., Dynamic analysis of contact face seals, SAE Trans., September 1987. Shapiro, W., Lee, C, and Jones, H. F., Analysis and design of a gas lubricated, sectored, floating ring seal, J. Tribol., 110, July 1988, 525. Lebeck, A. O., Principles and Design of Mechanical Face Seals, John Wiley & Sons, New York, 1991. Muijderman, E. A., Spiral Groove Bearings, Springer-Verlag, New York, 1966. Smalley, A. J., The narrow groove theory of spiral grooved gas bearings: development and application of a generalized formulation for numerical solution, J. Lubr. Technol., 94(1), 86, 1972.
Copyright © 1994 CRC Press, LLC
ROLLING BEARING FATIGUE LIFE
Charles A. Moyer and William J. Denier
INTRODUCTION
Bearing selection is appropriate and successful if the bearing provides the performance expected in the application. This chapter will address the information required to predict this performance, primarily in terms of fatigue life. Life prediction also depends on several other key components: bearing ratings, actual load or load spectrum the bearing will endure, and the operating conditions under which a bearing must operate. We will address these factors and their interactions as they pertain to all rolling element bearings. In Volume II of the Handbook of Lubrication, Theory and Practice of Tribology, fatigue life calculations are not specifically included. However, the Rolling Element Bearings chapter1 does give a complete description of rolling bearing types and bearing materials. There is also information on rolling bearing theory, contact stresses, friction, rolling bearing rating equations, static capacity and bearing lubrication. In addition, Volume II includes consideration of rolling bearing operating conditions of load, speed, temperature, alignment, clearance, mounting, and an extended explanation of the role of various lubricant types. Finally, a resume of bearing failure modes is presented to help in proper application of a bearing to a particular environment.
THE ROLE OF BEARING RATINGS
Rating equations as shown in the newest American Rating Standards issued in July 1990 by the American National Standards Institute (ANSI) and the Anti-Friction Bearing Manufacturer’s Association (AFBMA),2,3 and in the ISO Standard issued in December 1990 by the International Standards Organization (ISO)4 are essentially the same equations as given in Volume II but from standards of the 1972 through 1978 era. Ball Bearing Ratings Radial rating equations, taken from Reference 2, for radial and angular contact ball bearings when the ball diameter Dw is equal to or less than 25.4 mm (1 in.), are (Symbols are defined in the Nomenclature Section.) When Dw is larger than 25.4 mm (1 in.) diameter, then the radial ratings are
Thrust rating equations for thrust ball bearings taken from Reference 2 are
0-8493-3903-0/94/$0.00 + $.50 Copyright © 1994 CRC Press, LLC © 1994 by CRC Press, Inc.
469
CRC Handbook of Lubrication and Tribology
470
For Equations 3a and 3b, ball diameter is equal to or less than 25.4 mm (1 in.) For bearings with balls larger than 25.4 mm (1 in.) diameter, the thrust rating equations are
Within the ball bearing standard Reference 2, tables are given for all the Fcm values needed in the above equations. These will result in higher ratings (1.3 times) than were in the 1978 ANSI/AFBMA ball bearing standard referred to in the bearing chapter, Reference 1.
Roller Bearing Ratings For radial roller bearings, comparable rating equations from Reference 3 are For thrust roller bearings, rating equations from Reference 3 are
Values of Fcm are given in Reference 3 and roller bearing ratings are increased by 1.1 times, comparing the 1978 to 1990 ANSI/AFBMA Roller Bearing Standards. The increases of ratings have been significant over the last 15 years (1978 to 1992), primarily because of the much lower number and smaller size of nonmetallic inclusions in quality bearing steels. Since these ratings are essentially consensus efforts that represent neither maximum nor minimum values that can be attained, the designer is recommended to review manufacturers’ most recent bearing handbooks and their specifics for more accurate life estimates. In spite of the significant theoretical background as reviewed in Reference 1, buried within the rating equations are factors that require significant experimental data. The first factor is the material constant within Fcm, often described as the empirical fatigue constant. The second factor is the load-life exponent or the experimentally determined relationship of stress level to fatigue life. Both of these factors have been derived over the years by extensive bearing fatigue testing. Improvements in material quality, surface roughness, and dimensional variability, especially over the last 10 years or so, have impacted the usual numbers used for these first two factors, although for reasonable engineering approximations the present values are considered acceptable. The third factor is the scatter or dispersion of the fatigue data (usually called the Weibull slope or Weibull dispersion statistical parameter). While the updating of these factors would provide more current standard rating values, such changes have not been made in the bearing standards. Contact with the manufacturers of specific Copyright © 1994 CRC Press, LLC
Volume III
471
bearing types will establish the latest factors they have available and thus the more realistic ratings.
THE LOAD-LIFE RELATIONSHIP FOR BEARINGS
Basic rating life, L-10, for either radial or thrust bearings is determined within the standards2–4 by this simple equation:
For ball bearings n = 3, for roller bearings n = 10/3. When F equals C, the L-10 life calculates to be one million revolutions. At 500 rpm, L10 is only 33 1/3 hours. In fact, depending on bearing design, this L-10 generally would be above the elastic deformation limit for the bearing. Bearing ratings within the standards are in reality reference loads only used to calculate L-10 life for bearings using Equation 8 and load ranges recommended in Table 1. Values of the load-life exponent n were determined experimentally. For deep-groove ball bearings, Lieblein and Zelen5 indicated that the exponent 3 was acceptable for most of 213 test groups (4948 bearings) from four manufacturers. Early tests of line contact to determine the load-life exponent for rollers6,7 indicated n = 4. However, with rollers and/or raceways crowned or profiled for “elliptical contact” or “optimized contact”, under low load and no misalignment, an exponent of 10/3 has been confirmed with recent tests8 and has been used in the Roller Bearing Standard for many years. ANSI/AFBMA Standard 11–903 recognizes that even the crowned or profiled contact may have truncation and end-of-contact stress concentration at high loads. However, since actual roller bearing contact lengths and end modifications are not standardized, how specific bearings are influenced by truncation must be obtained from the bearing manufacturer. Table 1 is a summary of load in percent of rating “C” vs. life ratio for both ball and roller bearings. Life ratio is taken from Equation 8. If load F is 10% of rating C, then OF equals 1.0/0.1 or 10. If this load ratio is raised to the power 3, the life ratio resulting is 1000, as shown in Table 1 for the normal load for ball bearings. For ball bearings, from 4% C to 40% C covers over 975 to 1 difference in life; for roller bearings, from 7% C to 50% C covers over 700 to 1 difference in life. Therefore, it is important that load be determined carefully, especially when a range of loads are to be expected. If bearing loading exceeds the 40 and 50% limits suggested in Table 1, even for a portion of a loading cycle, then the possibility of damaging plastic deformation should be considered. Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
472
In the event of multiple loads, actual time under each load is important. The following equation shows how to use variable loads, speeds and time to obtain a weighted average load Fwt. In Equation 9, speed S is in rpm; very often for comparison purposes S in the denominator is chosen as 500 rpm. The general equation is
F1, F2, Fi equal the actual loads, either all equivalent radial or all equivalent thrust, that are applied to a bearing for time fractions T1, T2, Ti, such that T1 + T2 + Ti equals 1.0. S1, S2 and Si represent the corresponding speed in rpm at each load and time fraction. Fwt can be used within Equation 8 to determine an approximate L-10 life under any given histogram of loading. Caution is advised in using Equation 8 to consider too wide a range of loading, especially in the heavy load region where permanent plastic deformation within the bearing would invalidate Equation 8.
FACTORS INFLUENCING RATING LIFE
For Equation 8, the L-10 life calculated is called rating life, which means the bearing is running under operating conditions that provide an adequate lubricant film based on sufficient lubricant viscosity at the operating temperature and the speed of operation. Within the bearing standards, adequate lubricant film is defined as that film equal to or greater than the composite roughness of the two contacting surfaces. That is, the elastohydrodynamic lubricant (EHL) film divided by composite roughness determines the lambda ratio A. When A is larger than 1.0, subsurface contact fatigue limited by the bearing material is expected to occur, and rating life should be realized. The ANSI/AFBMA Standards also state mat kinematic viscosity at operating temperature of less than 13 cSt may not be adequate for rating life, and that some minimum speed of operation may be required. These were combined as DN (bearing pitch diameter D in mm times speed N in rpm) with a lower limit of 10,000. Figure 1 indicates what this minimum speed means for a range of bearing pitch diameters. These two conditions combined (minimum 13 cSt viscosity and minimum surface velocity from the DN value) indicate an EHL film of less than 0.1 mm (4 microinches) could occur for either a ball or roller bearing within the size range of Figure 1. Considering how A depends on surface roughness, load, bearing size, and design, the standards criterion on Figure 1 will in itself only provide a rough estimate that a given operating condition will provide rating life.
LIFE ADJUSTMENT FROM RATING CONDITIONS
In 1972, ANSI/AFBMA Bearing Standards910 incorporated the concept of life adjustment factors. The concept recognized the interrelationship of load, life, and reliability. As life is related to load by the appropriate load-life exponent, life is related to reliability by the Weibull dispersion parameter (m) or Weibull slope. There is some evidence that Weibull slope increases with load,8 so this may need to be taken into account in using the interrelationships of load, life, and reliability now assumed. To determine adjusted or application life (L-na) the following equation was provided in References 9 and 10 to introduce the three factors for modifying the fatigue rating life:
Copyright © 1994 CRC Press, LLC
Volume III
473
FIGURE 1. Plot of ANSI/AFBMA minimum limit of DN; bearing size (mm) times speed (rpm) equals 10,000 DN.2,3
where: L-na = calculated raring life L-10 adjusted by the combination of the factors:
a1 a2 a3
= reliability factor for other than 90% reliability, = material factor for other than standard or nominal steels in current use, and = application or environmental conditions factor, primarily recognizing the lubricant condition within a bearing. Also includes alignment, internal load distribution, or load zone and contamination, liquid or solid.
The standards caution that when expected life L-na is lower because of poor lubrication conditions with a3 less than 1.0 (thin film operation with significant surface asperity interaction), then an increase in the second factor a2 from a change to better material should not be expected to counteract such loss in life. For many, classical rolling contact fatigue is defined as rating life, as calculated with Equation 8. This involves initiation of fatigue cracks at subsurface depths corresponding to the range of orthogonal shear stress (τxz), the usual criterion, or maximum shear (τ45), or even the von Mises yield parameter. When lubricant conditions deteriorate, fatigue initiation sites can move to near surface or even on the surface so that the operating conditions assumed for rating life calculations are no longer valid. Considerable experimental and analytical work has been done to understand the influence of these three life-adjustment factors on bearing life and almost every bearing catalog now has specific versions of Equations 10 and 11. Reliability Factor a1 Reliability is the percent of a group of identical bearings that is expected to reach some specified life when they are operating under the same load and identical conditions. The reliability of a single bearing is the probability it will attain or exceed the specified life. For rating life, L-10, there is 90% probability that a bearing will achieve this life. The reliability factor a1as given in References 2 and 3 and many bearing catalogues is based on a Weibull slope value of 1.5. The factor can be determined with the equation:
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
474
FIGURE 2. Variation in the reliability factor a1 for different values of Weibull slope, m. (From Mover, C. A., STLE S. P. 31, April, 1991, 89.)
where:
a1= R= m=
life factor (ratio) used in Equation 10 or 11 reliability expressed as percent survival Weibull slope; its usual value is 1.5, but other values, if warranted, can be used in Equation 12.
Figure 2 indicates how the a, value can vary over the probability of survival range of 90 to 99% (the plot is a segment of Weibull probability paper with the abscissa a normal log scale). This figure includes Weibull slopes of 1.0, 2.0, and 3.0 besides the 1.5 value. This range of Weibull slope depends on the bearing type, operating conditions, and stress level. An indication of how the Weibull slope can change with stress in Figure 3 is based on tests on a 19-mm bore tapered roller bearing with Hertzian contact stresses of 1420 MPa to 2365 MPa.8 The line is a regression with 0.83 correlation coefficient tit. In evaluating values for the Weibull slope, Tallian11 included standardized lives of 2520 bearings from 93 test groups, of which about 91% were ball bearings (i.e., point contact). In the survival range between 40 to 90%, the fatigue failures fit the Weibull distribution very well with a slope about 1.0, very close to the 1.125 Weibull value that is part of the experimental data input in the standard rating equations given above. In the survival range of 95 to 99.9%, a deviant line close to a Weibull slope of 1.5 was put through the lowest standardized bearing lives (Figure 4). Thus for ball bearings a Weibull slope of 1.5 is suitable for a1 in this higher probability of survival range. A collection of tapered roller bearings representing standard production lots over about a 5-year period has been reported.8 Figure 5 shows the results of these fatigue tests on a Weibull paper segment covering 70 to 99.99% survival. Relative life is the ratio of experimental life to expected life based on bearing ratings and test loads of each bearing lot. Of a composite sample of 3465 bearings, 2095 fatigued to a 6 mm squared spall or larger. The Weibull slope in Figure 5 is 1.6, with little deviation from the calculated population line. Thus 1.5 seems acceptable for use in a, for roller (line contact) bearings and is perhaps conservative. For roller bearings, Equation 12 can be used, with m equal to 1.5, from about 99.95 to 50% reliability, i.e., a1 equals 0.03 to 3.5, based on the data available. For ball bearings, Equation Copyright © 1994 CRC Press, LLC
Volume III
475
FIGURE 3. Relation of hertzian contact stress to Weibull slope, m. Correlation coefficient of line = 0.83. (From Moyer, C. A., STLE S. P. 31, April, 1991, 89.)
FIGURE 4. Probability of survival vs. standardized bearing lives from 95 to 99.97% surviving. Deviant line for primarily ball bearings, m = 1.5. (From Tallian, T., ASLE Trans., 5, 183, 1962.)
Copyright © 1994 CRC Press, LLC
476
CRC Handbook of Lubrication and Tribology
FIGURE 5. Composite Weibull distribution of 3465 tapered roller bearings ranging from 12.0 mm to 85.0 mm bore, all run with SAE 20 mineral oil and 38°C inlet oil temperature. Rationalized to sample L-10 equal to 1.0. (From Moyer, C. A., STLE S. P. 31, April, 1991, 89.)
12, with m equal to 1.5, is suitable from 99.9 to 90% reliability. Figure 4 also indicates that there may be a minimum life at about a1 equal to 0.04, based on the two earliest failures. The earliest failure in Figure 5 is below the line but is not an indication of minimum life. From a statistical standpoint, the trends in estimating likelihood of survival at these extremes have fairly low confidence levels and should be used with extreme caution.
Material Factor a2 When the material factor was considered in the bearing standards in the mid 1970s, the fatigue damage was inclusion initiated and fatigue limits for carburized steels such as AISI 8620 and 4620 or through hardened AISI 52100 were based on the cleanness level of the conventional steel then used. Significantly cleaner versions of these steels were available then as now, such as vacuum-arc remelted (VAR). Other melting practices such as vacuum induction melted vacuum-arc remelted (VIMVAR) have been in use for critical life applications, particularly in the aerospace industry. Many of these practices involve hightemperature steels such as M-50 or M50NIL, CBS-600, and CBS-1000M so that bearings could be operated in the 150°C (300°F) to 475°C (890°F) temperature range. The bearing standards give no life values for a2, but in 1971 the ASME design guide12 included some relative values to be used with its material factor “D”. Air melt steels have improved significantly since, and more current experience is provided in Chapter 3 of the recently issued STLE Life Factors for Rolling Bearings.13 This new book includes information on material chemistries, related life factors, and discussions on melting practices, surface processing, hardness, residual stress, coatings, and surface treatments. The chapter also covers ceramic materials such as silicon nitride and hybrid bearings (ferrous and nonferrous components) and ends with 140 material-related references. The substantially cleaner steels in the present standard bearings are now recognized by the revised ratings, so the bearing user needs to contact the manufacturer to determine specific ratings and new a2, factors for current special steels. Values of a2 range from 0.6 to over 6 for different material chemistries, melting practices (cleanness levels), metal working, heat Copyright © 1994 CRC Press, LLC
Volume III
477
treatment, and surface modification.13 For standard vacuum-processed steels, the a2 life factor is 1.0 for conventional applications. Each manufacturer is responsible for developing and providing substantiating data for material fatigue life in his specific bearing design.
Application/Environmental Conditions Factor a3 Bearing standards have not given values for a3 but define in general the conditions that deviate from those that produce rating life, as reflected in Equation 8. Application factor a3 can encompass all environmental conditions that impact bearing performance: temperature, alignment, in-and-out-of-plane loading distortion, liquid and solid contaminants, clearance settings and preload (internal load distributions), and the lubricant. Table 2 lists the primary operating variables that can impact the a3 factor. Of these the lubricant probably involves the most complicating factors.
Lubricant-Topography Subfactors The lubricant can be a mineral, synthetic or other fluid, with properties of viscosity, surface wetting ability, surface tension, chemical surface reactivity, and shear strain-stress relationships that are enhanced or degraded by the additives used and the particular lubricant formulation. Lubricant properties, combined with load and surface velocities, are used to determine the EHL film within a bearing that tends to separate the surfaces. The lubricant film and surface roughness determine the lambda ratio A that has been used more than 20 years to describe the lubricating condition within a bearing: where: h=
σ=
σq= σa=
film thickness within a bearing determined by primarily a central film thickness equation, although minimum film equations also are used; composite roughness of the contacting surfaces (usually measured in the rolling direction), σq or σa; [((Rq1)2+ (Rq2)2]1/2when root mean square roughness parameters are used in the measurement; and [Ra1+ Ra2] when arithmetic average parameters are used.
Copyright © 1994 CRC Press, LLC
478
CRC Handbook of Lubrication and Tribology
For central or minimum lubricant film calculations, STLE Life Factors for Rolling Bearings,13 Chapter 5 can be used. To improve the film calculation to better match actual measured films, a thermal correction factor, a starvation factor, and an ellipticity factor can be considered.14– 17 A review of the lambda ratio is given in Reference 18 along with the possibilities of functional filtering done by the actual contacts within a bearing depending on their size, especially in the rolling direction. For composite roughness, R,, is preferred because of the compatibility of combining this with other roughness parameters such as asperity slope or average wave length (spacing between asperities). If only R2 values are available, the recommended summation is indicated under Equation 13. Because of the trend of decreasing fatigue life that accompanies a reduced lambda ratio (A), especially with A less than 1.0, or increased life with increased A, the lambda ratio has been used to develop a3 life factors. Most successful attempts to develop a3 factors have been with one specific type bearing.18–22 In these and others a relationship between A and fatigue life L-na is represented by: where x = a unique constant for each test group, and y = an experimental exponent that has ranged from about 0.5 to over 1.0. The specific equations developed following Equation 14 were each only valid for a specific range of A values. For example, Reference 19 covers A from about 0.05 to 0.40, Reference 20 from 0.5 to 18.0, and Reference 21 about 0.05 to 1.10. While the combined data in Figure 6 from Reference 23 (including References 19 through 22) indicate a distinct trend, scatter in the region below A equal to 1.0 dictates caution in using any average curve to estimate an a3 value. Again, the bearing manufacturer should be consulted for appropriate a3 factors for specific bearing types based on their own substantiating data.
Temperature-Speed Subfactors Temperature within a bearing is influenced by the load and speed as a reflection of heat generation Qgen: where S = rotational speed and M = bearing running torque. Within M, speed is also present, along with load and viscosity (somewhat as in EHL equations), so that roughly Qgen can be related to speed (S), viscosity (µ) and load (F) by: This equation simply recognizes how actual bearing temperature depends on increases of speed, load, and viscosity, and that their contribution to increasing temperature is not the same.24 Lubricant churning, lubricant types, e.g., grease vs. oil or lubricant sumps, arid other rolling/sliding contacts such as gears may also influence the bearing temperature and related operating conditions. Also, extremes of speed contribute centrifugal force effects that can cause both changes in internal loading and wear from lubricant starvation of the inner bearing ring and rolling element retainer. Load Zone—Alignment Subfactors Internal clearance or preload change can cause internal load variations simply by shifting the loading on each rolling element. “Pure” radial load is usually defined as 180° load zone,
Copyright © 1994 CRC Press, LLC
Volume III
479
FIGURE 6. Relation of lambda ratio A and relative life as N(A)/N-catalog. The solid curves are for rms asperity slopes, (1) 2°, (2) 48’, and (3) 3° 30’. (From Tallian, T. E., ASME Trans., Lubr. Technol, 103, 509, 1981.)
and one half of the rolling elements are considered to be carrying the load (Figure 7). Equations for calculating load on the most heavily loaded rolling element are given in Reference 1, (“Rolling Element Bearings”, p. 512). As clearance is increased, the top roller carries more of the load, and fatigue life will decrease as shown by the right portion of the curve in Figure 7. A slight preload as indicated will actually give increased life, but higher preload will result in a drastic loss of life as shown by the severely dropping line. Misalignment can occur from shaft bending or mismatch of the components supporting the bearing inner and outer races. Misalignment within a bearing is usually measured by the angle formed by the center line of the outer bearing ring to the center line of the inner ring as shown in Figure 8. Table 4.3 from Reference 13 gives allowable misalignment for minimal life reduction for various bearing types. Misalignment influences fatigue life by load level, i.e., under higher loads the greater total elastic deformation may produce a more relative uniform stress along the contact length than under lower loads. Misalignment is also influenced by the contact, which changes with bearing type and internal contact design. It is difficult to produce a single curve that represents the relation between misalignment and life. Figure 9 from Reference 25 shows relative life vs. misalignment for one type of cylindrical bearing. Curves for other bearing types are given in Chapter 4 of Reference 13. Contaminants—Liquid and Solid Subfactors Contaminants within a bearing can cause considerable damage other than fatigue, as illustrated in Reference 1 (Table 18, p. 533). Rough running, excessive friction, overheating,
Copyright © 1994 CRC Press, LLC
480
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
FIGURE 7. Percent rated life vs. load zone varying from endplay (radial clearance) to heavy preload.
Volume III
481
FIGURE 8. Bearing misalignment, the angle formed when the outer race axis is not aligned with the inner race axis.
FIGURE 9. Approximate misalignment life reduction (L-na/L-10) due to misalignment 0 to 0.005 rad (0 to about 17’). (From Denier, W. J., SAE 700559, Society of Automotive Engineers, New York, 1970.)
Copyright © 1994 CRC Press, LLC
482
CRC Handbook of Lubrication and Tribology
FIGURE 10. Effect of water concentration (PPM) on roller bearing life, with life at 100 PPM used as reference (L = 1.0). (From Cantley, R. E., SIZE Trans., 20, 244, 1977.
and wear can occur from contaminants so that proper filtration and sealing are important, the extent needed depending on bearing size and type. Acid formation from lubricant breakdown and the presence of water can be especially damaging. An indication of the impact from water is shown in Figure 10.26 Solids, such as steel wear debris and fatigue or cutting fragments, are also damaging in their ability either to accumulate in the inlet region of a bearing contact and cause lubricant starvation27 or to cause denting large enough to be stress raisers and initiate fatigue. Reference 28 summarizes recent work concerning debris influence on bearing performance. Especially for bearings with small contact lengths (below approximately 5 mm), the lubricant system should be as clean as realistically possible. If debris enters the system, filtering helps but will not return performance life to that obtained with clean systems. Under a debris environment, bearings with smaller contact have shown improved life with 3-micron filters.29–30 On the other hand, larger bearings (contact lengths roughly above IS mm) appear more tolerant to debris31 and performance may not improve by going to a filter below 40 micron.28 Even starting with a meticulously clean system, debris can be generated within the numerous rolling/sliding contacts and break loose from other surfaces. Many particles can be formed, their numbers fitting an essentially exponential distribution commonly ranging from only a few large particles (over 50 or 100 µm in size) to millions of small particles (less than 5 µ in size). The fine, smallest particles cause mild wear. The intermediate (5 to 40 µ) cause wear, directly related to the amount of debris present, and for small bearings can cause dent sites leading to fatigue. The earliest fatigue-related failures seem directly related to denting caused by the largest particles that are able to pass through the bearing contacts. From a fatigue standpoint, a filter should remove all the large particles that can cause life-limiting dents. Early experimental work in this area bases conclusions on testing of artificially induced denting by a diamond indenter which fractured the surface and left a work-hardened rim about the depression.32 The fatiguing induced may not be related to field applications of bearings where the majority of experiences have noted smoother indentations without embedment or hard protruding rims. This is generally the case with frangible or softer contaminants. Examination of hardware returned from field testing will provide the best evidence of the nature of a life-limiting contamination. Indentations induced in an actual hertzian rolling contact generally differ from those caused by an isolated indenter pressed into an otherwise unconstrained surface. Bearings with Copyright © 1994 CRC Press, LLC
Volume III
483
smaller hertzian contacts are more sensitive to the effects of hard particulate contaminants than are most roller bearings and larger size ball bearings. Also, some applications and operating conditions may be more or less sensitive to the effects of contaminants. Of particular importance is the class of lubricant and its additive package. Since many bearings are grease lubricated, it is well to recognize the beneficial capability of the grease volume outside the bearing contacts to trap foreign material. Leenders and Houpert33 and Tripp et al.34 demonstrated that surface chemistry characteristics of the lubricant play a significant role in delaying or preventing the onset of surface nucleated fatigue and wear. This is an important consideration in assessing liquid and total system contaminants. A practical specification establishing definitions and limit of contaminants should include the following: 1. 2. 3. 4. 5.
Total quantity (by weight) per bearing size (or contact) Quantity (by particle size) for individual objectional debris materials Limits of particle sizes and hardness Intolerable means of access to a bearing Durability and/or replenishment to be tolerated
Reference to bearing manufacturers, catalogs and application engineers can provide guidance in these areas of concern. There are already available sealing arrangements and purging lubrication systems or procedures for minimizing and/or controlling contamination problems. For example, baffles or riffles may be built into the housing or sump. Utilization of seals in series, with purging lubrication plans, are being used to prevent harmful solid and fluid contaminant ingression.
Fretting Wear Fretting is wear occurring between two surfaces having oscillatory relative motion of small amplitude. A prime example is an inner ring on a rotating shaft continuously subjected to fretting under a wide range of conditions. Considering the higher loads resulting from rating increases, it is important to review the interference fit levels used for newer applications. Poor fitting practices for inner or outer rings can lead to excessive fretting wear debris or even fatigue.
Fatigue Spall Criteria Subfactors Very often, for less demanding applications, bearings can be run beyond the first evidence of fatigue. Under essentially mild operating conditions with moderate temperature, reasonable speeds, and lubricant films within a bearing that provide A values above 1.0, it may be possible to run until 10% or so of the raceway surface is spalled. This extended life can more than double the life from the first evidence of fatigue. This life extension is very dependent on the application, whether noise or vibration will cause problems, and whether flaked off material will interfere with functioning of other machine elements. Also bearing material may be a factor, e.g., through hardened steel and ceramic, especially for the inner ring, may crack through the cross section and lead to catastrophic failure. Carburized material, primarily from compressive residual hoop stresses, will resist such cracking except under excessively high loads and press fits. Under thin film conditions, especially at higher temperatures and loads considered heavy by Table 1, fatigue propagation may proceed at a rapid pace and no useful life will exist beyond the first detection of a fatigue spall. For thin film conditions, with lambda ratio less than 1.0, fatigue initiation may be surface related, occur early, and propagate rapidly. A review of failure modes helps explain some of these differences.1,13
Copyright © 1994 CRC Press, LLC
484
CRC Handbook of Lubrication and Tribology
INFLUENCE OF SURFACE TOPOGRAPHY
In some applications, surface nucleated micro-spalling is encountered, indicating the need for better surface finishes and/or increased lubricant viscosity and improved boundary lubrication properties. Increased bearing dynamic capacities encourage the selection of smaller bearings and shafts, operating at higher stress levels and deflections with the potential for accelerated inner ring/shaft fret wear. While micro-spalling may be associated with low lambda ratios alone, it is often the result of aggravated misalignment which generates significant slip in the abberated hertzian contact.25 Extensive studies of elastohydrodynamic contacts have demonstrated the need for more meaningful descriptions of the metallic surfaces.33,35 No longer is a meter reading from the typical stylus trace sufficient.35–38 Finishing of modern rolling element bearing surfaces is usually a two-step process which requires a statistical topographical model based on gaussian or non-gaussian distributions (asperity reduction/flattening due to superfinishing after initial grinding).38,39 Also significant are the stylus tip dimensions and the length of trace (cutoff ),18,35,40 as discussed by Thomas,40 Mover and Bahney,18 and Leaver et al.35 To make practical the utilization of modern EHL theory and new bearing catalogs and standards, the following are needed: 1.
2. 3.
4.
5.
6.
Incorporation into ISO/ANSI Load Rating Standards of new specifications for gaussian or nongaussian surface topography/surface finish definitions. These must be expanded to consider both axial and circumferential traces. More detailed surface-tracing capability with stylus tip instruments and cutoff length selected for the type and size of bearing being considered. (Typical bearing ground surface shown in Figure 11.) Mathematical models which incorporate realistic concepts of actual bearing surfaces (large, nearly flat asperities separated by deep valleys). An example of such a superfinished bearing surface is Figure 12. On a one-to-one scale, the valleys and peaks of the topography are mere undulations (Figure 13) subject to significant flattening under operating stresses of 1380 MPa (200 KSI) and more. Under these conditions, elastic deformations are several times greater than the average peak to valley distances. With this more realistic depiction of the contact areas, it can be appreciated that EHL films are enhanced by surface chemistry effects at the lubricant-to-metal interface. Continuous films up to 0.25 µm (10 µ in.) thick found in loaded hertzian contacts call for new studies of fluid rheological and chemical properties and their influence on operating life.33,34 These enhanced descriptions of bearing surfaces help explain experience in field applications where bearing life exceeded that predicted. It must be appreciated that occasional asperity contact in a well-lubricated application has generally not led to early failure. Determination of the traction shear properties of thin lubricant films in hertzian contacts and comparison to traction assumed in present-day models. Development of acceptable contaminant descriptions and listing those contaminants which will not be tolerated in field applications based on how they degrade surface topography.
STATUS OF EXTREMES OF OPERATION
Since catalog ratings continue to increase, bearing applications are evolving with more severe operating conditions. Elevated contact stresses, greater misalignment, increased slip in the hertzian contacts, and lower lubricant viscosities now demand more realistic approaches to the determination and maintenance of protective EHL films. An example of this may be Copyright © 1994 CRC Press, LLC
Volume III
485
FIGURE 11. Typical ground bearing contact surface SEM (#125) at 100 x magnification and (# 126) at 500 x magnification. (Courtesy of The Timken Company, Canton, OH.)
found in over-the-highway truck transmissions where oil temperatures may exceed 150°C (300°F), requiring new gear oils to be qualified at 163°C (325°F). This requirement could be the precursor of inner ring growth with concurrent loosening, and of retainer wear due to lower operating oil viscosity. In the low load end of bearing applications, the possibility of an endurance limit for bearings has recently been proposed.41,42 Studies by Lorosch31 and Zwirlein and Schlicht41 at stress levels typical of field applications have shown improved bearing lives exceeding that predicted by AFBMA and ISO load rating standards, and their data indicate a theoretical endurance strength value may exist. A new fatigue life model for bearings proposed by ioannides and Harris42 also includes a “fatigue limit” stress. These concepts indicate that the load-life exponent (n) may increase or be a variable8,41 or that a load exists below which fatigue does not occur.42 The ideal conditions for this consider lambda ratios above 3.0, or even above 10.0, indicating that the contact surfaces are completely separated, mat temperatures are controlled with continuously cooled lubricants, filtered to keep debris particles larger than the Copyright © 1994 CRC Press, LLC
486
CRC Handbook of Lubrication and Tribology
FIGURE 12. Typical superfinished (ground and honed) bearing contact surface SEM (#001) 100 x magnification and (#000) 500 x magnification. (Courtesy of The Timken Company, Canton, OH.)
EHL film from entering the contact areas; and that loads are below the cyclic yield strength so that even localized material stress raisers (e.g., inclusions or carbides) cannot lead to selfpropagating cracks within the material. This fatigue concept is based on experimental work with AISI 52100 through-hardened steel in which the appearance of “dark etching bands” and “white bands”41 is taken as evidence that classical subsurface fatigue sites are being formed. The absence of these indications suggests the attainment of an operating state that will not generate fatigue cracks beneath the operating surface. Similar work has yet to be reported on carburized materials. If rolling contact fatigue actually follows an S-N curve, one would expect that life scatter would increase at lower stresses as with other fatigue.43 Limited bearing data, Figure 3,8 and gear data44 indicate the Weibull slope decreases with lower stress so that at the low loads near an endurance limit, extended scatter would make estimates of the likelihood of a few early fatigue failures difficult. Therefore, care is required in determining the actual reliability for low Copyright © 1994 CRC Press, LLC
Volume III
487
FIGURE 13. Effect of horizontal magnification on depiction of bearing contact surface by profile. Vertical/ horizontal magnification equal to: top, V/H = 0.96/1.0; middle V/H = 24/1.0; bottom V/H = 240/1.0. Stylus tip radius: 2µm.
load, extended life estimates until additional experimental data under these conditions are developed. Although most realistic applications are considerably degraded from the ideal conditions, deliberate design and environmental improvements can help bearings in the field exceed rating life and be limited only by the quality of the material. Improved bearings capable of extended life cannot make up for poor housing design (nonuniform stiffness), excessive misalignment, radial or axial preloading excessive operating temperatures and temperature distribution, vibratory and/or rotating loading, exaggerated mounting conditions, contamination, or abusive ambient conditions. On the other hand, there are minimum loads below which bearings cannot be operated effectively. To achieve extremely extended lives without experiencing wear or failures, other than fatigue, such minimum loads may need to be revised. The new fatigue life theories are more comprehensive than those on which the bearing standards are now based. However, considerably more experimental work will be required before they can gain general acceptance. It is the significant experimental foundation of the past that has provided the technology now available in our current quality level of bearings. a1 a2 a3 C Cr Ca D Dw Dwe F Fcm
NOMENCLATURE
= life adjustment factor for reliability. = life adjustment factor for materials. = life adjustment factor for application or environmental conditions. = basic dynamic rating either radial or thrust load. = basic dynamic radial load rating, Newtons (pounds). = basic dynamic axial load rating, Newtons (pounds). = rolling bearing pitch diameter. = ball diameter, mm (inches). = roller diameter applicable in the calculation of load ratings, mm (inches). = bearing load either radial or radial equivalent, or thrust or thrust equivalent. = factor based on bearing component geometry, accuracy, manufacturing quality and normal bearing steel.
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
488
i Z L-10 L-na Lwe m n S T α ∆
= number of rows of rolling elements in a bearing. = number of rolling elements per row. = basic rating life in million revolutions. = adjusted/application life in million revolutions. = roller length used in the calculation of roller bearing load ratings, mm (inches). = Weibull dispersion or shape parameter or “slope”. = load-life exponent, n = 3 for ball bearings and n = 10/3 for roller bearings. = bearing speed of rotation, rpm. = time (T1 T2... Ti = fractions of time during a loading cycle that totals 1.0). = nominal contact angle of the bearing, degrees. = lambda ratio, EHL film divided by composite roughness.
REFERENCES 1. Derner, W. J. and Pfaffenberger, E. E., Rolling element bearings, in CRC Handbook of Lubrication, Vol. 2, Theory and Practice of Tribology, Booser, E. R., Ed., 1983, 495. 2. Load ratings and fatigue life for ball bearings, American National Standard, ANSI/AFBMA Std. 9–1990, July 17, 1990. 3. Load ratings and fatigue life for roller bearings, American National Standard, ANSI/AFBMA Std. 11–1990, July 17, 1990. 4. Rolling bearings—dynamic load ratings and rating life, International Standard ISO 281, First ed., 1990–12–01. 5. Lieblein, J. and Zelen, M., Statistical investigation of the fatigue life of deep-groove ball bearings, Res. Pap. 2719, J. Res. Nat. Bur. Stand., 57(5), 273, 1956. 6. Lundberg, G. and Palmgren, A., Dynamic capacity of roller bearings, Acta Polytech., Mech. Eng. Ser., 2(4), 96, 1951. 7. Lohmann, G. and Schreiber, H. H., Determination of the rating exponent for ball and roller bearings, Werkstatt Betr., 92(4), 188, 1959. 8. Moyer, C. A., The status and future of roller bearing life prediction, in Proc. Int. Ind. Tribol. Syrup. (Northwestern University). August 1990, STLE SP-31, Advances in Eng. Tribology, Chung, Y., and Cheng, H. S., Eds., April 1991,89. 9. Load ratings and fatigue life for ball bearings, AFBMA Standard 9, Anti-Friction Bearing Manufacturer’s Association, Arlington, VA, June 1972. 10. Load ratings and fatigue life for roller bearings, AFBMA Standard 11, Anti-Friction Bearing Manufacturers’ Association, Arlington, VA, June 1972. 11. Tallian, T., Weibull distribution of rolling contact fatigue life and deviations therefrom, ASLE Trans., 5, 183, 1962. 12. Bamberger, E. N., Harris, T. A., Kacmarsky, W. M., Moyer, C. A., Parker, R. J., Sherlock, J. J., and Zaretsky, E. V., Life Adjustment Factors for Ball and Roller Bearings, An Engineering Design Guide, ASME Spec. Pub., American Society of Mechanical Engineers, New York, 1971. 13. Zaretsky, E. V., STLE Life Factors for Rolling Bearings, STLE Pub. SP-34, Society of Tribologists and Lubrication Engineers, Park Ridge, IL, 1992. 14. Hamrock, B. J. and Dowson, D., Isothermal elastohydrodynamic lubrication of point contacts, II. Ellipticity parameter results, ASME Trans., J. Lubr. Technol., 98(3), 375, 1976. 15. Hamrock, B. J. and Dowson, D., Isothermal elastohydrodynamic lubrication of point contacts, III. Fully flooded results, ASME Trans., J. Lubr. Technol., 99(2), 264, 1977. 16. Hamrock, B. J. and Dowson, D., Isothermal elastohydrodynamic lubrication of point contacts. IV. Starvation results, ASME Trans., J. Lubr. Technol., 99(1), 15, 1977. 17. Dowson, D. and Toyoda, S., A central film thickness formula for elastohydrodynamic line contacts, Proc. 5th Leeds-Lyon Symp., Mechanical Eng. Publications Ltd. (MEP), London, U.K., 1979, p. 60.
Copyright © 1994 CRC Press, LLC
Volume III
489
18. Moyer, C. A. and Bahney, L. L., Modifying the lambda ratio to functional line contacts, STLE Tribol. Trans., 33(4), 535, 1990. 19. Danner, C. A., Fatigue life of tapered roller bearings under minimal lubricant films, ASLE Trans., 13(4), 241, 1970. 20. Andreason, S. and Lund, T., Ball bearing endurance testing considering elastohydrodynamic lubrication, Pap. C36, in Proc. Symp. Int. Mech. Eng. EHD, 1972, 138. 21. Skurka, J. C., Elastohydrodynamic lubrication of roller bearings, ASME Trans. J. Lubr. Technol., 92, Ser. F, 2, April 1970, 281. 22. Moyer, C. A. and Bianchi, A., Relating lubricant parameters and bearing fatigue life in the mixed EHL regime, Pap. C15, in Proc. Symp. Int. Mech. Eng. EHD, 1972, 95. 23. Tallian, T. E., Rolling bearing life modifying factors for film thickness, surface roughness, and friction, ASME Trans., J. Lubr. Technol., 103, 509, 1981. 24. Witte, D. C., Predicting bearing temperature, Mach. Des., 48(12), 110, 1976. 25. Derner, W. J., Misalignment problems in cylindrical roller bearings, SAE 700559, April 1970, Society of Automotive Engineers, New York. 26. Cantley, R. E., The effect of water in lubricating oil on bearing fatigue life, STLE Trans., 20(3), 244, 1977. 27. Wan, G. T. Y. and Spikes, H. A., The behavior of suspended particles in rolling and sliding elastohydrodynamic contacts, STLE Tribol. Trans., 31(1), 12, 1988. 28. Moyer, C. A., The influence of debris on rolling bearing performance: identifying the relevant factors, SAE Pap. No. 871687, Off-Highway/Power Plant Congress Milwaukee, WI, September 14 to 17, 1987. 29. Loewenthal, S. H. and Moyer, D. W., Filtration effects on ball bearing life and condition in a contaminated lubricant, ASME Trans., J. Lubr. Technol., 101, 171, 1979. 30. Sayles, R. S. and MacPherson, P. B., Influence of wear debris on rolling contact fatigue, ASTM STP 771, Rolling Fatigue Testing of Bearing Steels, Hoo, J., Ed., American Society of Testing and Materials, Philadelphia, 1982. 31. Lorosch, H., Research on longer life for rolling element bearings, Lubr. Eng., 41(1), 37, 1985. 32. Dalal, H. and Senholzi, P., Characteristics of wear particles generated during failure progression of rolling bearings, ASLE Trans., 20(3), 233, 1977. 33. Leenders, P. and Houpert, L., Study of the lubricant film in rolling bearings: effects of roughness, in 13th LeedsLyon Conf. Fluid Film Lubr., September 1986, 629. 34. Tripp, J. H., Houpert, L. G., Ioannides, E., and Lubrecht, A. A., Dry and lubricated contact of rough surfaces, in Int. Mech. Eng. Conf. Tribology-Friction, Lubr. and Wear, Fifty Years On, MEP, London, U.K., 1987, 71. 35. Leaver, R. H., Sayles, R. S., and Thomas, T. R., Mixed lubrication and surface topography of rolling contacts, Proc. Inst. Mech. Eng., Vol. 188, 461, 1984. 36. Sayles, R. S. and Thomas, T. R., Measurements of the statistical microgeometry of engineering surfaces, ASME Trans., J. Lubr. Technol., 101, 409, 1979. 37. Tallian, T. E., Rolling bearing life modifying factors for film thickness, surface roughness and friction, ASME Trans., J. Lubr. Technol., 103, 509, 1981 [note discussion by Kauzlarich, p. 516]. 38. Seabra, J. and Berthe, D., Influence of surface waviness and roughness on the normal distribution in the Hertzian contact, ASME Trans., J. Lubr. Technol., 109, 462, 1987. 39. Whitehouse, D. J., Assessment of surface finish profiles produced by multi-process manufacture, in Proc. Inst. Mech. Eng., 199(B4), 263, 1985. 40. Thomas, T. R., The characterization of changes in surface topography during running in, in Proc. Inst. Mech. Eng., 4th Leeds-Lyon Conf., September 1978, 99. 41. Zwirlein, D. and Schlicht, H., Rolling contact fatigue mechanisms-accelerated testing versus field performance, RC Fatigue Testing of Bearing Steels, ASTM 771, Hou, J. J. C., Ed., American Society of Testing and Materials, Philadelphia, 1982, 358. 42. Ioannides, E. and Harris, T. A., A new fatigue life model for rolling bearings, ASME Trans., J. Lubr. Technol., 107, 367, 1985. 43. Rice, R. C., Fatigue data analysis, Statistics and Data Analyses, Metals Handbook, 9th Ed., Vol. 8, Mechanical Testing, American Society for Metals, Metals Park, OH, 1989, 675. 44. Moyer, C. A., The role of reliability for bearings and gears, AGMA Tech. Pap. 92 FTM 8, October 1992.
Copyright © 1994 CRC Press, LLC
POROUS METAL BEARINGS Cris Cusano
INTRODUCTION
Porous metal (P/M) bearings have been commercially manufactured by powder metallurgy since the mid-1920s. Because of their self-contained “oil reservoir” and their simplicity, millions of porous bearings are now used each day in a wide variety of applications. Porous bearings production follows five main steps: (1) The desired metal powders are initially mixed, sometimes with a solid lubricant, to produce a uniform blend. (2) This blend is fed into dies and compacted at room temperature under pressures of 207 to 690 N/mm2 (15 to 150 tons/in.2). (3) The green compact is then sintered at about 80% of the melting point of the base metal, in a reducing atmosphere, to produce metallurgical bonds between powder particles. (4) To improve dimensional accuracy and surface finish, the sintered bearings are repressed with sizing tools. (5) Oil is then introduced into the myriad of interconnecting channels and reservoirs. Porous metal bearings have a number of advantages. They are easily fitted, available in a wide range of stock sizes at relatively low costs, forced-feed lubrication is eliminated, often they offer simplified design involving less space, maintenance is reduced, and oil contamination is reduced. The main limitation of porous metal bearings is that they commonly operate with only boundary or mixed lubrication. Other limitations include: (a) porosity causes reduction of mechanical strength and thermal conductivity; (b) very small bearings are limited by the difficulty of controlling density and by oil content requirements, while very large bearings are limited by the capacity of compacting presses; (c) porous bearing shapes and length-to-wall thickness ratios are limited to those which can be formed by axial compression; (d) alloys are limited to those suitable for powder metallurgy; (e) debris or metal-cutting action such as reaming, grinding, or machining can close pores and significantly reduce selflubricating properties; and (f) specific geometries which are not available off-the-shelf can be prohibitively expensive.
MATERIAL COMPOSITION AND SPECIFICATIONS
Materials standards for porous bearings are given by the American Society for Testing and Materials (ASTM) and by the Metal Powder Industries Federation (MPIF). Even though these standards are voluntary and may be incomplete, they provide useful information from which materials can be selected for specific applications. The ASTM standards are as follows:
B438–83a (reapproved 1989) B438M-84 (reapproved 1989): B439–83 (reapproved 1989): B782–88: B612–90: B328–73 (reapproved 1986):
Sintered Bronze Bearings (Oil-Impregnated) Sintered Bronze Bearings (Oil-Impregnated) [Metric] Iron-Base Sintered Bearings (Oil-Impregnated) Iron Graphite Sintered Bearings (Oil-Impregnated) Iron Bronze Sintered Bearings (Oil-Impregnated) Standard Test Method for Density and Interconnected Porosity of Sintered Powder Metal Structural Parts and Oil-Impregnated Bearings
The MPIF standard 35 entitled “Material Standards for P/M Self-Lubricating Bearings” gives similar information.1 The MPIF standards for bronze bearings are given in Table 1. In this table, the first letter in the material description code gives the basic element, followed by major and minor elements and their percentages, and finally the radial crushing strength. Copyright © 1994 CRC Press, LLC
491
492
CRC Handbook of Lubricants and Tribology
Copyright © 1994 CRC Press, LLC
Volume III 493
Copyright © 1994 CRC Press, LLC
494
CRC Handbook of Lubricants and Tribology
Percentages of other minor elements are excluded from the code and given under “chemical composition”. A material composition code for a bronze bearing would be
The radial crushing constant is related to the radial crushing force by
where Pc = radial crushing force, Ibf; Do = outside diameter of bearing, in.; T = wall thickness of bearing, in.; L = bearing length, in.; and K = radial crushing strength constant, psi. Table 1 also gives minimum value of oil content, Po, and wet density, Dwet. These two quantities are given mathematically as:
where Po = oil content, interconnecting porosity in volume percent; A = weight of oil-free sample in air, g; B = weight of oil-impregnated sample in air, g; C = weight of oil-impregnated sample suspended and immersed in water, g; S = specific gravity of impregnant at the test temperature, g/cm3; E = weight (tare) of suspending wire or basket immersed in water, g; pw = specific gravity of distilled water with wetting agent addition at the test temperature; and Dwet, = wet density, g/cm3. Among the critical characteristics to be considered in choosing a material for a specific application are friction and wear behavior; mechanical strength, including impact; beat dissipating characteristics; conformability; oil content and type; corrosion resistance; and costs. Materials standards for bearings made from bronze, iron and iron-carbon, iron-copper, iron-copper-carbon, diluted bronze and iron-graphite can be obtained from MPIF.1 Similar standards can also be found in ASTM. A brief description of basic characteristics of each of these materials follows.
Bronze Bearings These are the most widely used and give a good balance between strength, wear resistance, conformability and ease of manufacture. By reducing tin content, cost is reduced as well as hardness at some expense to performance under heavy loads. Graphite is added to improve lubrication under oscillatory or intermittent motion, in high temperature applications, and to reduce noise. Relatively low graphite content and lower density imply higher oil content for use in higher speeds-lower loads. Higher graphite content and densities are used at higher loads-lower speeds. Higher density porous bronze bearings with no graphite are good for shock and high loads at low speeds. They should not be used for oscillatory motion or where noise is a major consideration. Bearings with both high graphite and low density should carry only light, nonshock loads, since they have low strength.
Iron and Iron-Carbon Bearings Since these are less expensive and stronger than bronze, iron-base materials have been widely used since the 1950s in noncorrosive environments. Addition of carbon to the iron
Copyright © 1994 CRC Press, LLC
Volume III
495
provides better mechanical strength and better wear resistance. With the higher carbon content, heat treatments improve mechanical properties. Since iron and iron-carbon bearing materials are not as good tribologically as bronze, they are generally restricted to low and medium load applications where no oscillatory motion exists.
Iron-Copper Bearings These are economical, with good high load-low speed characteristics. Bearings with significant copper content are harder, stronger and have better shock-loading capabilities than bronze. Although their friction and wear characteristics are inferior to bronze, these bearings are often used for their higher structural strength. Iron-Copper-Carbon Bearings The iron-copper alloy can be significantly strengthened by adding 0.3 to 0.9% carbon to obtain improved shock, compressive strength, and wear resistance. As with iron-copper bearings, these bearings also find use in high load-low speed applications.
Diluted Bronze Bearings Bronze diluted with 40 to 60% iron can be a good compromise between bronze and iron as related to cost and lubricating quality. Generally, they will have higher coefficient of friction than bronze bearings. They can often replace bronze bearings in less demanding applications under light to medium loads and medium to high speeds. Carbon content is limited to avoid having bearings that are very hard and noisy.1
Iron-Graphite Bearings Being less expensive than bronze, iron-graphite bearings are widely used in noncorrosive environments. Graphite aids lubrication under boundary lubrication. These bearings tend to have good damping characteristics, and they are quiet unless the maximum dissolved carbon is exceeded, in which case very noisy operations can result. Both oil content and load capacity are generally lower than those of bronze bearings.
SOME TYPICAL PROPERTIES
One of the more important properties of porous bearings is porosity. Total porosity is defined as the ratio of pore volume to bulk volume. In a bearing, not all pores are interconnected. Ideally, interconnected porosity is the same as oil content. However, with incomplete oil impregnation during bearing manufacture, oil content can be less than interconnected pore volume. For porous bronze bearings with no graphite content, porosity can be 30% or more, while bronze bearings with 3% graphite may have under 10% porosity. While high porosity is beneficial in storing more oil, tensile and compressive strengths and thermal conductivity decrease with increasing porosity. Thus, choice of porosity is based on a compromise between strength, heat dissipating capacity and oil storage capacity. Mechanical properties of porous bearings vary over a wide range. Based on the MPIF standard for bronze bearings, strength constant, oil content, and density are given in Table 1. Modulus of elasticity is a function of material composition, porosity, and density, and also increases with stresses and strains.2 Typical effects of material composition, density, and porosity on tensile strength are given in Table 2. Some nominal values of coefficient of thermal expansion are given in Table 3. For a given material group, coefficient of thermal expansion also varies with material composition.
LUBRICATING OILS
Porous bearings can be impregnated up to 95% of their available porosity by using heat, vacuum, or pressure, or a combination of all three. The vacuum method provides the greatest Copyright © 1994 CRC Press, LLC
496
CRC Handbook of Lubricants and Tribology
impregnation in the shortest time. Turbine mineral oils are the most commonly used lubricants for impregnating porous bearing. Lower viscosity oils are used in low load and higher speed applications at moderate operating temperatures, while higher viscosities are used when oil migration or noise4 are a problem, or at high temperature and in heavy duty applications involving high loads, low speeds, and oscillating motion where extreme boundary lubrication is expected. Figure 1 gives a guide for selecting oil viscosity as a function of shaft peripheral velocity.3 Oils to impregnate porous bearings have an ISO viscosity grade which ranges from 32 to 150, the most common being about 68. A safe maximum operating temperature range for mineral oils is approximately 93 to 121°C (200 to 250°F). In practice, the maximum temperature at the shaft-bearing interface is not easily determined and an “average” bearing temperature is often used to estimate safe Copyright © 1994 CRC Press, LLC
Volume III
497
FIGURE 1. A general guide to the choice of oil viscosity as a function of shaft velocity.
conditions. Limited data suggest that the O.D. bearing temperature, on the loaded side, should not exceed about 66°C (150°F) for satisfactory operation.5 According to Whiting,6 if operating temperature exceeds approximately 88°C (190°F), mineral oil degradation can lead to early bearing failure. If bearing temperature is high enough to cause mineral oil degradation, or for operation in low or high temperature environments, more expensive synthetic lubricants should be considered for their lower pour point, improved oxidation or thermal stability, or better viscosity-temperature characteristics. Base synthetic fluids commonly used are polyalphaolefin synthetic hydrocarbons (PAOs), diesters, polyolesters and to a small extent silicones. Some advantages and limitations of these lubricants are given in Table 4. It is usually wise to select synthetics which are miscible with common mineral oils unless replenishment by the user with the wrong oil can be safeguarded. Almost all mineral and synthetic oils used to impregnate bearings contain additives to improve specific properties. These additives include polymers to reduce migration, oxidation Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
498
inhibitors, rust protectors, extreme pressure additives, and antiwear additives. In some applications, submicron particles of graphite are added to the oils to improve boundary lubrication. In general, suspensions of solid lubricants can cause pore closure and should be avoided unless experience in special applications indicates otherwise. Grease might be preferred in some applications, as with water repellent grease acting both as a lubricant and sealant. If grease is used, it should be fed directly to the I.D. of a porous bearing, since its soap content will plug the pores if fed through the O.D. Grooving is often used to distribute grease along the entire length of the bearing.
LUBRICATION AND FRICTION
Role of Porosity and Permeability Both porosity and permeability play a critical role in oil retention and circulation in a porous bearing. Permeability defines the ease with which a fluid may be made to flow through the material by an applied pressure gradient, in other words, its fluid conductivity. For porous sleeve bearings, permeability varies along their length to a much greater degree than porosity. The permeability at the center of a porous bearing can be five times that at its ends.2 The higher the friction at the die-compact and punch-compact interfaces, the lower the compressibility of the powder and the higher the permeability variations. Because of the relatively thin wall, permeability variation is much less in the radial direction. If the fluid flow is symmetric about the axis of a sleeve porous bearing, average permeability is approximately given by
where k = permeability, m2; q = radial volumetric flow rate, m3/s; µ = absolute viscosity of fluid, Pa*S; rir0 = inside and outside radius of bearing, respectively, m; L = length of bearing, m; and pi, po = fluid pressure on inside and outside wall of bearing, respectively, Pa. A typical value of k, for a porous bronze bearing with little or no graphite, is 8 x 10-14 m2. Relationship between porosity and permeability depends on tortuosity of the channels and on pore geometry. There is an approximate linear logarithmic relationship between permeability and porosity.8 Permeability is also a function of time, since particles in the oil and air, or those which have come loose due to wear, can clog the pores. In addition, smearing of the bearing surface takes place, if not continuously, then during starts and stops under load. Olexa9 claims that rapid decrease of permeability can also be caused by a boundary layer formed by lubricant-metal interactions at the shaft-bearing interface. Rate of decrease depends on materials contact pairs, pore size and shape, and especially oil type. Oils with EP additives seem to cause the greatest decrease in permeability.
Oil Circulation and Loss Oil flow to the bearing surface can be explained as follows. If a nonrotating shaft is loaded, some oil will seep out of a porous bearing simply due to compressibility of its structure, as with placing a load on a saturated sponge. As a loaded shaft starts to rotate, heat generated at the bearing-shaft interface decreases both oil viscosity and capillary forces and also expands volume of oil contained in the bearing. These factors tend to bring oil out of the bearing matrix and into the clearance space between the bearing and shaft. If conditions favor hydrodynamic pressure generation in the loaded region, oil then circulates from the loaded to the unloaded regions of the bearing, as shown in Figure 2. Unless oil is adequately replenished, the film extent will decrease with continued operation until asperity contact and finally boundary lubrication conditions exist. Copyright © 1994 CRC Press, LLC
Volume III
FIGURE 2. Schematic representation of mechanisms of oil flow in the sintered porous bearing.
499
Oil circulation in porous bearings has been studied by Braun10 and Kaneko and Obara.11 Braun examined oil circulation by placing a mirror inside a hollow glass shaft. He observed that when the shaft rotates, oil disappears from the bearing gap into the pores on the loaded part of the bearing. The bearing gap is refilled with oil from the unloaded part of the bearing, partly as the result of the capillary action of the bearing gap itself. Hydrodynamic lubrication occurred only when very small radial loads were applied and only in newly impregnated bearings. For the operation conditions involved, evaporation and creep were the dominant oil loss mechanisms. Kaneko and Obara injected a dye to estimate oil circulation in a porous matrix. Under hydrodynamic conditions, oil flowed away from the load line toward the unloaded region; but under boundary lubrication conditions, most of the oil in the porous matrix flowed toward the region encompassed by an angle 180° ≤ θ ≤ 225° from the load vector in the direction of shaft rotation. Inflow of oil from the clearance into the porous matrix occurs in the loaded region. When a porous bearing is received from a manufacturer, it is about 90% saturated with oil. With the progressive loss of oil during use, the pores are emptied, size by size, until the increased surface tension forces produced by the smaller pores are sufficient to reabsorb some of the oil lost. Even though capillary forces increase slowly with oil loss when the oil saturation is relatively high, they increase very rapidly when the saturation is less than 50%. Figure 3, taken from Morgan,12 shows typical oil loss curves. Morgan suggests that before the oil content of the bearing falls below 65%, it should be replenished by means of felt or wool wicks, oil reservoirs, or oil cups. Typical oil fittings and reservoirs are shown in Figure 4.12 Any hydrodynamic pressure developed in a porous bearing causes oil loss from its ends. Other factors which tend to cause oil loss include centrifugal action flinging oil from the shaft surface at the ends of the bearing, lower viscosity and lower surface tension of the oil, creepage of oil along the shaft, relatively rough surfaces, higher temperature and associated oil evaporation, and large pore size. Youssef and Eudier13 suggest that machining small recesses on both ends of the bearing will reduce flinging oil losses. Oil losses can also be reduced by (a) adding felt washers or slingers on the shaft at the ends of the bearing to reduce oil migration, (b) using a shaft finish which will reduce migration (see subsection “Shafts for Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
500
FIGURE 3. Typical oil loss curves for porous metal bearings.
FIGURE 4. Typical methods of supplementing and replenishing the oil in the pores of a bearing.
Porous Bearings”), (c) choosing oils which promote oil retention (higher viscosity and surface tension, for example), (d) reducing effective pore size, and (e) machining very thin slits along the axial direction near the ends of the bearing.
Friction Friction in porous bearings is predominantly due to surface interaction rather than shearing of the oil film. Experiments by Cusano and Phelan,5 for bronze porous bearings with an I.D. of 25.4 mm, indicated that coefficient of friction changed much more with load than with speed. After all excess impregnation oil was removed from the bearing, the coefficient of friction ranged from 0.05 to 0.08 for loads per unit projected area of 0.52 to 1.55 MPa (75 to 225 psi) at 850 rpm. For a load of 1.04 MPa (150 psi) and speeds from 425 to 1275 rpm, the coefficient of friction was within a narrow range of 0.05 to 0.055. The coefficient of friction obtained by Morgan14 under boundary lubrication varied from 0.06 to 0.09. For lightly loaded miniature porous bearings, Braun10 reported that friction is a strong function of operating time and speed during the running-in process. With the light loads and speed range involved, the
Copyright © 1994 CRC Press, LLC
Volume III
501
coefficient of friction varied from very small values (hydrodynamic) up to 0.2. Under boundary lubrication, the coefficient of friction depends on loads, speeds, ability of lubricant to form tough surface films, material properties, and temperature. Typical values range from 0.05 to 0.15. When bearings are operating near the PV limit suggested by manufacturers, a coefficient of friction of about 0.1 is a reasonable estimate. For given operating conditions, coefficient of friction can be reduced by using a hardened and ground or chrome-plated shaft and by making sure that oil exists at the bearing surface during initial running-in.
THERMAL CONSIDERATIONS
Energy Balance An energy balance to estimate bearing temperature is important since (a) dimensions of the bearing and shaft change with changes in temperature, (b) permanent distortion can occur due to creep of the metals maintained at high temperatures for prolonged periods, (c) bearing load capacity can significantly drop due to accelerated oil loss and lower viscosity at increased operating temperatures, and (d) oil oxidation and thermal breakdown occur at high temperatures. Lubricant breakdown will result, in turn, in excessive noise, vibration, accelerated wear, high operating temperature, and the possibility of smoke. For porous bearings, the operating temperature must be kept as low as possible by minimizing the heat generated at the contact and designing the assembly for optimum heat dissipating characteristics. Heat generated can be calculated as follows: where Hg = heat generated, watts; W = load, N; V = peripheral velocity of shaft, m/s; and f = coefficient of friction. Ability of an assembly to dissipate this heat by convection, conduction, and radiation depends mainly on materials used, air velocity around the assembly, exposed hot area of the assembly, and ambient temperature. The heat dissipated, in its simplest form, can be expressed as where Hd = heat dissipated, watts; K = heat dissipating characteristics of the assembly, watts/°C; T = approximate bearing temperature, °C; and Ta = ambient temperature, °C. For a given assembly operating in a given environment, the value of K can be obtained experimentally as suggested by Morgan.3 In this approach, a thermocouple measures the bearing temperature near the shaft-bearing interface while an electric heating element simulates heat generated in the bearing. Watts input can be plotted against T—Ta; the slope of this plot gives the value of K. Having K, Equations 4 and 5 can be equated to get an approximate steady-state bearing temperature, i.e., If K cannot be obtained experimentally, an equation developed by Glaeser and Dufrane,15 based on the work of Crease,16 can be used for metal bearings and steel shafts.
Copyright © 1994 CRC Press, LLC
502
CRC Handbook of Lubricants and Tribology
where T,Ta = bearing and ambient temperatures, respectively, °F; f = average coefficient of friction; W = load, lb.; D, Dh = diameter of shaft, and equivalent housing diameter, respectively, in.; N = shaft speed, rpm; Rt = thermal resistance factor (values of 0.7 and 0.9 suggested for 1 in. and 10 in. shaft diameters, respectively); h = convective heat transfer coefficient BTU/h/ft2/°F (values of 6 and 4 suggested for 1 in. and 10 in. shaft diameters, respectively); L,Lt = length of bearing, and total length of shaft from center of bearing to free end, respectively, in.; kbh, = thermal conductivity of bearing-housing combination BTU/ h/ft/°F (for cast iron or steel housings, a value of 30 is recommended); and C = shaft cooling factor, dimensionless (recommended values are 0.2 and 0.1 for 1 in. and 10 in. shaft diameters, respectively). Equations 6 and 7 will only give some average bearing temperature. In using the heater to estimate K in Equation 6, for example, the heat is transferred uniformly to the wall of the bearing. In an operating bearing, the heat is usually generated over an arc of less than 180° and has an approximate parabolic distribution. The average temperature calculated, therefore, could be significantly lower than the maximum bearing temperature experienced in service. In the above relationship, both coefficient of friction and the heat-dissipating characteristics for the assembly require some guessing unless experience dictates otherwise. Maximum tolerable bearing temperature in a given application mainly depends on the oil and its replenishment, if any. Oxidation and thermal stability, evaporation, creep, and viscosity of mineral oils limit prolonged exposure to temperatures below 121°C (250°F) for reasonable life. To increase the heat-dissipating characteristics of a bearing assembly, the following methods can be considered:16 (a) position the housing with unrestricted access to surrounding air, (b) provide heat flow paths between the bearing and the exterior housing surface, (c) use housing materials with high thermal conductivity, (d) use fins on the housing surface, (e) insulate joints between the housing and other heat sources, and (f) use forced air cooling, as with a fan on the shaft. PV Factor Load capacity of porous bearings, as with many other self-lubricating bearings, has been related for many years to a maximum PV factor (P = load W/(bearing length L x bore diameter D) with a unit of psi, V = peripheral shaft velocity in ft/min.). While the PV factor is not an ideal design criterion, it is usually related to two important limiting criteria for acceptable porous bearing performance: operating temperature and wear rate. Relation of PV factor to heat input into the bearing, Hg, can be shown by arranging Equation 4 in the form:
Applying a heat balance, from Equations 5 and 8,
Thus, PV is an indicator of the bearing temperature rise relative to the ambient. For a given material, if it is assumed that volume wear rate (and radial wear rate) is proportional to the rate of energy generated, then radial wear is also proportional to PV. Suggested PV limits plus limiting values of P and V for bronze sleeve bearings are shown in Table 5. While difficult to generalize for specific applications, bearing operation should normally be well within these load, speed, and PV limits. If hydrodynamic lubrication exists, Copyright © 1994 CRC Press, LLC
Volume III
503
PV value has little meaning. For a PV value of 10,000 or less, difficulties are unlikely. For operation at or near maximum PV values, periodic lubricant replenishment should be considered. Generally, if additional lubrication is provided, if the bearing is cooled by forced air or other means, and if the shaft is hardened or chrome-plated, the PV value can be near or above maximum quoted values. On the other hand, for continuous high speeds or high loads; shock loading; relatively large bearing clearances; assemblies with limited heat-dissipating capabilities; a misaligned or out-of-round shaft; long life, rotating bearings; bearings with high length-to-diameter ratios; oscillating or reciprocating motion; abnormally high or low operating temperature; or dirty and humid operating environments; the PV value should be lower than the recommended values. Pratt,17 Cusano and Phelan,5 and Braun10 found a PV of 50,000, the value generally recommended by manufacturers, was too high for their porous bronze test bearings. On the other hand, some of the tests conducted by Cheng et al.,18 ran satisfactorily for 5 h at a PV value of 50,000. Cheng et al. also found that the lowest operating temperatures were associated with relatively large pores which resist closure during sizing and testing and are therefore more likely to sustain the oil delivery than smaller pores. It is generally accepted mat, for a given PV value, the heat generated at high P in combination with a low V will be less than that generated for a low P and high V. For selflubricating bearings, Cooper19 suggests modifying factors to compensate for specific attributes of the application in question. Because of the many factors which affect the tribological behavior of porous bearings, life-predictive models are nonexistent in the open literature. When data are lacking for a particular application, the bearing should be tested for the expected operating conditions, not under accelerated conditions (especially velocity), where failure in a relatively short time might not have much relationship to actual operation.
Seizure Seizure is a very complicated function of the mechanical, thermal, and wear properties; radial clearance; heat generated at the contact; and heat dissipating properties of the assembly. Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
504
Two possible mechanisms can cause seizure. The first is related to thermal expansion and contraction, which can eventually take up the initial clearance between the shaft and bearing.20 To minimize this possibility, it is necessary to check tolerances on the bearing I.D. and shaft plus differential expansion over expected extreme temperature ranges to make sure that an acceptable clearance exists at operating conditions. The second type, called galling seizure by Rabinowicz,21 is caused by wear particles generated during sliding. As more and more wear particles are formed, clearance is lost and both surfaces are roughened. Rabinowicz suggests that galling seizure can be minimized by using materials with a low ratio of surface energy of adhesion to hardness of the softer contacting surface and by using a clearance which is at least three times the average diameter of the wear particles.
BEARING ASSEMBLY AND GEOMETRY
Machining In many applications, proper dimensional accuracy in porous bearings is obtained by using a fitting mandrel (sizing plug) while the bearing is press fitted into the housing. Due to springback, the diameter of the mandrel should be larger than the desired fitted bearing bore size by 2.5 to 5 µm for high porosity bronze and 5 to 7.5 µm for low porosity bronze and cast iron.12 The exact springback, however, might have to be determined by trial and error. The mandrel should be hardened steel and have a surface finish of better than 0.4 µm CLA. With this approach, post assembly machining is generally not required and bearing clearance is dictated only by the nominal shaft size and its tolerance. If bore diameter needs to be increased after a bearing is press-fitted into a housing, a reamer or a burnishing tool can be used. Both tools should be sharp in order to minimize pore closure. Reamers are usually used for small volume operations. Tolerances of 12.5 µm or less can be maintained on a production basis using burnishing tools. Burnishing is preferred to reaming to minimize pore closure. Boring can also be used when significant material removal is required or for extremely close tolerances or alignment of holes. Boring should be at high speeds with fine feeds using carbide or diamond-tipped tools since they retain a sharp cutting edge which minimizes pore closure.
Interference and I.D. Close-in Unless porous bearings are of the self-aligning type, they are press-fitted into the housing. A typical press-fit range for bronze bearings is given in Figure 5.22 For other materials, the manufacturer should be consulted. These data are for relatively rigid housings and for applications which operate near ambient temperatures. The “minimum” line can be approached for bearings supporting moderate, unidirectional loads, for longer bearings since they have more surface area for holding power, and for thin-walled bearings to reduce distortion. Both housing bore and bearing O.D. should be chamfered to facilitate alignment and insertion. Attention should be given to the possibility of a press-fitted bearing becoming loose in its housing, as the result of an appreciable temperature change. The interference fit in the housing produces a reduction in bore diameter which is a function of outside diameter of the bearing, wall thickness, and bearing materials. For bronze bearings, typical values for this “close-in” are given in Figure 6.22 For other bearing materials, the manufacturer should be consulted. If no information is available, the equations for compound long cylinders can be used.23 Bearing Diametrical Clearance Typical clearance values for bronze bearings at room temperature are given in Figure 7.22 For other materials, the manufacturer should be consulted. Low running clearances are more likely to cause problems with seizure, misalignment, and high operating temperature. On the
Copyright © 1994 CRC Press, LLC
Volume III
505
FIGURE 5. Typical press-fit values for porous bronze bearings. (Oilube Powdered Metal, Bearing Sales Corporation, Chicago, IL.)
FIGURE 6. Typical I.D. close-in for porous bronze bearings. (Oilube Powdered Metal, Bearing Sales Corporation, Chicago, IL.)
FIGURE 7. Typical bearing clearance for porous bronze bearings. (Oilube Powdered Metal, Bearing Sales Corporation, Chicago, IL.)
Copyright © 1994 CRC Press, LLC
506
CRC Handbook of Lubricants and Tribology
other hand, peak pressure, oil loss, wear, and noise can all be reduced and more precise radial location can be expected with smaller clearances. For self-aligning bearings, lower clearances are usually recommended. While it is not shown in Figure 7, Morgan3 recommends that clearance should increase with shaft peripheral velocity. Although clearance is normally specified at room temperature, bearing performance is based on clearance at the actual operating temperature. Since a shaft, bearing, and housing seldom have the same coefficient of thermal expansion, dimensional changes can occur at operating temperatures which appreciably alter the room temperature clearance. For example, if the coefficient of thermal expansion of housing (αh bearing (αb), and shaft (αs) are such that αh < αb < αs, a temperature increase could result in complete loss of clearance. Also, if αh > αb, the interference between the housing and bearing might be lost at elevated temperatures. A systematic evaluation of combinations of coefficient of thermal expansion for shaft, bearing, and housing, and their effect on running clearance and on O.D. bearing interference are given by Morgan.14 With self-aligning porous bearings, the thermal expansion or contraction of the housing can be neglected. Assuming a coefficient of friction, radial clearance changes can be predicted by a full thermal analysis of the assembly using a commercially available finite element package. If such an analysis cannot be done, change in radial clearance can be estimated by using equations developed by Nica.24 These equations assume uniform temperatures at the I.D. and O.D. of the bearing and at an equivalent O.D. of the housing, and that radial stress at the I.D. of the bearing is also uniform and equal to the load per unit projected area on the bearing. Shafts for Porous Bearings Surface finish, geometry, and hardness of the shaft are critical to the life of porous bearings. To reduce oil pumping action of the shaft and thus increase bearing life, its radial outof-roundness should not be more than about 2.5 µm, with care taken that no spiral machine tool marks are on its surface. Shafts should have a surface finish of approximately 0.4 µm RMS or better. As a rule of thumb, the shaft should be at least 30 points (RB scale) harder than particles making up the bearing. For bronze bearings, for example, the particle hardness is approximately 65 RB. Therefore, shaft hardness should be 95 RB or better. Hardened shafts are especially important when using porous bearings with a significant amount of iron. Hardened and ground or chrome-plated shafts invariably give better performance than soft shafts. Recommended materials for shafts are carbon steels containing approximately 0.4% carbon. Steels with lower carbon content, free machining types, should be avoided. As a general rule, 300 series stainless steel should not be used against bronze bearings; 400 series stainless steels have better characteristics but not as good as straight carbon steels. Where corrosion resistance is required, chrome plating is recommended. Length-to-Diameter Ratio For a given diameter, load capacity of a porous bearing increases with length. Operating mainly in the boundary lubrication regime, their length-to-diameter ratios tend to be larger than those of equivalent solid bearings used under hydrodynamic lubrication. Elastic deformation of the shaft, however, can neutralize the advantage of longer bearings unless the bearing is selfaligning. Bearing length is also limited by the P/M process itself. The longer the bearing, the more likely large porosity and permeability gradients will exist in the axial direction. For very long bearings, buckling or breakage of the core can occur during the sizing operation. As a general rule, if the assembly is relatively rigid, misalignment is not a problem and the radial clearance is not expected to be low, longer bearings can be used. By considering the heat Copyright © 1994 CRC Press, LLC
Volume III
507
FIGURE 8. Shaft deflection in a bearing.
balance, for the same PV and assembly, shorter bearings will operate at a lower temperature.
Wall Thickness Effect of wall thickness on the life of porous bearings is unclear.25 Bearings with a thicker wall can store more lubricant, but a thicker wall often increases unit cost and reduces both thermal conductivity and radial stiffness. On the other hand, bearings with a very thin wall can be expensive due to high cost of the manufacturing tools needed. In addition, thin walls can cause concentricity problems during manufacturing. As a general rule, bearings less than 6.3 mm I.D. should have a minimum wall thickness of 1.2 mm. For larger bearings, wall thickness should be about 0.125 x shaft diameter.
MISCELLANEOUS CONSIDERATIONS
Misalignments Some misalignment always exists in a bearing assembly from distortion, accumulation of tolerances, manufacturing errors, and deflections due to loaded members. The misalignment causes a nonuniform pressure distribution and possible excessive pressures near one end of the bearing, resulting in shorter bearing life.26 If misalignment is a major problem, either selfaligning bearings or a flexible bearing mount should be considered. Misalignment problems can also be reduced by reducing bearing length, increasing radial clearance, and increasing shaft rigidity. According to Morgan,12 shaft deflection over one half of the bearing length should be less than one quarter of the diametrical running clearance. If a centrifugal force exists due to an imbalance, its contribution to shaft deflection must also be considered. Referring to Figure 8, for bearing A, this deflection can be calculated from (using any consistent system of units)
where Es is modulus of elasticity of the shaft, W is the load, and “a” represents the shorter distance from the load vector to the bearing since the larger deflection occurs in the bearing closer to the load vector.
Noise High graphite content bronze bearings or iron-graphite bearings should be considered in applications where bearing noise is important. For both iron-graphite and diluted bronze Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
508
bearings, exceeding the maximum recommended dissolved carbon can also result in noisy bearings.1 In general, noise can be reduced by reducing radial clearance and using oils with good boundary lubrication characteristics. Watanabe et al.27 indicated that, for porous bronze bearings, noise level decreased as permeability decreased to approximately 2 x 10-15 m2. As permeability decreased further, noise level began to increase. They related this phenomenon to the ease with which oil could flow from the bearing structure. Another parameter which affects noise is oil viscosity, as discussed earlier.
Thrust Bronze Bearings Thrust bronze bearings or washers are designed for PV values of 10,000 or less. Mean velocity, based on the average radius, is used to determine V. To improve the friction and wear behavior of these washers, their mating surface should be made of steel which has been ground to a surface finish of 0.4 µm RMS or better and its hardness should not be less than 95 RB.
Improving Load Capacity With reduction in permeability, reduced seepage keeps oil in the clearance gap where it more effectively generates a pressure to support a load. Permeability can be reduced only so far, however, since its reduction usually results in reduction of porosity and oil content within the bearing. Permeability has been reduced on the loaded side of the bearing by deliberate overload to smear the bearing surface.28 While this local reduction should theoretically improve both friction and load characteristics, surface damage to the shaft and temperature rise caused by this overloading and its effect on oil loss and degradation might do more harm than good. Another approach for reducing the permeability on the bore of the bearing has been reported by Youssef and Eudier13 and Eudier and Margerand.29 They experimentally found that a layer of ultra fine powder of lead and tellurium alloy, approximately 100 µm thick, on the bore of a conventional porous bearing, significantly reduced coefficient of friction and increased allowable PV. The ultra fine powder, in addition to reducing oil seepage, was believed to improve feeding of oil to the bearing bore due to stronger capillary forces. Bocchini et al.30 and Quan and Wang31 have claimed that a stepwise variation of permeability in the circumferential direction significantly improves load capacity and decreases the coefficient of friction.
THEORETICAL BEHAVIOR BASED ON FILM FORMATION
Background While porous bearings are likely to operate in the boundary regime, numerous analytical investigations since the late 1950s have been directed at their hydrodynamic operation which may occur when oil is periodically supplied to the bearings, or when bearings are saturated with oil during initial stages of their lives and operate at relatively low loads and high speeds. Two important differences exist in the analysis of porous and solid bearings. First, unlike solid bearings which theoretically can carry an infinite load as the eccentricity ratio approaches one, porous bearings have finite load capacity. For porous bearings, as seen in Figure 9, the circumferential flow is reduced or even completely eliminated by radial oil seepage q1 through the wall and the total side flow qs out of the ends of the bearing. For a given eccentricity ratio, the more this seepage, i.e., the higher the permeability, the less load a bearing can carry. Second, because of the porous structure, the no-slip boundary condition used at the surface of solid bearings is not applicable. Representative analytical solutions for journal bearings have been obtained by Prakash and Vij32 and Reason and Siew.33 Copyright © 1994 CRC Press, LLC
Volume III
509
FIGURE 9. Hydrodynamic (a), and boundary (b) lubrication in a porous bearing.
FIGURE 10. Sommerfeld number vs. permeability factor for ε = 0.7.
Porous Bearings with a Copious Oil Supply Figures 10 and 11 give typical results for the load capacity of journal porous bearings with a copious supply of oil.33 In these figures, ε is eccentricity ratio; T the bearing wall thickness; D the I.D.; L the bearing length; C is the radial clearance; ri is D/2; Ψ is a permeability factor (dimensionless) defined as kri/C3 with k representing the permeability; and S the Sommerfeld number (dimensionless) defined as µ-N/P (ri/C)2 with N the shaft rotational speed (rev/s); µ an average absolute viscosity of the lubricant; and P the load per unit projected area. As expected, as the permeability increases, bearing load capacity is reduced (Sommerfeld number increases). Although not shown, the trend for coefficient of friction is similar to that for Sommerfeld number for the two eccentricity ratios given.
Starved Porous Bearings With subsequent lubrication from sources such as wicks, felt pads, drop-feed oil cups, and reservoirs to replenish all the oil loss, a porous bearing can operate under continuous but steady starved conditions. Of course, even if oil is completely replenished by means of an external oil source, the film extent of the starved film might still be too small to support the load. Thus, mixed or boundary lubrication conditions are possible. Representative performance results for starved, steady-state, hydrodynamic conditions are shown in Figures 12 and 13.34 In these figures, additional symbols not defined for Figures Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
510
FIGURE 11. Sommerfeld number vs. permeability factor for ε = 0.9.
FIGURE 12. Total side flow vs. active film arc for ε = 0.8.
Copyright © 1994 CRC Press, LLC
Volume III
511
FIGURE 13. Summerfeld number vs. active film arc for ε = 0.8.
10 and 11 include Φ, defined as k/C2, the total dimensionless side flow, Qs, defined as qsCLriN), the active film arc, β, and the OD of the bearing, Do. Figures 12 and 13 can be used by knowing either the oil flow into the bearing from an external source or the active film arc under given operating conditions. Under steady-state conditions, the amount of oil externally fed to the bearing is assumed to equal the total side flow, qs. Knowing this flow and permeability parameter, Φ, the active film arc, p, can be estimated. For a given β and eccentricity ratio, ε, the Sommerfeld number can be obtained from Figure 13 and the load supported at this eccentricity ratio can be calculated.
Stability and Transmissibility Behavior Chattopadhyay and Majumdar35 and Conry and Cusano36 show that, for the same steadystate eccentricity ratio, porous journal bearings supporting rigid rotors are less stable than solid bearings. Porous bearings, however, do show promising transmissibility characteristics. Simply as a sleeve bearing37 or especially as a squeeze film damper for rolling element bearings,38 force transmissibility characteristics of porous bearings are predicted to be better than those of solid bearings. Elastohydrodynamic Lubrication For a given load, a porous bearing of the same length and inside diameter deforms more than a solid bearing. Nevertheless, only limited studies have been done of elasticity effects with
Copyright © 1994 CRC Press, LLC
512
CRC Handbook of Lubricants and Tribology
porous bearings since they are generally not used for critical, heavy load capacity applications requiring an external oil supply. The effects of matrix deformation on bearing characteristics has been considered by Mak and Conway39 as well as Jain et al.40 The data generally show that, for a given permeability, the more the deformation in the porous matrix, the lower the load capacity, and this difference increases with eccentricity ratio. Based on rigid rotor analysis and for a given permeability and eccentricity ratio, porous bearings which deform more are also more stable. This effect also increases with increasing eccentricity ratio.
ACKNOWLEDGMENT
I would like to thank Mr. Michael Schloder of Keystone Carbon Company, St. Mary, PA and Mr. Irvin Zaleski of Stackpole Ltd., USA, Brownsville, TN for supplying some of the information included in this chapter and for their comments and time taken to read the manuscript.
REFERENCES 1. Metal Powder Industries Federation, MPIF Standard 35, Materials Standards for PIM Self-Lubricating Bearings, Princeton, NJ, 1991–1992. 2. Morgan, V. T., The effect of porosity on some of the physical properties of powder metallurgy components, Powder Metall., 12(24), 426, 1963. 3. Morgan, V. T., Self-lubricating bearings—2 porous-metal bearings, Engineering, 220(8), Tech. File No. 80, 1980. 4. Watanabe, T. and Shimizu, T., Effects of lubricating oil on the sliding noise of sintered porous bearings of bronze, Rep. Casting Res. Lab., No. 32, Waseda University, Tokyo, Japan, 1981. 5. Cusano, C. and Phelan, R. M., Experimental investigation of porous bronze bearings, J. Lubr. Technol., Trans. ASME, Ser. F, 95(2), 173, 1973. 6. Whiting, R., Selecting synthetic impregnating oils, paper presented at Annu. Meet., Small Motor Manufacturing Association, Saddelbrook, FL, 1985. 7. Whiting, R., Ultrachem, Inc., Wilmington, DE, personal communications, 1991. 8. Grootenhuis, P. and Leadbeater, C. J., Discussion, in Symp. Powder Metall., Iron Steel Inst. Spec. Rep., 58, 361–363 and 367–369, 1954. 9. Olexa, J., Investigation of the relations between the permeability and the service life of porous self-lubricating bearings. Wear, 58, 1, 1980. 10. Braun, A. L., Porous bearings, Tribol. Int., 5(5), 235, 1982. 11. Kaneko, S. and Obara, S., Experimental investigation of mechanism of lubrication in porous journal bearings. I. Observation of oil flow in porous matrix, J. Tribol., Trans. ASME, 112, 618, 1990. 12. Morgan, V. T., Porous metal bearings and their application, in Conf. on Bearings: searching for a longer line, held at Cheltenham, England, sponsored by Chartered Mechanical Engineer, London, 1984. 13. Yonssef, H. and Eudier, M., Production and properties of a new porous bearing, Powder Metall. Conf., Pap. 53, American Institute of Mining Engineers, NY, 1965. 14. Morgan, V. T., Porous metal bearings, Powder Metall., 12(24), 426, 1969. 15. Glaeser, W. A. and Dufrane, K. F., The design of boundary lubricated cast bronze bearings, in Handbook, Cast Bronze Bearing Institute, Inc., Cleveland, OH, 1978. 16. Crease, A. B., in Heat dissipation from bearing assemblies, in Tribology Handbook, M. J. Neale, Ed., John Wiley, 1973, F7. 17. Pratt, G. C, A review of sintered metal bearings: their production, properties, and performance, Powder Metall., 12(24), 356, 1969.
Copyright © 1994 CRC Press, LLC
Volume III
513
18. Cheng, J. A., Lawley, A., Smith, W. E., and Robertson, J. M., Structure, property and performance relations in self-lubricating bronze bearings: commercial premixes, Int. J. Powder Metall., 22(3), 149, 1986. 19. Cooper, J. H., Why, when where, and how to use self-lubricating bearings, in Proc. Nat. Conf. Power Transmission, November, 1981, Chicago, 1981, 167. 20. Khonsari, M. and Kim, H. J., On thermally induced seizure in journal bearings, J. Tribol., Trans. ASME, 111,662, 1989. 21. Rabinowicz, E., Friction seizure and galling seizure, Wear, 25, 357, 1973. 22. Oilube Powdered Metal, Bearing Sales Corporation, Chicago, IL 23. Burr, A. H., Mechanical Analysis and Design, Elsevier, New York, 1981, 318 and 325. 24. Nica, A., The real clearance in sliding bearings, J. Basic Eng., Trans. ASME, 87(3), 781, 1965. 25. Cusano, C., Theoretical and Experimental Analyses of Porous Metal Bearings, Ph.D. thesis, Cornell University, Ithaca, NY, 1970. 26. Cooper, J. H., The influence of misalignment on self-lubricated bearings, SAE Trans., 92, No. 831370, 1983. 27. Watanabe, T., Shimizu, T., and Endo, H., Effects of permeability of sintered porous bearings of bronze on the sliding noise, Rep. No. 30, Casting Res. Lab., Waseda University, Tokyo, Japan, 1979. 28. Morgan, V. T., Hydrodynamic porous metal bearings, Lubr. Eng., 20(12), 449, 1964. 29. Eudier, M. and Margerand, R., Practical working conditions for sintered bearings, Powder Metall., 12(24), 417, 1969. 30. Bocchini, G. F., Capone, A., Capone, E., and Niola, V., The four step porous bearing, Tribol. Int., 11, 330, 1978. 31. Quan, Y. X. and Wang, P. M., Theoretical analysis and experimental investigation of the porous metal bearing, Tribol. Int., 18, 67, 1985. 32. Prakash, J. and Vij, S. K., Analysis of narrow porous journal bearing using Beavers-Joseph criterion of. velocity slip, J. Appl. Mech., Trans. ASME, 41(2), 348, 1974. 33. Reason, B. R. and Siew, A. H., A refined numerical solution for the hydrodynamic lubrication of finite porous journal bearings, Proc. Inst. Mech. Eng., 199, C2, 85, 1985. 34. Cusano, C, An analytical study of starved porous bearings, J. Lubr. Technol. Trans. ASME, Ser. F, 101(1), 38, 1979. 35. Chattopadhyay, A. K. and Majumdar, B. C, On the stability of a rigid rotor in finite porous journal bearings with slip, J. Tribol., Trans. ASME, 108(2), 190, 1986. 36. Conry, T. F. and Cusano, C., On the stability of porous journal bearings, J. Eng. Ind., Trans. ASME, Ser. B, 96(2), 585, 1974. 37. Cusano, C. and Conry, T. F., On the transmissibility of short porous journal bearings, J. Lub. Technol., Trans. ASME, Ser. F, 100(2), 296, 1978. 38. Cusano, C. and Funk, P. E., Transmissibility study of a flexibility-mounted rolling element bearing in porous bearing squeeze-film damper, J. Lubr. Technol., Trans. ASME, Ser. F, 99(1), 50, 1977. 39. Mak, W. C. and Conway, H. D., The lubrication of a long, porous, flexible journal bearing, J. Lubr. Technol., Trans. ASME, 99, 449, 1977. 40. Jain, S. L., Sinhasan, R., and Singh, D. V., Elasto-hydrodynamic study of vertically loaded flexible porous partial bearings, Indian J. Tribol., 26, 307, 1988.
Copyright © 1994 CRC Press, LLC
SELF-CONTAINED BEARING LUBRICATION: RINGS, DISKS, AND WICKS Richard C. Elwell
INTRODUCTION
Oil is supplied to many journal bearings by self-contained devices such as oil rings, wicks, disks, pumping holes, drip feeders, and chains in order to achieve simplicity and reliability. These self-lubricated systems are generally limited to modest shaft speeds because oil delivery is less than is required for a full oil film. Some thrust bearings can be lubricated by disks; wicks and rings are generally not used. Design of bearing systems lubricated by such devices can be supported by “starved bearing” performance calculations. While background for these analyses follows in this chapter, nearly all self-lubricating designs require careful prototype testing. Minimum oil film thicknesses between shaft surfaces and self-lubricated bearings are much smaller than in the pressure-fed bearings; the surfaces may be only a few micrometers apart. Further, instead of heat being carried from the film by oil flow, it now must be conducted away through the metal parts. Exceptions to this rule are certain disk-lubricated designs. Some general design rules for these bearings are as follows: •
• •
•
•
Journal surface finish must be excellent; 0.4 µm (16 µ-in.) CLA is common, 0.2 down to 0.1 µm (8 to 4 µ-in.) CLA may be required in heavily loaded applications. In extreme cases, grinding or lapping needs to be supplemented by special techniques such as rolling. Self-lubricated journal bearings need axial grooves in the bore to distribute oil longitudinally from the feeding device into the working film. Downstream edges of these grooves must be carefully blended to induce as much oil inflow as possible. Oil distribution grooves must have vents at their outboard ends to release any accumulated air. Otherwise the small oil flows in the grooves may be blocked. Cooling the bearing housing by either air or water will improve bearing load capacity and it may be necessary in a borderline design. To prevent running out of oil, install an easy-to-see oil level indicator. For added safety, put a temperature sensor in me bearing loaded zone.
OIL-RING BEARINGS
Oil-ring bearings have been in wide use for over a century in electric motors, pumps, fans, and other large machines (Figure 1). Preferred bearing proportions are as follows: L/d ratio: 0.6 to 1.1 (based upon net bearing working length) D/d ratio: 1.5 to 1.9 Bearing bore/shaft journal clearance ratio: 0.0020/0.00275
Common ring cross sections are shown in Figure 2 with typical dimensions given in Table 1. The trapezoidal ring is often a zinc die-casting in bores up to about 170 mm. Larger rings are machined from brass or bronze. The T-section type is generally made from rolled brass sheet and is used in large rings because it can be split and hinged. Ring bores are sometimes grooved to increase oil delivery at low speeds, especially in long bearings; high speed performance is not improved. The design in Figure 3 has been used 0-8493-3903-0/94/$0.00 + $0.50 Copyright © 1994 CRC Press, LLC © 1994 by CRC Press, Inc.
515
516
CRC Handbook of Lubrication and Tribology
FIGURE 1. Oil-ring bearing elements.
FIGURE 2. Ring cross-sections.
successfully in pairs to supply a 127 mm bearing (L/d = 1.4), using ISO VG32 oil. Ring bore diameter is 200 mm. Note crowned bore and edge radii. Sharp-edged grooves or teeth are not recommended out of concern for shaft scoring or damage in bearing load zone. In a typical application, the bearing is split horizontally for assembly in a split housing (Figure 4). The housing in turn is contained in a casing which also forms an oil sump beneath the bearing. The ring must be reasonably flat and round to insure even rotation. Typical bearing grooving for rings, oil distribution, and collection of end flow is also shown in Figure 4. Often, one or both ends of the journal bearing will have babbitted, grooved thrust faces for carrying light axial loads. Copyright © 1994 CRC Press, LLC
517
Volume III
FIGURE 3. SAE 62 ring with 1.5 x 1.5 grooves (dimensions in mm).
FIGURE 4. Oil-ring journal bearing with 2 ring slots.
FIGURE 5. Ring speed vs. shaft speed.
Bearing shell material is usually steel or bronze, with allowance for thermal expansion. Use of cast iron is diminishing because of environmental problems with babbitting processes. Babbitt is either tin- or lead-based, nominally 0.8 mm (0.03 in.) thick.
Oil-Ring Bearing Operation Basic operation is evident from Figure 1. As the shaft turns, the ring follows, drawing oil up from the sump below. Two major investigations1,2 of a variety of ring types and sizes demonstrated their general behavior to be as shown in simplified manner in Figures 5 and 6. Typical bearing oil requirements are superimposed on ring delivery in Figure 6. The upper curve shows flow needed for a full fluid film, sometimes called the “classical supply rate”. The
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
518
FIGURE 6. Oil flows over speed range. estimated minimum rate shown for “starved” lubrication is 10% of the classical rate. Note that ring delivery falls below this minimum bearing requirement and approaches zero at high speeds. This is a fundamental limitation of ring-oiled bearings.
Ring Speed Calculations The ring speed/shaft speed relationship in Figure 5 can be calculated from Lemmon and Booser.1 In the “no-slip” range, ring speed (N) and shaft journal speed (n) are simply related as follows: The “no-slip” range ends at ring slip speed Ns, when viscous oil drag on the lower arc of the ring begins to exceed driving friction force from the rotating journal. Ns may be derived from Lemmon and Booser’s Equation 7, as follows:
Full-film drive (nf) begins at about 2.9 times the slip speed (ns). Once established, the following relation1 for full-film ring speed reflects a balance between viscous drive and viscous drag forces: Equation 3 is based upon the ring being immersed in the oil bath to a depth of 15% of its diameter. For other immersion depths, the correct factors1 are as follows:
Note that the ring may be slowed by as much as 27% by increasing its immersion in the oil bath from 5 to 25% of its diameter. Copyright © 1994 CRC Press, LLC
Volume III
519
Oil rings in most smaller machines operate in the full-film drive region. Large, slowerspeed applications such as steel mill drive motors or ship lineshaft bearings may involve noslip or partial-film ring drives.
Ring Oil Delivery Rate Having calculated ring speed, oil delivery to the bearing can now be computed. Computation methods change at a ring “transition speed” (Vt), above which an increasing percentage of oil picked up from the sump is thrown off and never reaches the bearing. Transition speed is obtained as follows, by rearranging Lemmon and Booser’s Equation 2 for ungrooved rings:
Below Nt, compute oil delivery by the ring from Equation 5:
Above the transition speed, use Equation 6 after first calculating Qt and nt from Equations 3 to 5:
Equation 6 does not allow for oil lost from the ring by centrifugal action or by air blowing through the bearing upper ring slot. As higher values of shaft speed (n) are introduced, the equation will continue to predict more oil delivery, even though in small increments. As a safeguard against inaccuracies in Equations 5 and 6, standard practice assumes that only 50% of the oil delivered by the ring actually gets into the bearing working film. A practical speed limit has been found in design experience and also in the experimental work:1,2 Do not use ring-oiled bearings at peripheral velocities greater than 14 m/s (46 ft/s) without a rigorous experimental program to support the design.
Oil Ring Calculations-Numerical Example Calculate oil delivered by a pair of trapezoidal-section rings to a 127 mm (5 in.) diameter by 140 mm (5.5 in.) long bearing operating at 1200 rpm. Referring to Figure 2, ring dimensions are as follows: A = 15.75 mm (0.62 in.), B = 12.7 mm (0.50 in.), C = 1.27 mm (0.05 in.), and D = 203 mm (8.0 in.). Each SAE 62 bronze ring weighs 1.03 kg (2.27 lb.). Oil is ISO VG 32 at a sump temperature of 54.4°C (130°F); absolute and kinematic viscosities at this temperature are 0.0138 Pa s (13.8 cP) and 16.4 (10-6) m2/s (16.4 cSt), respectively.3 (See Godfrey4 for SI units and conversion factors.)
Ring Speeds Ring speed at the end of the “no-slip” range from Equation 2, after converting the variables to consistent units (µ = 0.0138 Pa s/9.8 m/s2 = 0.0014 kg s/m2, D = 0.203 m) is as follows: Shaft speed at the beginning of slip is, from Equation 1: Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
520
FIGURE 7. Ring speeds in partial-film drive.
Full-film ring drive begins at shaft speed 2.9 times the speed at the beginning of slip: From Equation 3, ring speed is as below at nf = 39.7 rps, and v = 16.4 x 10-6 m2/s: Because full-film ring drive would begin at a higher shaft speed (39.7 rps) than the operating speed (20 rps), the ring is running with partial-film drive (Figure 7). Ring speed in this range is obtained by interpolation on a log-log plot as shown. For n = 20, N is found to be 6 rps. For operating speed either above or below the partial-film drive range, ring speed would have been calculated directly from either Equation 1 or 3, as appropriate.
Bearing Oil Flow Now that operating ring speed has been found to be 6 rps, oil delivery to the bearing can be computed. First, ring transition speed is calculated from Equation 4:
Since the transition speed occurs in the no-slip range, shaft speed at transition is obtained from Equation 1:
Oil delivery at the transition speed is, from Equation 5: Equation 6 finally gives oil delivery by a single ring at the operating speed: Copyright © 1994 CRC Press, LLC
521
Volume III
FIGURE 8. Disk-lubricated lineshaft bearing assembly. (From Gardner, W. W., J. Lubr. Technol., Trans. ASME, 99(2), 174, 1977.)
Applying the usual assumption that only 50% of ring delivery gets into the bearing film, bearing inlet oil flow equals 5.55 X 10-6 m3/s (0.087 gal/min). Since this is marginal oil feed for a 127 mm (5 in.) diameter bearing operating at 1200 rpm, two rings will be used in starved bearing calculations later in this chapter.
DISK-OILED BEARINGS
Disks usually require more axial space than oil rings in a self-lubricated design but offer the following advantages: (1) they operate better on tilted shafts, such as in marine or earthmoving equipment; (2) they are a more positive supply method at very low speeds; and (3) they can be arranged to deliver oil at high speeds and under substantial pressure. Limited data for design use is available from laboratory test programs5–7 and manufacturers’ experience.8 In a simple disk-oiling arrangement for a low-speed marine lineshaft application (Figure 8), a scraper block with an antirotation pin floats on top of the disk and diverts oil from the outside surface of the disk into channels feeding the bearing. This system is self-contained with its own circulating and sealing features. Gardner5 found for a large disk (D = 0.94 m) at low speeds that oil delivered by this type of disk-scraper was proportional to the following parameter, which resembles Equation 5 for oil-ring delivery:
Copyright © 1994 CRC Press, LLC
522
CRC Handbook of Lubrication and Tribology
Oil flow is directly proportional to the DFP only up to a value of 2 x 109, at which point the oil flow is approximately 111.6 x 10-6 m3/s (1.77 gal/min). At higher speeds, an increasing percentage of the oil taken up from the sump is blown away by a combination of centrifugal force and air flow circulating inside the housing. As the DFP increases beyond 2 x 109, actual oil flow begins to fall further and further below that which would be calculated. At 6 x 109, for instance, measured flow was about 75% of theoretical.
Disk Flow Parameter-Numerical Example Calculate the speed at which the DFP reaches 2 x 109 for the case of Gardner’s large disk using his “light oil”. Values of the variables are as follows: disk width B = 0.127 m, diameter D = 0.94 m, and viscosity µ, = 0.065 Pa s/9.8 m/s2 = 0.00663 kg s/m2. Substituting in Equation 7:
In this case, for a large disk with highly viscous oil, transition from a linear flow function of the Disk Flow Parameter begins at low rotational speed, as may be expected. Oil delivery at higher speeds may be estimated from Gardner’s Figure 11. Gardner’s work was extended by Ettles et al.,6 who tested a smaller disk (153 mm) over a wider speed range (0 to 6000 rpm). Although scraper shapes, disk cross section, and shrouding were different, oil was again scraped from the disk outside diameter by both fixed and floating scrapers. Ettles also included data from Albert Rose for 250- and 500-mm disks. Ettles’ apparatus was in a transparent enclosure, allowing observations over the range of speeds. Qualitative descriptions of oil flows are given, along with rough correlations among the major experimental results. For both fixed and floating scrapers, Ettles defines a “critical speed”, nc, above which oil flow decreases with increasing speed. Oil delivery ceases completely at about 2.5 times the critical speed. The critical speed is a strong function of depth of disk immersion in the oil bath, as follows:
At speeds below nc, disk oil delivery to a fixed-gap scraper with a gap up to 3% of the disk diameter can be estimated from:
This equation is of limited usefulness because fixed scrapers are seldom used; it is usually too difficult to maintain the rotating position of the disk precisely enough to keep it from contacting the scraper under any condition. It nevertheless does allow an estimate of oil delivery for a fixed-scraper/disk arrangement.
Fixed-Scraper/Disk Oil Delivery-Numerical Example Calculate oil delivered by the following disk geometry for comparison to data measured by Rose:6 D = 0.25 m, H = 0.04 m (16% immersion), B = 0.025 m, µ, = 0.0818 Pa s/9.8 m/s2 = 0.00835 kg s/m2, n = 800 RPM/60 s = 13.33 rps, U = πDn = 10.47 m/s. In Equation 8:
Copyright © 1994 CRC Press, LLC
Volume III
523
FIGURE 9. Centrifugal pumping disk assembly. (From Kaufman, H. N., Szeri, A. Z., and Raimondi, A. A., Trans. ASLE, 21(4), 315, 1978.)
Since critical speed is greater than operating speed, use Equation 9 to calculate the oil delivery:
Measured flow was 57.2 mL/s, which gives some indication of the conservatism in the flow equation. The Ettles publication should be examined carefully before using the above equations in detail design calculations. An entirely different disk concept, intended for bearing lubrication at higher speeds, was developed by Kaufman et al.7 This design scoops oil from inside a hollow disk (Figure 9). The disk is shrouded at the bottom to minimize power loss from churning. Flow through the whole circulating system is controlled by orifices metering inflow to the disk from the bath. Oil metered to the bottom of the disk is carried by disk rotation to the top, where it is scooped and diverted through a trough into bearing feed holes. The fixed scoop clears the inner disk surface by a small gap, of the order of 3 mm (0.12 in.). With its positive collection, this type of disk provides a more dependable high-speed oil supply than an oil ring. Since centrifugal force on the oil increases with speed, this arrangement overcomes the high-speed problems with externally scraped disks. One way to compensate for limited lowspeed delivery is to add a scraper “dam” on or near the disk outside surface. At low speeds, the disk carries a layer of oil on its outer surface; this is thrown away at higher speeds and kept from reforming by the shroud. Figure 10 shows oil delivery of an assembly with and without the dam for a disk of 312.4 mm (12.3 in.) outside diameter, which would typically be used to supply a journal bearing of 125 mm (4.92 in.) diameter. Oil is ISO VG68 at a bath temperature of 24°C. With the external dam, full flow is achieved at only 50 rpm vs. 100 rpm with the scoop only. Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
524
FIGURE 10. Oil delivery with and without external dam. (From Kaufman, H. N., Szeri, A. Z., and Raimondi, A. A., Trans. ASLE, 21(4), 315, 1978.)
FIGURE 11. Bronze oil circulator ring. (Courtesy of Kingsbury, Inc., Philadelphia.)
Still another way of using a disk to supply lubricating oil is inside an “oil circulator” (Figure 11), sometimes called a “viscous pump”. In contrast to the disk arrangements shown thus far, this device sometimes furnishes oil to both journal and thrust bearings at substantial pressure. Oil from the reservoir shown in Figure 12 is drawn into the circulator surrounding the thrust collar through one of the two holes in the bottom (marked “A” in Figure 12) and is then carried around in a shallow groove by collar rotation. At the end of its travel, a dam in the groove at the bottom centerline generates oil pressure to push it through hole “B” to spaces between the lowest thrust bearing pads on both sides. This oil is circulated to all pads by collar rotation, as is usual in thrust bearings. Some oil from the circulator discharge hole “B” on the Copyright © 1994 CRC Press, LLC
525
Volume III
FIGURE 12. Cross section through standard bearing assembly with Oil Circulator. (Courtesy of Kingsbury, Inc., Philadelphia.)
bottom center is led through connecting holes to the journal bearing next to the thrust bearings, where it is admitted to conventional horizontal distribution grooves. A second journal bearing can also be connected to this supply. The circulator and its mating oil passages are designed for rotation in either direction. When rotation starts, the antirotation lug at the top bears against the side of a slot to align the oil entrance and discharge holes at the bottom. Most of the oil flow from the circulator goes through the thrust bearings and is thrown radially outward by the collar. This flow is collected in a passage running over the top and down a pipe to a cooler beneath the assembly. Oil delivery performance of a standard circulator is presented in Figure 13. In this assembly, there are a pair of 6-pad thrust bearings with a collar between them. The journal bearing in the assembly is 114 mm (4 1/2 in.) by 114 mm. Both flow and pressure are almost linear functions of speed culminating with a maximum flow of 1.7 L/s (27 gpm) and pressure of 110 kPa (16 psi) at 3600 rpm. These are substantial values, comparable to deliveries by mechanical pumps except at lower mechanical efficiency.
STARVED BEARING ANALYSIS
In theoretical analysis of the starved journal bearing, lubricant inlet flow to the bearing film is reduced in a series of steps below the value required by continuity for a full (or “classical”) fluid film. The reduced flow results in a working bearing film that is not as wide or thick as the full film. At each step the computer calculates the resulting journal eccentricity, load and friction. Two prominent analyses exist: Connors data9 will be used here for convenience, even though Artiles and Heshmat10 employ a more modern computational approach. Effects of varying oil supply rates on load capacity of a cylindrical journal bearing may be seen in Figure 14. For a given bearing (fixed viscosity, speed, load, diameter, and Copyright © 1994 CRC Press, LLC
526
CRC Handbook of Lubrication and Tribology
FIGURE 13. Measured performance of 229 mm diameter Oil Circulator using ISO VG 32 oil at 49°C. (Courtesy of Kingsbury, Inc., Philadelphia.)
FIGURE 14. Chart for determining minimum film thickness in starved journal bearing. (From Connors, H. J., Trans. ASLE, 5(2), 404, November, 1962.)
clearance), the Bearing Characteristic Number, S, is constant. The “Classical Input Flow” line gives the minimum oil film thickness (hn) for this value of S, assuming that precisely the right amount of oil is supplied for equilibrium. Supplying greater amounts of oil (“flooding”) has no effect, but reducing flow below the classical value is seen to substantially reduce minimum film thickness between the bearing and journal. This is the regime in which ring-oiled, wickfed, and other self-lubricated bearings typically operate. Copyright © 1994 CRC Press, LLC
Volume III
527
Figure 15. Chart for determining friction in starved journal bearing. (From Connors, H. J., Trans. ASLE, 5(2), 404, November, 1962.)
Effect of oil supply rate on bearing power loss is not dramatic, as Figure 15 shows. Here again, for a fixed value of S, the “Classical Input Flow” line divides the operating regimes between flooded and starved, or “reduced lubricant supply”. Putting more oil into the film than required does not change theoretical friction. Since the film is starved, power loss is only slightly affected because the decreased circumferential extent of the film is commonly offset by a higher shear rate in a thinning film. Because of the many assumptions, a calculated operating temperature may be regarded as only an approximation of the bearing metal beneath the load. It is useful for comparative purposes, however, in examining effects of different oil viscosities and design changes such as clearance increase. The procedure is as follows: 1.
2. 3. 4. 5.
Assume an operating temperature (Tav) somewhat higher than bearing oil inlet temperature (Ti), which in turn is usually assumed equal to the sump temperature. Obtain absolute viscosity for the oil at Tav from a source such as Reference 3. Compute S, Input Flow Variable, and oil flow. Obtain value of f(R/C) in Figure 15. Multiply bearing load by f to obtain friction force on the journal. Combine friction force at radius R and shaft speed n to compute power loss.
Copyright © 1994 CRC Press, LLC
528
6. 7
8.
CRC Handbook of Lubrication and Tribology
Calculate oil temperature rise by assuming all the power (or some traction of it) calculated in step 5 is carried out by bearing oil flow
Add one-half the temperature rise from step 6 to film inlet temperature (Tj) to obtain operating temperature. Compare to value assumed in step 1; recalculate if necessary. Go to Figure 14, look up hu, calculate minimum film thickness from journal eccentricity. Solve a number of cases to construct load vs. film thickness curves.
Starved Bearing Calculations — Numerical Example Calculate load capacity, power loss, and approximate temperature rise of the bearing described in the oil-ring analysis earlier in this chapter: R = 0.0635 m (2.5 in.), L = 0.140 m (5.5 in.), n = 20 rps, C = 165 x 10-6 m (0.0065 in.), P - 58,000 kg/m2 (82.3 lb/in.2), oil = ISO VG 32 (see Reference 3, p. 414), Ti = 54.5°C (130°F), Qi = 11.1 x 10-6 m3/s (0.68 in.3/s) (flow into bearing from two rings), p = 831 kg/m3 (0.030 lb/in.3) (nominal density), specific heat = 1675 J/kg °C (0.40 BTU/lb °F).
Assuming AT = 13°C, Tav = 54.4 + 13/2 = 61°C, µ = 0.0012 kg s/m2 at Tav: From Figure 15, f(R/C) = 1.58; friction coefficient f = 1.58 x (0.000165/0.0635) = 0.0041 Bearing load = 58,000 x DL = 58000(0.127)(0.140) = 1031 kg Friction torque T = f X load x R = 0.0041 X 1031 x 0.0635 m = 0.268 kg m
Power loss = 2imT(9.8) = 2ir(20)(0.268)(9.8) = 330 J/s (0.44 HP) This is higher than the originally assumed δ T of 13°C. Assuming 20°C and repeating: S = (0.0011/0.0012)0.061 = 0.056 Qi/RCnL = 0.378 From Figure 15, f(R/C) = 1.50, f = 0.0039 Power loss = 330(0.0039/0.0041) = 314 J/s δT = 21.4(314/330) = 20.4°C; this is accepted as the solution Tav = 54.5 + 10.2 = 64.7°C (149°F)
Minimum oil film thickness between the bearing and journal can now be calculated: from Figure 14, for S = 0.056 and Input How Variable = 0.378, hn/C= 0.182. hn, = 0.182(165)(106 ) = 30(10-6) m (0.0012 in.). This is acceptable in usual practice.
WICK-LUBRICATED BEARINGS
In many early industrial bearings, notably in machine tool, railroad, and small electric motor service, oil was fed by wicks, primarily made of wool felt. Though this technique is confined to low surface speeds, say 4 m/s (13 fps), surprisingly heavy loads can be carried with proper attention to certain manufacturing details. There has never been a systematic study of wick-lubricated bearings, as there has been of ring-oiled bearings. Copyright © 1994 CRC Press, LLC
529
Volume III
FIGURE 16. Wick-lubricated journal bearing.
FIGURE 17. Bottom-fed oil delivery to moving journal by SAE F1 wick, for viscosities shown in centistokes.
Wick-oiled bearings are usually arranged as in Figure 16, although contact between the wick and journal is often on either the top or bottom centerline. Oil is lifted a height “h” above a reservoir level and is wiped off the working end of the wick by shaft motion. The reservoir is generally not an open container as sketched, but more usually another, larger wick made of material specifically designed for oil storage.12 Wool has been largely supplanted by synthetic materials, but is still covered in many felt specifications.13 SAE F1 (ASTM 16R1) is the most commonly used grade for delivering oil directly to the shaft journal. Amount of oil lifted and delivered to a moving shaft varies as shown in Figure 17. These data and others given by Fuller14 illustrate effects of the main design variables. The limited oil delivery relegates wick-fed bearings to low speeds because the oil flow cannot carry away a significant portion of the heat generated. Since bearing power loss is largely unaffected by load, however, high loads can still be carried at low speeds. Equation 10 has been proposed by E. R. Booser as an approximation to the Figure 17 data allowing for the effect of the developed wick length: Oil Wick Delivery Calculation — Numerical Example Calculate oil delivered to the same bearing analyzed earlier in the starved bearing analysis: R = 0.0635 m, L = 0.140 m, C = 165 x 10-6 m, n = 20 rps, increase load to P = 420,000 kg/m2. Assume a wick of SAE Fl felt, with A = 0.02 x 0.1 m = 0.002 m2 and 1 = 0.13 m, lifting oil through a height h = 0.051 m. Oil is ISO VG 32 at an effective wick temperature of 30°C; viscosity is 50 cSt, µ. = 0.00428 kg s/m2. From Figure 17, for h = 5.1 cm and viscosity of 50 cSt, delivery rate = 12 cm3/h cm2 = 33.3 x 10-6 m3/ s m2. Oil flow into the bearing is therefore: If, instead, Equation 10 is used to estimate wick oil delivery to the bearing (for SAE F1, Vo = 0.745 and hu = 0.191 m): Copyright © 1994 CRC Press, LLC
530
CRC Handbook of Lubrication and Tribology
Equation 10 predicts wick length (0.13 m) reduces oil delivery by about half from the value predicted from Figure 17. In either case, oil delivery is far less man by oil rings in the starved bearing analysis example above where flow rate was 11.1 x 10-6 m3/s. To use the starved bearing analytical data in Figure 14, compute the Input Flow Variable as follows:
Reference to Figures 14 and 15 will show that it is impossible to use an Input Flow Variable this small to calculate bearing performance, and another approach is needed. One design with which the author is familiar employed a bearing 229 mm (9 in.) diameter by 121 mm (4.75 in.) long, nominal diametral clearance of 0.051 mm (0.002 in.), a load of 11,600 kg (5,272 lb) at 1 rev/s (60 rpm), and using an ISO VG 68 oil with special extremepressure additives. Laboratory tests showed that metal temperature in the load zone was 70°C. The lead babbitt bearing bore was contoured to accommodate shaft bending, and the journal was ultrasmooth finished. Lesser loads could be carried without resorting to such extreme machining practices. Since this design achieved thousands of hours of service in many applications, a thin fluid film must exist between the bearing and journal. As an alternative to the Connors method and since oil starvation gives a circumferentially short oil film in which elastic deflections of the journal and bearing surfaces play a major role, the minimum film thickness can be estimated from elastohydrodynamic (EHD) theory, as presented by Cheng.11 Elastohydrodynamic Lubrication Calculation - Numerical Example A number of assumed values must be used in the computations, as follows:
Cheng’s Equation 13,11 generated by Hamrock and Dowson for the minimum film thickness between two surfaces in fully flooded EHD contact, is rearranged as follows:
The multiplying term (1 - e-0.68k) in the original equation approaches unity for an uncrowned journal and bearing. The dimensionless ratios U, G, and W in the equation are calculated as below:
Copyright © 1994 CRC Press, LLC
531
Volume III
FIGURE 18. Small bearing in FHP motor end shield.
Now solving for the minimum oil film thickness between the bearing and journal:
While this calculated film thickness is approximate, the analysis does give a procedure for evaluating effects of changing the many variables in the problem. Small Wick-Fed Bearings In this century, hundreds of millions of wick-lubricated electric motors and their driven equipment (fans, pumps, office machines) have accomplished billions of hours of service all over the world. Wick lubrication has been ideal for these small machines because of its compactness and low oil flows. In the first STLE Handbook, Booser’s15 Figure 8 showed a 1968 state-of-the-art wick-fed bearing system for fractional-horsepower (FHP) motor service with the following features: • • • •
A babbitted steel bearing insert, with wick window on top. A feed wick contacting the journal through the wick window. Two separate backup reservoir wicks. Cast motor end shield with spring-cap oiler for relubrication.
Developments in bearing and wick materials over the past 25 years have allowed much different designs, such as in Figure 18, which shows an FHP motor end shield employing the following: Copyright © 1994 CRC Press, LLC
532
• • •
CRC Handbook of Lubrication and Tribology
A porous bronze or iron bearing with no feed window A single reservoir, filled either with felt12 or injectable oil/fiber wicking mixture Pressed-steel end shield and cover with no re-oiling provision
Whereas the older design had the virtues of rugged construction and very long life (if reoiled periodically), significantly more costs were incurred through the use of castings, three manually assembled wicks, and the oiler. The old and new designs are not completely equivalent. Because oil feed to the working bearing film is now through the porous metal wall, oil is admitted at a very low rate. This limits load and speed; see DeHart16 for design data on porous metal bearings and Connors17 for data on felt wicks. Journal size does not usually exceed 19 mm (3/4 in.). On the plus side, some alignment capability is built in, using rounded ends on the bearing and a spring plate to clamp the bearing against the end shield.
A B C D d E F f g h hn hu H I L N N f, N s , N t n nc nf’na’nt P Q,Qi’Qt R S Ti’To’Tav U Vo Vt W w µ v p
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
NOMENCLATURE
Oil ring width; also wick contact area on journal Oil ring height; also disk width Bearing/journal radial clearance; also oil ring cheek width Oil ring bore diameter, also disk diameter Shaft (or journal) diameter Width of T-section oil ring; also elastic modulus Friction force on surface of journal; also web thickness of oil ring Coefficient of friction, F/W Gravitational constant Oil wick delivery height Minimum oil film thickness between journal and bearing Ultimate wicking height Depth of disk immersion in bath, also T-section oil ring height Wick developed length between bath and journal Net bearing working length Ring rotation speed; journal rotation speed in starved bearing analysis Ring speed at start of full-film drive, at beginning of slip, at transition Shaft rotational speed Disk critical speed Shaft speed at start of full-film ring drive, at beginning of ring slip, at Nt Bearing load on projected area, W/LD Oil flow rate, flow rate into bearing film, delivery rate at Nt d/2 Bearing Characteristic Number, µ-N/P (R/C)2 Temperature of bearing inlet oil, outlet oil, average (Ti+ To)/2 Disk surface velocity Volume fraction of oil in saturated wick Ring transition velocity Ring weight; dimensionless load in EHD analysis Width of ring riding on journal; load in EHD analysis Oil absolute viscosity Oil kinematic viscosity Oil density
Copyright © 1994 CRC Press, LLC
Volume III
REFERENCES
533
1. Lemmon, D. C. and Booser, E. R., Bearing oil-ring performance, Trans. ASME, Ser. D, 82(2), 327, 1960. 2. Heshmat, H. and Pinkos, 0., Experimental study of stable high-speed oil rings, J. Tribal., Trans. ASME, 107(1), 14, 1985. 3. Raimondi, A. A. and Szeri, A. Z., Journal and thrust bearings, in CRC Handbook of Lubrication, Vol. II, Theory and Design, Booser, E. R., Ed., CRC Press, Boca Raton, FL, 1984, 413. 4. Godfrey, D., SI units and conversion factors, Appendix n, CRC Handbook of Lubrication, Vol. I, Application and Maintenance, Booser, E. R., Ed., CRC Press, Boca Raton, FL, 1983, 571. 5. Gardner, W. W., Bearing oil delivery by disk-scraper means, J. Lubr. Technol., Trans. ASME, 90(2), 174, 1977. 6. Ettles, C. M. McC., Adamson, W. R., and Yiallouros, M., Some characteristics of the disk-scraper oil-feed mechanism, Trans. ASLE, 23(4), 442, 1980. 7. Kaufman, H. N., Szeri, A. Z., and Raimondi, A. A., Performance of a centrifugal disklubricated bearing, Trans. ASLE, 21(4), 315, 1978. 8. Anon., Six-shoe thrust bearings and journal bearings. Catalog CP, Kingsbury, Inc., Philadelphia, PA. 9. Connors, H. J., An Analysis of the Effect of Lubricant Supply Rate on the Performance of the 360° Journal Bearing, Trans. ASLE, 5(2), 404, 1962. 10. Artiles, A. and Heshmat, H., Analysis of starved journal bearings including temperature and cavitation effects, J. Tribal., Trans. ASME, 107(1), 1, 1985. 11. Cheng, H. S., Elastohydrodynamic lubrication, in CRC Handbook of Lubrication, Vol. II, Theory and Design, Booser, E. R., Ed., CRC Press, Boca Raton, FL, 1984, 139. 12. Anon., The Fiberstruct® Way, American Felt & Filter Company, New Windsor, NY. 13. Anon., Technical Data Bulletin on Wool Felt, Booth Felt Company, Chicago, IL. 14. Fuller, D. D., Theory and Practice of Lubrication for Engineers, 2nd ed., John Wiley & Sons, New York, 1984. 15. Booser, E. R., Electric motors, in Standard Handbook of Lubrication Engineering, O’Connor, J. J., Boyd, J. and Avallone, E. E., Eds., McGraw-Hill, New York, 1968, 35– 1. 16. DeHart, A. O., Sliding bearing materials, in CRC Handbook of Lubrication, Vol. II, Theory and Design, Booser, E. R., Ed., CRC Press, Boca Raton, FL, 1984, 413. 17. Connors, H. J., Fundamentals of wick lubrication for small sleeve bearings, Trans. ASLE, 9(2), 299, 1966.
Copyright © 1994 CRC Press, LLC
WATER AND PROCESS FLUID BEARINGS R. L. Orndorff, Jr.
INTRODUCTION
Many bearing materials can be used in a water-or process fluid-lubricated application. Some of these materials are discussed in Part 2 of this Volume and in Volume 2 of this series.1 This chapter will mainly discuss elastomer (rubber) bearings, because these are the most commonly used and they are a good engineering choice in most respects. Most rubber bearings consist of multiple flat lands,2 either in full molded form (Figure 1a) or, in the case of large shafts, assembled staves (Figure 1b). Also discussed are design of the bearing surface, desirable shaft characteristics, design of the overall bearing, lubrication regimes, and other commonly used materials.
FUNDAMENTALS
Lubricants Water is a constituent of many process fluids because it offers a number of advantages. It is an inexhaustible natural resource in most locations, is low in cost, has high specific heat, and usually causes no pollution or reclamation problems. Since water viscosity is many fold less than that of oil, however, minimum water film thickness values are typically less than one tenth the thickness of oil films. This puts more emphasis on the boundary and mixed film operating characteristics of the bearing material and on the need for proper bearing design. Water can be considered incompressible for most bearing applications. Cavitation can take place when operating close to its relatively low boiling point, or at much lower temperatures if the lubricant supply is somewhat limited. Any process fluid can be used to lubricate a bearing so long as there are no chemical compatibility problems. A major problem is always the presence of solid particles in the lubricant. These can be metal, sand, grit, clay, or some partially dissolved solid from a chemical process. Traditional good engineering practice indicates that solid particles of a size above the minimum expected film thickness have to be filtered out of the lubricant in order to obtain acceptable wear. While this is true for harder polymer (plastic), ceramic, and carbon-graphite bearings, rubber bearings can be designed to handle much larger particles, as discussed in the wear section of this chapter.
Bearing Materials The most important consideration in a properly designed bearing is that the bearing material, the supporting rigid shell (if used), and the adhesive joining the two are compatible with the process fluid. Chemical resistance tables are available in various handbooks and from suppliers, listing the acceptability of various polymers (Table l).3 Figure 2 illustrates the operating temperature range and oil/fuel/weather resistance of the more common elastomer families.3 There are very different swell characteristics between elastomer families and even within a family depending on the recipe ingredients. Most process-lubricated rubber bearings are made of nitrile (NBR) compounds with a useful temperature range of approximately -55 to 120°C. For any given bearing, however, this range is highly dependent on the lubricating characteristics of the process fluid. Generally, with water the upper temperature limit is assumed to be 71°C. If the lubricant temperature exceeds 32°C, the bearing needs to be installed at room temperature with a looser fit on the Copyright © 1994 CRC Press, LLC
535
CRC Handbook of Lubricants and Tribology
536
FIGURE 1. Rubber bearings: (a) Landed full molded; (b) stave.
shaft than is customary to compensate for loss of clearance due to thermal expansion. While nitrile rubber tends to be the best all-around elastomer, other types used for specialized service include some polyurethanes, fluorocarbon (FKM), epochlorohydrene (ECO), chloroprene (CR) and occasionally natural rubber (NR). All of these elastomers tend to have higher wet friction than the nitrile equivalent because they do not have the same desirable elastic/creep characteristic, they swell too much, or their design practices in bearings are less well established. Hydrophobic elastomer compounds usually are better performers.
Running Clearance This depends on the shaft tolerance, minimum bearing-to-shaft clearance, and the permissible bearing manufacturing variation. Table 2 shows these values for a range of nominal shaft diameters used in commercial marine bearings, naval bearings, and industrial bearings.4 With any elastomer compound, too tight an installed bearing clearance or excessive swell results in a self-perpetuating catastrophic failure mode: a too-tight bearing heats up, the elastomer expands to increase the tightness and friction, resulting in more generated heat and still higher temperatures until failure eventually takes place. Polymer bearings cannot be operated with the close clearances of metal bearings because they have much higher coefficients of thermal expansion.2 In combination with the greatly reduced thermal conductivity, this can create problems if not properly considered in the design. The thermal conductivity problem can be alleviated by assuring a proper minimum process fluid flow rate through the bearing. Most process fluid lubricated bearings consist of bearing material adhered on its outside surface to a rigid metallic or fiberglass-reinforced plastic (FRP) shell (Figure 1). The rigid shell eliminates the need to consider installation closedown in the clearance calculation. However, the much lower coefficient of thermal expansion of the rigid shell restricts the bearing material to expanding only in one direction - toward the shaft. The coefficient of volumetric thermal expansion must then be used, which is three times the linear coefficient value reported in most handbooks. This approach works well except at the bearing ends, where the elastomer will not expand quite as much toward the shaft with increasing temperature because of the additional end expansion area. A suggested procedure is to perform a soak test of the prototype sample bearing submitted by the manufacturer in the process fluid at the intended operating temperature with a shaft installed. Such a test will evaluate the swell characteristics of the bearing material and shell, Copyright © 1994 CRC Press, LLC
Volume III
537
adhesion of the two, and general degradation as well as determine the required operating clearance. Plastic bearings can have definite handbook tabulated coefficients of thermal expansion values because of their homogenous nature. Every rubber compound, however, has a different coefficient of expansion, depending on the type of rubber, amounts of carbon black and other compounding ingredients, and the processing (mixing) and curing history.
Permissible Radial Load Customary recommended projected pressure is conservatively 276 kPa (40 psi) for most process and water-lubricated bearings. Projected pressure is calculated as the radial load divided by the product of the bearing length multiplied by the shaft diameter. Many bearings are loaded well under this pressure. The upper load limit is not a definite value for a particular elastomer bearing material since the compressive deflection also depends on the actual physical design details such as rubber hardness and thickness. Greater loads may be acceptable if operation is continuously Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
538
FIGURE 2. Guide to useful temperature range of commercial elastomers.
at higher shaft speeds. Lighter loads should be used if the shaft is frequently stopped and started or if the lubricant is extremely abrasive. Loads can be 689 kPa for some ship propeller shaft bearings and higher under special conditions. Rubber thickness and land width become very important for high loads. With too thick a rubber layer, the leading edge of the land approaches too closely to the shaft and scrapes off the lubricant before it can enter the contact patch. For a typical application, the ratio of shaft diameter to land circumferential width (D/W) should be in the range of 4 to 6. At a load of 689 kPa, the rubber thickness should be in the range of 3.18 to 6.35 mm. In some cases, the trailing edge of the land may be the critical point because of the negative pressure pulse in the fluid when it reaches the trailing edge.5 While the limits for D/W are somewhat vague because of Copyright © 1994 CRC Press, LLC
Volume III
Copyright © 1994 CRC Press, LLC
539
540
CRC Handbook of Lubricants and Tribology
all the variables involved, an example illustrates the situation. For a 584.2 mm diameter shaft, 276 kPa projected radial load, slow speed operation, rubber thickness of 7.95 mm, and 70 Shore A rubber hardness, the actual D/W ratio turned out to be 10. The bearing was overloaded. Ideally, for these conditions the D/W ratio of 4 would be best. Because this case was a retrofit, added stiffness to keep the shaft away from the land edges was accomplished by increasing the rubber hardness to 80 Shore A from 70 Shore A. (As a reference, tread rubber in a radial tire has a hardness of about 60 Shore A.) Rubber thickness could have been reduced to a lesser value to increase the stiffness, while retaining the softer rubber to give lower shaft wear. Copyright © 1994 CRC Press, LLC
Volume III
541
FIGURE 3. Tribological regimes.
One cautionary note concerns the shape of the leading edge of the bearing land. A quick and easy solution would be seemingly to simply round or bevel the leading edge to promote lubricant flow into the contact zone. This is actually the wrong thing to do. Rounding or beveling the edge dramatically increases friction, since this increases the approach angle of the shaft to the bearing land. The higher approach angle results in more exit flow area for the lubricant being hydrodynamically pumped by the rotating shaft to more easily leak out of the ends of the bearing. Just breaking the edge even seems to have a negative effect.
Self-Aligning Capability One inherent advantage of elastomer bearings is their alignment capability. This enables them to overcome many errors of omission and commission created by machine designers and millwrights. For very large bearings, however, still more alignment capability is sometimes required. This can be obtained by molding a circular ring of rubber around the external surface of the rigid shell of the bearing. The ring material is usually a natural rubber compound with superior dynamic and fatigue properties compared to the synthetic nitrile elastomer compound usually used for the bearing surface. The rubber mount is a spring and as such has radial, torsional, and axial natural frequencies at various points in the shaft speed spectrum. The radial and axial natural frequencies are generally constant. Torsional isolation performance is more complex because of the variation of the driving force (bearing friction) with shaft speed.
Shaft Speed Landed elastomer bearings can operate at high shaft speeds because they do not pound out as do metal and some plastic bearings. They cushion dynamic forces and allow an eccentric shaft to rotate at its center of gyration when this does not agree with the geometric center. Care must be taken to ensure adequate lubricant flow to remove the extra heat generated by the dynamically worked rubber. Slow speed operating performance of a water bearing is determined by the transition point between the mixed and hydrodynamic lubrication regimes on the coefficient of friction curve (Figure 3),6 generally at a surface speed of about 0.25 m/s for a well-designed bearing. Sometimes excessive wear can be a problem when the bearing is operating below the initial transition point. In many cases, the bearing will wear-in at the low speed, resulting in a new, lower shaft speed transition point and a rapidly reducing wear rate over a relatively short period of time. The key is to provide adequate lubrication to remove the higher frictional heat load. Copyright © 1994 CRC Press, LLC
542
CRC Handbook of Lubricants and Tribology
In other cases, slip/stick noise can result, due to the breakdown of the fluid film at very low shaft speed in the boundary-lubricated regime. Reducing elastomer thickness and making the surface very smooth will eliminate the stick/slip.
Lubricant Supply A water flow rate of 3.8 X 10-3 m3/min (1 gal/min) per inch of shaft diameter is common for a bearing in the normal load range of 207 to 276 kPa. This is conservative; bearings have operated successfully at less than 25% of this value. With extremely dirty water, flow rate may be doubled. In applications such as a marine strut bearing or an immersed pump bearing where the lubricant is free to steadily flow in and out of the bearing and there is no prior upstream lubricant heating, no pumped water is required. Normal ship or fluid motion is sufficient. There are sometimes problems in a ship application if the bearing is installed in a stem tube with limited free flow of the lubricant. In this case, pumped water must be supplied.
Surface Finishes Ideally, the surface of a rubber bearing should be relatively smooth: 0.13 µm or better Ra finish is desirable. Shaft surface finish should be in the range of 0.4 µm. Process-fluid lubricated bearings do not depend on a transfer film for successful operation as do most dry bearings, so the shaft finish can be very smooth. This reduces friction during break-in of the bearing. It is impossible to mold all rubber bearings to exact bore size because tooling is not available for every conceivable size, so in many cases the bore must be machined or ground. The rubber surfaces should then be finish-sanded with minimum material removal as a final step, to polish the rubber surface. Any resulting lips at edges of the ground region should be sanded off and faired into the flat land. Elastomer bearings, especially those having a hardness of around 70 Shore A, continue to polish or lap the shaft surface during the first 100 h of operation. There is a need for surface smoothness protective measures if installed bearings remain idle for a long time before a new ship or machine is put into service. Mussels and barnacles can build up in stern tube bearings of nonoperational ships at dockside; metal chips and dust can accumulate in a pump being assembled in a shop atmosphere. The first few rotations of such a contaminated bearing assembly will ruin the carefully prepared surface finish of bearings and shafts. A tarpaulin or heat-shrunk plastic wrap can protect a pump. A continuous flow of water through a ship bearing will help. So will filling it with a water-soluble grease that can be washed out later with a high-pressure water jet. Conventional greases should be avoided since they will not wash out and may block lubricant flow in the water grooves between lands once the vessel is operating.
Shaft Material and Hardness The best situation is the hardest possible shaft surface running against the softest bearing material. Figure 47 shows the shaft wear of various materials operating against a particular water-lubricated elastomer bearing material. The lubricant included abrasive. If the bearing material is made harder or the shaft softer, the combined system wear will increase. It is the combined wear that determines the increase in clearance during operation. Successful operation of landed bearings is not nearly as sensitive to large changes in bearing-to-shaft clearance as is true with circular bore bearings. Each land functions as an independent bearing surface. For water-lubricated applications, shaft sleeves of the proper hardness are shrunk onto steel shafts. Sleeves of 410 stainless steel of 300 Brinell hardness are usually suitable, while softer 300 series stainless tends to wear in contaminated water. Nickel-containing stainless of Copyright © 1994 CRC Press, LLC
Volume III
Copyright © 1994 CRC Press, LLC
543
FIGURE 4. Shaft wear.
544
CRC Handbook of Lubricants and Tribology
intermediate hardness is a common choice. One innovation employs a nickel-chrome-boron sleeve fused to the outer surface of a carbon steel base material It has a hardness of 60 Rockwell C. Radial thickness of the hard coating is 1.73 mm and it is ground to a 0.15 to 0.20 µm finish. Hardness of at least 40 Rockwell C is preferred, though Rockwell C values of about 20 are common.
Length/Diameter Ratio Marine bearings traditionally have employed length-to-diameter ratios of 4:1 with water lubrication. Some now operate in ships with 2:1 L/D ratio. The main reason for using a longer bearing is to keep bearing pressure down to the recommended 276 kPa. However, a long bearing can create problems because of overhung loads and misalignment. As better materials and designs are developed, L/D ratio can be dropped to 2:1 or lower. The key is to keep the proper relationship between pressure, rubber thickness and hardness, and land width.
LUBRICATION REGIMES
Process fluid-lubricated bearings cover a wide range of operating conditions, often much broader than that for oil-lubricated metal bearings. Historically there have been three defined regimes: boundary, mixed, and hydrodynamic. Figure 3 includes a recent addition, dry friction. While the usual custom has been to treat these regimes as separate entities, every bearing goes through the boundary, mixed and hydrodynamic regimes every time the machine in which it is mounted starts and stops. The dry regime will come into play if the machine has not been operating for a long period of time. Tremendous amounts of damage can be done to processfluid lubricated bearings if the requirements of these different operating regimes are not considered, especially if the machine operating speed is continuously varied or it is frequently turned on and off.
Dry Friction Regime Dry friction is defined as the resistance to motion of one unlubricated body sliding over another. The classical equation applies,f = F/N, where f is the coefficient of friction, F the frictional force, and N the load. Most materials tend to obey the Amontons-Coulomb laws of dry friction:5 friction force is directly proportional to the load, friction force is independent of gross area of contact, friction force depends on nature of the sliding surfaces, and is independent of sliding velocity. Coefficient of friction values under dry friction conditions are on the order of 0.1 to 0.3, 1000 times the hydrodynamic regime values for the same bearing/shaft combination. Reference 8 covers the design of dry friction bearings.
Boundary Lubricated Regime This condition results when the surfaces are moving relatively at slow speed and/or with a heavy load. They are not in intimate contact but are separated by at least one molecular layer of lubricant. This condition occurs briefly during starting and stopping in process fluidlubricated bearings and sometimes at extremely slow speeds, well under one revolution per minute, when a ship at dockside has its propeller shaft being driven by the jacking motor. Heat removal is critical here since lubricating fluid flow is usually minimal under slow shaft rotation. Pumped additional water is usually required.
Mixed Film Regime This regime results when the fluid film is spotty or incomplete. Parts of the film are experiencing boundary lubrication, others are operating hydrodynamically. In process fluidlubricated bearings, it occurs as shaft speed is lowered below the hydrodynamic minimum Copyright © 1994 CRC Press, LLC
Volume III
545
friction point. Considerable tribological development work has been concerned with determining ways to lower the minimum friction hydrodynamic point, thereby reducing the extent of the mixed film lubricated regime.
Hydrodynamic Lubrication In this most desirable regime, the surfaces are completely separated by a fluid film. The fluid is hydrodynamically pumped between the surfaces by the relative motion between shaft and bearing. Due to the very low viscosity of water and many other process fluids, the minimum fluid film thickness values are extremely small and therefore require very expensive machining and installation techniques when conventional metal, carbon-graphite, or ceramic bearings are being used. Rubber bearing surfaces, on the other hand, are very forgiving and automatically adjust to give hydrodynamic lubrication so long as the designer controls the rubber properties and dimensions such as hardness, thickness, length, and width.
Other Comments A successful elastomer-bearing compound must have the ability to deflect and wear-in under mixed Film-operating conditions so that it can reach the hydrodynamic regime and give satisfactory life. The desired result is that the relatively soft elastomer bearing will lap and polish the shaft and itself sufficiently to shift the minimum friction point of the coefficient of friction vs. shaft velocity curve to a lower shaft velocity. Very smooth elastomer bearing and shaft surfaces eliminate or shorten this break-in period. Most process fluid lubricated bearings will only operate for short periods of time in the dry friction regime before damage occurs. Much larger clearances are required in the dry friction regime because of the greater generated heat. Good dry friction regime bearing materials such as polytetrafluoroethylene (PTFE) depend on formation of some sort of transfer film. The film cannot properly form in a water-lubricated application, resulting in a significant reduction in wear life and increase in friction.
PLASTO-ELASTOHYDRODYNAMIC LUBRICATION
Elastohydrodynamic lubrication is defined as the interaction between elastic stresses in a sliding body with the hydrodynamic film pressure in order to form very favorable lubricating conditions. It usually is a transient process involved with items such as gear teeth, ball bearing elements, or cam lobes that are rapidly moving in and out of contact. The simplest case, though, involves a metal shaft rotating in a fluid against a rubber block. The shaft will deform the rubber and, under the right conditions, develop hydrodynamic lubrication. Important factors in this example of elastohydrodynamic lubrication are the geometry of the rubber block, its elastic properties, the load, speed, and the lubricant viscosity. The creep and permanent set of the rubber in an elastomer bearing are also very important factors and probably a necessity for successful operation. Several investigators have reported the formation of a lubricant-trapping pocket in the elastomer polymer bearing face (Figure 5) during operation due to creep and set.9–11 Reducing the rubber thickness to no more than about 7.94 mm has been found to promote the formation of this pocket due to creep and set. The heavier the load or the larger the shaft, the thinner should be the rubber. Thickness values down to 0.79 mm have been successfully tested, but the coefficient of friction curve does increase. The optimum low friction rubber thickness for shafts in the 254 mm diameter range was 4.78 mm. For a 171 mm diameter shaft the optimum elastomer thickness was 3.18 mm. Wear rate also decreases with reduced rubber thickness, but not as dramatically as the friction decrease. Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
546
FIGURE 5. Carbon paper footprint of plasto-elastohydrodynamic pocket resulting from rubber creep and set.
This wear and friction reduction action of rubber bearings involving the formation of pockets is radically different than that of harder plastic and metallic bearing materials. The name “plasto-elastohydrodynamic lubrication” has been given to this type of bearing action. One distinction between plastics and elastomers in the polymer family is that plastics usually creep or distort under radial load with very little, if any, elastic recovery after the load is removed. Elastomers under load creep, but a large part of the creep is recoverable when the load is removed. The part of creep or deformation that does not elastically return is called “permanent set”. Nitrile elastomers tend to have substantial creep and permanent set compared with natural rubber compounds, as do most synthetic elastomers. Most water-and process fluid-lubricated bearings are made of nitrile elastomer compounds. Visual observation of the contact patch indicates that the minimum film thickness occurs at the trailing edge of the pocket. This agrees with the case for foil bearings and generally for all compliant surface bearings where calculated film thickness is at a minimum in the exit region. This is because there is a negative pressure gradient at the trailing edge and the pressure in the film drops to below the ambient value just before the gap increases beyond the trailing edge. Landed rubber bearings mat have run successfully for a number of years have, upon removal, shown a characteristic feathered or scalloped contact patch trailing edge caused by the cavitation/erosion resulting from the negative pressure. This failure mode is very slow. The bearing lands are able to operate for a long time before feathering becomes serious, and only a small percentage of harder elastomer bearings (usually over 80 Shore A) ever exhibit feathering. The cavitation/erosion is more likely to occur in cases where lubricant flow is at a minimum value.
WEAR
Suitable resistance to wear so that adequate life can be obtained is a fundamental requirement for a bearing/shaft system. There are a number of philosophies regarding the Copyright © 1994 CRC Press, LLC
Volume III
547
distribution of wear between bearing and shaft. One is to deliberately sacrifice the bearing to protect the expensive shaft. Some bearings used in process fluids are self-contained materials mat include lubricants such as graphite in their recipe. Even though the process fluid is available to develop hydrodynamic lubrication in a well-designed bearing, these bearings are designed (hopefully) to slowly consume themselves by wearing away to furnish the needed lubricant. Hydropower water-lubricated elastomer bearings have operated for 30 years or more in the presence of grit and sand. Their bearings deflect and adjust under load (rather than tilt) to form efficient fluid-lubricated bearing lands, while protecting the shaft.
Abrasive Wear Theory Theory of how wear starts and proceeds under conditions of particle-laden process fluid lubrication is the same for all materials and bearing designs. First, particles enter the bearing through clearance spaces or axial lubricant supply grooves and soon are pumped into the contact patch where the film thickness is at a minimum. The particles are pressed into the bearing material, where they are free to operate as miniature cutting tools, machining sharpedged grooves into the shaft surface. Next, the sharp-edged shaft grooves machine the bearing surfaces to complete the cycle. To eliminate wear, the chain of events has to be broken. One way to do this is to make the bearing material soft enough so that the pressure on the solid particle buries it completely so that there is no exposed cutting-tool to machine the shaft surface. One limitation of this approach is that only a certain number of solid particles can be embedded before the bearing layer is completely covered. Given the steady diet of particle-laden process fluid in many applications, economic life is usually too short for this design to be practical. Another method is used in the case of elastomer bearings. To interrupt the chain of events, the solid particle is elastically depressed or buried into the rubber.4 Because of the elasticity of the material, the particle is rolled through the contact patch to exit out the other side without damaging either the shaft or the rubber. One wear advantage of flat landed elastomer bearings is that much more fluid is pumped into the decreasing flow area than the bearing can accept. The excess fluid back-flows and proceeds toward the bearing ends to flush out the solid particles. Another way to interrupt the wear chain is to make both the bearing surface and the shaft coating out of very hard silicon carbide so that the particles are literally ground up. Such hard materials are brittle and difficult to make and machine, expensive, and require very careful installation and adjustment. Hand-lapping is often required after installation.
Wear Rate Total bearing and shaft wear is customarily measured by inserting feeler gages between the bearing and shaft to determine the increase in diametric clearance. This is used as an indication of when bearing replacement is required and possibly a new shaft. Most designers elect to use a bearing that wears out before the shaft; usually a number of bearing replacements are made for every replacement shaft required. For rigid plastic and metallic bearings, increase in clearance is an accurate indication of system wear. For elastomer bearings, however, change in clearance is not a good indication of system wear because of the initial elastic deflection of the rubber, followed by creep, and then permanent set to form the favorable plasto-elastohydrodynamic lubrication pocket. As much as 90% of the increase in diametric clearance during a clean water break-in period is due to creep and set of the elastomer without any loss of bearing material.
FRICTION
Friction performance of the bearing material is important because process fluid bearings often operate in the mixed fluid regime and occasionally under boundary lubrication. Film Copyright © 1994 CRC Press, LLC
548
CRC Handbook of Lubricants and Tribology
thickness is also at a minimum compared with oil values. These two conditions place critic emphasis on surface finish of bearing and shaft, ductility/elasticity properties of the bearing and, to a lesser degree, the shaft. Bearing friction torque is usually of little interest in itself because it is such a low percentage of total torque being transmitted by a shaft in a ship, pump, or process machine. But the heat generation rate depends on the factional energy loss. Most of this heat must be remove by the lubricating process fluid since heat conducted out through an elastomer bearing was is low because of its very low thermal conductivity, and the shaft assembly has too small a mass to be useful in absorbing the heat load. Another consideration is the upstream heat history of the process fluid before it reached the bearing. For example, in the case of water-lubricated elastomer bearings installed in the stem tubes of large ships, the cooling and lubricating water supplied to the bearing often has first lubricated and cooled the propeller shaft main seal. Typical break away (static coefficient of friction value for rubber bearings is 0.6. Reduced rubber thickness can drop it to 0.3.
Power Loss and Coefficient of Friction Power loss due to friction is given by the equation: P = (5.25 X 15-8)/WD(rpm) = kilowatts, where/is the coefficient of friction, W is the radial force (N), D is the diameter (mm) and rpm is the shaft speed (1/min). A coefficient of friction of 0.001 is a good assumption for the heaviest loaded land of a water-lubricated elastomer bearing operating in the 0.5 to 3 m/s shaft velocity range, 482 kPa projected pressure, running against a bronze shaft sleeve having a surface finish of about 0.24 µm and a water temperature of 21°C. Softer rubber generally makes a better bearing than does harder rubber when the operational regime falls in the mixed film regime for any length of time (Figure 6).” The more flexible surface asperities of the softer elastomer are believed to bend over more readily under load so that they will not penetrate the water film to the same degree as do those of a harder rubber compound. Note in Figure 7 the high friction in the mixed regime (0.2 to 1 m/s) shaft velocity and boundary lubricated regime (0.05 to 0.2 m/s) for the harder compound. For the softer elastomer in the mixed film region (0.05 to 0.51 m/s), the friction coefficient is much lower. Figure 7 shows how a thinner rubber bearing behaves when extrapolated to a longer length.” Also plotted are published curves for two traditional process fluid-bearing materials, lignum, vitas, wood and & laminated fabric-phenolic, not as widely used now because of their poor wear quality in grit-laden water. The two harder materials have much higher shaft speeds for their minimum friction transition point. Other studies have shown similar superior lowspeed performance of thin rubber bearings at slower shaft speeds with water lubrication. As discussed earlier, rubber bearing controlling design factors are the rubber type and hardness, thickness, land width, and surface finish. The limitations of elastomer bearings are related to temperature and load.
Theoretical Coefficient of Friction For comparison purposes, it is possible to calculate a theoretical coefficient of friction curve for a single land as in Figure 6. The equation used is developed from that used for tilting shoe bearings,3 which are geometrically similar to flat land rubber bearings. Assuming zero end leakage, the following formula results for water-lubricated conditions:
Copyright © 1994 CRC Press, LLC
Volume III
549
FIGURE 6. Coefficient of friction comparison, softer vs. harder rubber.
where µ, is the fluid viscosity (Pa s), L is the axial length of the bearing (m), V is the shaft velocity (m/s), and F is the radial force (N). This theoretical hydrodynamic coefficient of friction gives the designer a quick check on bearing efficiency. Load per land is determined by trigonometric analysis.
OTHER BEARING MATERIALS
Many nonmetallic bearing materials have been used for water lubrication. Lignum vitae wood was the very first of these (1850s) and is still used. Due to its limited availability, seasoned kiln-dried hard maple is a typical replacement.12 The wood is impregnated with a blend of fluid lubricants before being machined into various required shapes. The impregnated lubricant makes up to 40% of the total final weight of the wood bearing. Exuded lubricant at the bearing/shaft interface allows for easy initial start-up. As a water film develops, the fluid lubricants are re-absorbed by the wood. Upper temperature limit is 83°C. Load rating is given by a pressure-velocity upper limit of 25,217 kPa mpm. One of the earliest plastics, phenolic, is still used in some water-lubricated applications.13 Various textiles used to reinforce the phenolic include cotton fibers and cloth, glass fibers and cloth, and still occasionally asbestos fibers, paper, and cloth. In recent years, most asbestos has Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
550
FIGURE 7. Comparison of full-length bearings.
been replaced because of its carcinogenic nature. Lubricating fillers such as graphite and molybdenum disulfide are used to give wet/dry operational capabilities. Phenolic bearings can operate as high as 31,522 kPa mpm. Phenolic bearings do not have as low a hydrodynamic transition point as do elastomer bearings (Figure 7) and are manufactured with round bores as well as with flat lands.
Copyright © 1994 CRC Press, LLC
Volume III
551
REFERENCES 1. Booser, E. R., Ed., CRC Handbook of Lubrication (Theory and Practice of Tribology), Vol. II, Theory and Design, CRC Press, Boca Raton, Florida, 1984. 2. Orndorff, R. L., Jr., Polymer bearing and wear surfaces, Part A, in Plastic Products Design Handbook, Miller, E., Ed., Marcel Dekker, 1981, pp. 252–253, 260–262. 3. Military Standardization Handbook: Rubber MIL-HDBK-149B, U.S. Department of Defense, February 1, 1984, pp. 13, 15, 59. 4. Application Guide for Engineers, BFGoodrich Cutless® Brand Bearings, BFGoodrich Aerospace, Wilmington, N.C., 1988, pp. 2, 9. 5. Fuller, D. D., Theory and Practice of Lubrication for Engineers, 2nd ed., John Wiley & Sons, New York, 1984, pp. 449, 549. 6. Gardner, W. W., Hydrodynamic Oil Film Bearings, Fundamentals, Limits and Applications, Waukesha Bearings Corporation, Waukesha, Wisconsin. 7. Thordon Engineering Manual, Thomson-Gordon, Burlington, Ontario, 28, 1991. 8. Waters, K. A., Design and Material Selection for Dry Rubbing Bearings, ESDU International, London, 1987. 9. Abramovitz, S., Fluid-film bearings, fundamentals and design criteria and pitfalls, in Proc. 6th Turbomachinery Symp., Texas A&M University, Houston, 1977, 193. 10. Pollock, M., Water Lubricated Stem Tube Bearings, Recent Developments, Thomson-Gordon, Burlington, Ontario, 1985, 7. 11. Orndorff, R. L., Jr., Water-Lubricated Rubber Bearings, History and New Developments, Naval Eng. J., Nov. 1985, 97, 7, 39. 12. Hughes, T., For bearings, consider Nature’s plastic hardwood, Power Transm. Des., November 1989. 13. Smith, W. V. and Schneider, L. G., Lubrication in a sea-water environment, Naval Eng. J., October 1963,841.
Copyright © 1994 CRC Press, LLC
GAS BEARINGS
Michael M. Khonsari, Lee A. Matsch, and Wilbur Shapiro
INTRODUCTION
Gas bearings are successfully utilized in a wide variety of applications from small instruments to large turbomachinery. There are major applications in the computer industry, particularly for magnetic tapes and flying heads in read/write devices. They are also used extensively on aircraft air cycle machines for providing cabin pressurization and climate control. In general, gases offer fairly useful lubrication from cryogenic to elevated temperatures of 1600°C.1.2 Gas bearings can be very attractive for high-speed applications by virtue of low gas viscosity and its relative intensitivity to temperature. Whereas the viscosity of the conventional oils drops exponentially with increase in temperature, viscosity of gases increases. In addition, use of the ambient gas medium as a lubricant can ease or eliminate the need for lubricant seals. Air-lubricated journal bearing dental drills operate smoothly at speeds over 600,000 rpm.2 Fuller puts the practical limit of 700,000 rpm for gas bearings with no cooling requirements. Gas bearings are not free of disadvantages. Because gas viscosities are orders of magnitudes lower than oil viscosities, the load-carrying capacities are inherently lower. Nonetheless, there are reports describing gas bearings that support a massive spectrometer under a load of 236 tons.3 Gas bearings are complicated by compressibility effects, which can aggravate system stability and must be carefully analyzed during the design phase. There are basically two broad categories of gas bearings: (1) self-acting bearings where relative surface motion generates hydrodynamic pressure; and (2) externally pressurized bearings where pressurized gas is supplied from an external source. Some typical examples of self-acting thrust bearings are shown in Figure 1. The externally pressurized bearings can be of either me thrust or journal type. Many factors must be considered in bearing selection. For self-acting bearings, the surfaces are susceptible to wear as they come into direct contact during stopping and starting operations. Another potential problem with self-acting journal bearings is self-excited whirl instability. The externally pressurized bearings, on the other hand, require more elaborate supply arrangements with continuous gas feed through holes in the bearing surface. With pressurized gas turned on, relatively high loads can be sustained in the absence of relative surface motion at the expense of gas consumption. Externally pressurized bearings, however, can be susceptible to pneumatic instabilities. This chapter will concentrate on self-acting bearings. Externally pressurized gas bearings are covered by Tang and Gross4 and Wilcock.5 A thorough discussion of this subject is given by Gross and co-authors.3 General design procedure for hybrid gas journal bearings that utilize a combination of hydrostatic and hydrodynamic action is described in Reference 2. Gross and his contributors3 discuss a class of compliant surface bearings where the surface deformation plays a significant role. These include bearings that utilize coating with low modulus of elasticity and also foil bearings where one surface is made of flexible material under tension. Recent contribution in the field of gas lubricated foil bearings are given by Heshmat and co-workers.6–9
Copyright © 1994 CRC Press, LLC
553
CRC Handbook of Lubricants and Tribology
554
FIGURE 1. Various thrust bearing configurations.
REYNOLDS EQUATION
The Reynolds equation characterizes the pressure distribution in gas-lubricated bearings. Assuming that the working fluid is an ideal isothermal gas, this equation takes on the following form3,10 where H, P, and T are dimensionless film thickness, pressure, and time, respectively. Parameter Λ represents the bearing number or compressibility number defined for slider bearings as:
where U is the velocity, B is the length in the direction of motion, and hm is the minimum film thickness. The parameters Pa and µa are the ambient pressure and viscosity, respectively
where r is the shaft radius, c is the radial clearance, and σ is the rotational speed. The following parameter is called the squeeze number
The above form of the Reynolds equation is derived assuming that continuum holds; i.e., the molecular mean free path of the gas is negligibly small compared to the film gap. For cases where this assumption ceases to be valid (Knudsen number greater than 0.01), the result is slip flow of the boundary surfaces. As a result, a reduction in the pressure can be expected, akin to a reduction in viscosity.11,12 Consider for example a minimum film thickness of 20 µ Copyright © 1994 CRC Press, LLC
Volume III
555
FIGURE 2. Effect of film-thickness ratio and bearing number on load for plane slider bearings. (From Gross, W., Ed., Fluid Film Lubrication, Wiley-Interscience, New York, 1980.)
in (0.5 µm) in an air-lubricated slider bearing. Given that the molecular free path of air at atmospheric condition is 2.52 µ in (0.064 µm), the Knudsen number becomes Kn = 0.064/0.54 = 0.119, which signifies the importance of slip flow. Referring to Equation 1, the following limiting cases can be deduced:
1. 2.
When A → 0, and the steady-state performance is sought, the effect of compressibility diminishes and the solution to Reynolds equation reduces to the incompressible case. Subscript zero will be used to designate this limiting case. When A is large and in the limit Λ → ∞, Reynolds equation provides a limiting solution for compressible flow. The subscript ∞ will be used for this case.
The actual pressure (and therefore the load-carrying capacity) lies between these useful limiting first order approximations (see Figure 2). Typical closed-form solutions for specific geometries are presented in this chapter.
PROPERTY VALUES
Table 1 shows the viscosity of several gases and their variation with temperature. Volume II of this Handbook (pp. 291–300) gives a variety of other gas properties of interest. The compressibility number can experience a large variation, depending on the operating condition. To illustrate, consider the following journal bearing example:
For the top speed and maximum temperature, we have
Copyright © 1994 CRC Press, LLC
556
CRC Handbook of Lubricants and Tribology
Copyright © 1994 CRC Press, LLC
Volume III
557
For the minimum speed and temperature A = 0.22 and this can be assumed to be in the incompressible range.
CLOSED-FORM SOLUTIONS
Solutions of the Reynolds equation in which the side leakage is neglected yield fairly realistic results when bearing length in the axial direction is much greater than that in the sliding direction. Two classes of these solutions follow.
Infinitely Long Taper-Step Bearings Theoretical load-carrying capacity of an infinitely long slider bearing is predicted to be optimum if the shape is designed to form a combination tapered and step bearing as shown in Table 2.3 Performance parameters, including the load and friction force, can be evaluated from the expressions provided in the table. Performance results for simple geometric configurations such as taper (plane wedge), taper-flat, and step bearings are essentially simplified versions of the equations in Table 2. For some geometric shapes, such as Rayleigh step bearings, the theoretical load-carrying capacity is considerably reduced by relatively large leakage flow out of the bearing sides resulting from the maximum pressure occurring at the step. To improve the problem with excessive leakage, one can machine a curve surface in the place of the step3 or apply shrouds along the sides of the bearing to form a pocket as shown in Figure 1. Infinitely Long Journal Bearings For a full 360° journal bearing, similar limiting expressions are summarized in Table 3. Parameters W, and W2 are components of the total load W which is General expressions for evaluating the performance of partial bearings are given in Reference 3. The remainder of this chapter provides finite length-bearing solutions of the Reynolds equation for a variety of thrust and journal bearing designs.
RECTANGULAR SLIDER, SECTOR-PAD, AND TILTING-PAD THRUST BEARINGS
Load-carrying capacity as a function of the bearing number for a square-shape slider is presented in Figure 3. For comparison purposes, various solutions for different tilt angles are shown for slenderness ratios of L‘ = L/B = 1 and ∞. This figure is useful for assessment of the accuracy of the infinitely long solutions as well as the limiting solution when A →∞ A simple approach for analyzing a sector-pad thrust bearing is to replace the geometry by a rectangular pad whose width is equal to the mean value of the sector circumference and its length equal to the difference between the outer radius r0 and inner radius r0. This, however, distorts the sector shape geometry. Etsion and Fleming13 provide accurate solutions for flat sector-shape thrust bearings over a wide range of the compressibility numbers with Reynolds equation written in polar coordinates. When a pivot (assumed frictionless) is used, the film thickness will depend both on the pitch (tilting about a radial line) and roll (tilting about a tangent line). If the film thickness is decreased substantially, the perfect plane pivoted-load bearing surfaces become parallel, thereby leading to loss in load-carrying capacity.3 For this reason, and because of difficulties with starting of perfectly plane-pivoted bearings, a crown is commonly machined into the surface. Performance curves and appropriate design procedures for sector thrust bearings with crowned surfaces (parabolic, cylindrical, and spherical) are covered by Shapiro and Colsher.14 Copyright © 1994 CRC Press, LLC
558
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
Volume III
559
The dimensionless Reynolds equation in polar coordinates takes on the form given below14 where Q = P2 and T is the dimensionless time = ω1t/2. Parameter ω1, is a reference speed = 2T/t. The following design information is extracted from Shapiro and Colsher14 for a tiltingpad bearing with cylindrical crowns as shown in Figure 4. The parameters of interest for design are: position of pivot, optimum height, and bearing performance prediction. The parameters that characterize a crown are the crown height, δ, and clearance of aligned pad measured from the peak point of the crown, c. Location of the pivot is defined by the radius rpy and the angle γ measured from the pad leading edge. The angular extent of the pad by θP is assumed to be 45° (0.7854 rad). The pitch is represented by αr, = α and the roll by αθ When the pad is approximated as a rectangular shape, the pivot location is normally taken to be along the mean radius. Proper formulation of the problem shows that the actual position Copyright © 1994 CRC Press, LLC
560
CRC Handbook of Lubricants and Tribology
FIGURE 3. Effect of bearing number on slider bearing load. (From Gross, W., Ed., Fluid Film Lubrication, Wiley-Interscience, New York, 1980.)
is located only slightly outside the mean radius for a zero roll angle. Although it is a function of δ/c and also varies with Λ, a value of rpvro = 0.76 represents a typical pivot position.14 For representative compressibility numbers of Λ = 25 (incompressible) and Λ = 250 (compressible), this radial position of the pivot is about 1 to 2% beyond the mean radius and is rather insensitive to the crown height, inclination angle, and compressibility number.14 When the dimensionless minimum film thickness H,min = hmin/c is very small, the load coefficient cL = W/Par20 does indeed become sensitive to the pivot position. Figures 5 and 6 illustrate the importance of the dimensionless crown height, δ/C, for Λ = 25 and Λ = 250, respectively. An optimum load can be achieved by letting δ/c = 1. This results in a significant improvement over a flat shape (δ/c = 0). Figures 7 and 8 present field plots as Copyright © 1994 CRC Press, LLC
Volume III
561
FIGURE 4. Nomenclature of a tilting-pad thrust bearing. (From Shapiro, W. and Colsher, R., Rep. No. IA0249–30, Franklin Institute Research Laboratories, Philadelphia, 1968.)
a function of the pivot position for the representative bearing numbers. Figures 9 and 10 present the variation of the load coefficient and friction as a function of Λ, respectively. Additional field plots for Λ = 2500 are also available.14
Design Procedure for Tilting-Pad Thrust Bearings As an example, consider a tilting-pad thrust bearing with the following specifications: ri = 2 in. (5.08 x 10-2 m); r0 = 4 in. (1.02 X 10-1 m); θP = 45°; ω = 20,000 rpm (2094 rps); Pa = 15 psi (103 KPa); total load WT = 450 lb (2000 N); n = 6 pads; Wp = 75 lb per pad (334 N).
Design for Optimum Load Capacity 1.
Compute the load coefficient based on the maximum load per pad.
2. 3. 4.
Since crown parameters C and 8 are not known a priori, the bearing number cannot be computed directly. Therefore, we shall begin by selecting as a representative bearing number Λ = 250. Use Figure 9 to evaluate Hmin = 0.732. Compute C using the definition of the bearing number
5. 6.
For optimum load-carrying capacity, choose δ = C = 0.414 X 10-3 in. Compute dimensional minimum film thickness.
7.
It is crucial that the adequacy of this minimum film thickness be checked by comparing it to the surface roughness of bearing surfaces. Compute the friction moment factor from Figure 10.
Copyright © 1994 CRC Press, LLC
562
Copyright © 1994 CRC Press, LLC
FIGURE 6. Performance of tilting-pad thrust bearing as a function of crown height, A = 250. (From Shapiro, W. and Closher, R., Rep. No. I-A0249–30, Franklin Institute Research Laboratories, Philadelphia, 1968.)
CRC Handbook of Lubricants and Tribology
FIGURE 5 Performance of tilting-pad thrust bearing as a function of crown height, A =25. (From Shapiro, W, and Colsher, R., Rep. No. I-A0249–30, Franklin Institute Research Laboratories, Philadelphia, 1968.)
Volume III
563
FIGURE 7. Tilting-pad thrust bearing load coefficient and minimum film-thickness as a function of pivot position, constant crown height, varying film-thickness ( , cylindrical profile.) (From Shapiro, W. and Colsher, R., Rep. No. I A0249–30, Franklin Institute Research Laboratories, Philadelphia, 1968.)
8.
Determine pivot position. Use Figure 6 with δ/C = 1.0, and read γ/θP = 0.6. Since θP = 45°, γ= 27°. Take the radius of the pivot as 3.05 in. (1.5% greater than the 3 in. mean radius). The influence of bearing compressibility number on the load
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
564
FIGURE 8. Tilting pad-thrust bearing load coefficient and minimum film-thickness as a function of pivot position, constant crown height, varying film-thickness ( rad, cylindrical profile.) (From Shapiro, W. and Colsher, R., Rep. No. I-A0249–30, Franklin Institute Research Laboratories, Philadelphia, 1968.)
coefficient for θP = 45° and δ/C = 1.0 (optimum loading) is shown in Figure 11. Also shown are results based on the incompressible fluid assumption, which closely approximate compressible results up to roughly Λ = 50.
FULL JOURNAL BEARINGS
Steady-State Performance Raimondi15 first developed design charts for full journal bearings of finite length with various L/D ratios based on numerical solutions of the Reynolds equation. Results for L/D = 1 are given in Figure 12. The load and operating conditions enable one to compute Λ: = µUr,/c2Pa and P/Pa = W/2rLPa. Figure 12a then gives operating eccentricity ε. Entering Copyright © 1994 CRC Press, LLC
Volume III
565
FIGURE 9. Effect of varying Λ on friction load coefficient. (From Shapiro, W. and Colsher, R., Rep. No. IA0249–30, Franklin Institute Research Laboratories, Philadelphia, 1968.)
Figures 12b and 12c with the compressibility and ε then gives friction force (hence, power loss) and attitude angle, respectively. Alternatively, in a particular design, the minimum film thickness may be specified. Eccentricity ratio can then be directly computed from hmin = C(l - ε), and load-carrying capacity predicted from Figure 12a. Note in Figure 12 that nonlinearity sets in when Λ/6 ≥ 1; that is, the incompressible solutions deviate significantly from those of compressible when Λ ≥ 6. In contrast, in design curves of Shapiro and Colsher14 for tilting-pad thrust bearings, a limit of the compressibility number of FIGURE 9. Effect of varying Λ on friction load coefficient. (From Shapiro, W. and Colsher, R., Rep. No. I-A0249–30, Franklin Institute Research Laboratories, Philadelphia, 1968.) Λ ≥ 50 was recommended.
Angular Stiffness and Misalignment Torque For a system where a series of gas-lubricated journal bearings are used, one may have to take into account the effect of possible misalignment and angular stiffness. Figure 13 may be used to estimate the “moment-carrying capacity” of a bearing.16 For a given ≥, Figure 13 provides a relationship between the misalignment torque Tm and the misalignment angle β. Once a maximum misalignment angle is estimated, one can calculate the maximum allowable moment. Results shown in Figure 13 are based on a perturbation method and are valid for small eccentricity ratios (up to 0.5). Extension to higher ε values are provided by Rice,17 who solved the governing equations by numerical schemes. Whirl Instability Plain journal bearings are susceptible to whirl instability which therefore should be considered at the design stage. The threshold of instability depends on the eccentricity ratio and L/D ratio. When the compressibility number is known, Figure 14 can be used to determine the critical speed above which instability is likely to set in.18,19 The higher the eccentricity ratio, the higher the threshold of instability. The stability of plain journal bearings may be improved by incorporating multiple sections or lobed-shaped bore geometries (see Figure 15).
Copyright © 1994 CRC Press, LLC
566
CRC Handbook of Lubricants and Tribology
FIGURE 10. Effect of varying A on moment. (From Shapiro, W. and Colsher, R., Rep No. I-A0249–30, Franklin Institute Research Laboratories, Philadelphia, 1968.)
FIGURE 11. Influence of A on load coefficient and comparison of the incompressible and compressible theory. (From Shapiro, W. and Colsher, R., Rep No. I-A0249–30, Franklin Institute Research Laboratories, Philadelphia, 1968).
Copyright © 1994 CRC Press, LLC
Volume III
567
FIGURE 12. Plain journal bearing design charts for L/D = 1.0. a = load ratio vs. compressibility number. b = friction force vs. compressibility number, c = attitude angle vs. compressibility number. (From Raimondi, A., Trans. ASLE, 4, 131, 1961.)
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
568
FIGURE 13. Angular stiffness of misaligned 360° journal bearing. (From Ausman, J., in Proc. 1st Int. Symp. Gas Lubricated Bearings, 1959, 161.)
FIGURE 14. Stability chart for plain journal bearing L/D = 1.18.19
TILTING-PAD JOURNAL BEARINGS
Tilting-pad journal bearings have excellent performance from the point of view of stability and alignment. Extensive design curves are provided by Shapiro and Colsher.20 In this chapter, their design charts will be presented for three-pad journal bearings for the load directed toward one of the pads. Figure 16 shows the geometry of a three-pivoted-pad journal bearing with pivot film thickness values of hP1, hP2 and hP3. Parameters α and β denote the pad angle and angle from the load vector to pivot, respectively. There are two “clearances” involved: one is the Copyright © 1994 CRC Press, LLC
Volume III
FIGURE 15. Multiple section and elliptical bearings with improved stability characteristics (From Gross, W., Ed., Fluid Film Lubrication, Wiley-Interscience, New York, 1980.)
569
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
570
FIGURE 16. Nomenclature of a pivoted-pad journal bearing. (From Shapiro, W. and Colsher, R., Rep. No. AFFOL-TR-70–99, Air Force Flight Dynamics Laboratory, Wright-Patterson Air Force Base, OH, 1970.)
machine clearance C, and the other is the pivot clearance, C’. The ratio C’/C, the so-called preload factor, plays an important role on the bearing performance and stability. Let us consider a pivoted-pad journal bearing with all pads rigidly supported and with load directed toward a pad. Figure 17 shows the variation of the bearing coefficient, CLT = W/PaRL, vs. the bearing eccentricity ratio, E’ = e’/C’ for R/D = 0.5 and A = 1.5. For a given CLT and C’/C, one can predict the eccentricity ratio, e’, and the dimensionless film thickness at the pivot points. As shown HP,1 = HP,2 for all cases. Steady-State Performance The design procedure will be illustrated through the following example: R = 0.25 in. (6.35 x 10-3 m); Pa = 15 psia (103 KPa); N = 200,000 rpm; T = 703°F (373°C); W = 2.5 lb (11 N); µ. = 1.01 x 10-9 reyns (nitrogen gas). 1.
Bearing length, while not specified, plays an important role in load-carrying capacity and also start-up and roll stability. As a guide, W/DL ≤ 10 psi is recommended for maximum loading and L/D < 1.5 to prevent roll instability.
Copyright © 1994 CRC Press, LLC
Volume III
571
FIGURE 17. Bearing load coefficient vs. eccentricity ratio for a pivoted pad bearing. (From Shapiro, W. and Colsher, R., Rep. No. AFFOL-TR-70–99, Air Force Flight Dynamics Laboratory, Wright-Patterson Air Force Base, OH, 1970.)
2.
Choose a compressibility number A = 1.5 and compute “machined in” clearance C
3.
Compute load coefficient
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubricants and Tribology
572
FIGURE 18. Pad minimum clearance as a function of pivot clearance. (From Shapiro, W. and Colsher, R., Rep. No. AFFOL-TR-70–99, Air Force Flight Dynamics Laboratory, Wright-Patterson Air Force Base, OH, 1970.)
4.
Use Figure 17 with this value of CLT and tabulate the results for eccentricity and pivot film thickness for selected values of the pre-load factors C’/C. a. Evaluate bearing eccentricity ratio, ε’. move horizontally to intersect a selected C’/C curve and read ε’ at the intersection point. for example, for c’/c = 0.3, ε’ = 0.575 (extrapolation may be needed). b. Evaluate pivot film thicknesses. At the established eccentricity and appropriate c’/c curves (= 0.3 for this example), determine the dimensionless film thickness values c.
Evaluate the dimensional film thicknesses by multiplying hp1 by c; i.e., hps = c x hps for each pad. 5. Use Figure 18 to determine the minimum film thickness for each pad. In this example, the pad #3 has the smallest minimum film thickness @ Hm s= 0.1. Dimensionalizing the smallest minimum film thickness, 6.
Determine the pad coefficient of friction F = F,/P2CL for each pad by utilizing Figure 19. Solve for the friction force on each pad and sum them up to get the total
Copyright © 1994 CRC Press, LLC
Volume III
573
FIGURE 19. Pad friction coefficient as a function of pivot clearance. (From Shapiro, W. and Colsher, R., Rep. No. AFFOL-TR-70–99, Air Force Flight Dynamics Laboratory, Wright-Patterson Air Force Base, OH, 1970.)
7.
Compute frictional power loss in watts
8.
Repeats steps 4 through 7 for various preload factors, tabulate the results, and plot the dimensional minimum film thickness, hm, vs. the pre-load factor C’/C as well as the power loss, Pf, as a function of the pre-load factor to decide on the optimum value of the pre-load factor. A small C’/C value (tight film thickness) results in a large power loss. A large C’/C value, on the other hand, may cause vibration problems due to excessive pad loading and is susceptible to pad lock-up at the leading edge. Therefore, it is recommended to restrict the pre-load factor to 0.3 ≤ C’/C ≤ 0.8. For the example problem, minimum film thickness approaches a limiting maximum value approximately at C’/C = 0.5, which also gives reasonable power loss.
Copyright © 1994 CRC Press, LLC
574
CRC Handbook of Lubricants and Tribology
FIGURE 20. Dynamic stability for pivoted pad bearing A = 1.5. (From Shapiro, W. and Colsher, R., Rep. No. AFFOL-TR-70–99, Air Force Flight Dynamics Laboratory, Wright-Patterson Air Force Base, OH, 1970.)
For complete design charts with different compressibility numbers and for situations where the load vector is between the two pads, the designer is referred to Shapiro and Colsher.20,21 For certain applications, one of the pads may be spring-loaded to accommodate thermal expansion of the shaft. Design curves for such bearings are covered by Shapiro and Colsher.20
Dynamic Stability Considerations Once the steady-state analysis is complete, dynamic stability must be checked using Figure 20 with two parameters: the shoe pitch inertia parameter, Ip, and the dimensionless shaft mass parameter, Ms. Consider the following example, where the steady-state performance was evaluated in the previous section, and assuming that the shaft is made of steel, The pitch moment of inertia of the pad, assuming that it is made of titanium, is
The operating point corresponding to this bearing is in the stable region with either pad material. Copyright © 1994 CRC Press, LLC
Volume III
575
REFERENCES 1. Fuller, D. D., Theory and Practice of Lubrication for Engineers, Wiley, New York, 1980. 2. Grassam, N. and Powell, J., Eds., Gas Lubricated Bearings, Butterworths, London, 1964. 3. Gross, W., Ed., Fluid Film Lubrication, Wiley-Interscience, New York, 1980. 4. Tang, I. C. and Gross, W., Analysis and design of externally pressurized gas bearings, ASLE Trans., 5, 261, 1962. 5. Wilcock, D., Ed., Gas Bearing Design Manual, Mechanical Technology, Inc., Latham, NY, 1972. 6. Heshmat, H. and Shapiro, W., Advanced development of air-lubricated thrust bearings, ASLE Lubr. Eng., 40(1), 21, 1984. 7. Heshmat, H., Walowit, J., and Pinkus, O., Analysis of gas-lubricated compliant thrust bearings, J. Lubr. Technol. Trans. ASME, 105(4), 638, 1983. 8. Heshmat, H., Walowit, J., and Pinkus, O., Analysis of gas-lubricated compliant journal bearings, J. Lubr. Technol. Trans. ASME, 105(4), 647, 1983. 9. Ku, C.-P. R. and Heshmat, H., Compliant foil bearing structural stiffness analysis. I. Theoretical model including strip and variable bump foil geometry, ASME J. Tribal., 114(2), 394, 1992. 10. Hamrock, B. J., Fundamentals of Fluid Film Lubrication, NASA Ref. Pub. No. 1255, 1991. 11. Sereny, A. and Castelli, V., Experimental investigation of slider gas bearings with ultra-thin films. Trans. ASME J. Lubr. Technol., 101, 510, 1979. 12. Eshel, A., On controlling the film thickness in self-acting foil bearings, ASME Trans. J., Lubr. Technol., 92, 630, 1970. 13. Etsion, I. and Fleming, D., An accurate solution of the gas lubricated, flat sector thrust bearing, Trans. ASME, J. Lubr. Technol., 99, 82, 1977. 14. Shapiro, W. and Colsher, R., Analysis and performance of the gas-lubricated tilting pad thrust bearing, Rep. #I-A2049–30, Franklin Institute Research Laboratories, Philadelphia, 1968. 15. Raimondi, A. A., A numerical solution for the gas lubricated full journal bearing of finite length, Trans. ASLE, 4, 131, 1961. 16. Ausman, J., Theory and design of self-acting gas-lubricated journal bearings including misalignment effects, Proc. First Int. Symp. Gas Lubricated Bearings, Oct. 1959, Report ONR, ACR-49, U.S. Govt. Printing Office, Washington, D.C., 161. 17. Rice, J., Misalignment torque of hydrodynamic gas-lubricated journal bearings, Trans. ASME, J. of Basic Eng., 87, 193, 1965. 18. Cheng, H. and Pan, C, Stability analysis of gas-lubricated self-acting plain cylindrical bearing of finite length, Trans. ASME, J. Basic Eng., 87, 185, 1965. 19. O’Connor, J. J. and Boyd, J., Standard Handbook of Lubrication Engineering, McGraw Hill, New York, 1968. 20. Shapiro, W. and Colsher, R., Analysis and design of gas-lubricated tilting pad journal bearings for miniature cryogenic turbomachinery Rep. #AFFOL-TR-70–99, Air Force Flight Dynamics Laboratory, Wright-Patterson Air Force Base, OH, 1970. 21. Shapiro, W. and Colsher, R., Computer-aided design of gas-lubricated, tilting-pad, journal bearings, ComputerAided Design of Bearings and Seals, American Society of Mechanical Engineers, New York, 1976, 67.
Copyright © 1994 CRC Press, LLC
MAGNETIC BEARINGS
Paul E. Allaire Eric H. Maslen Robert R. Humphris Carl R. Knospe and David W. Lewis
INTRODUCTION
Magnetic bearings are a rather new concept in bearing technology. They take radial loads or thrust loads by utilizing a magnetic field to support the shaft rather than a mechanical force as in fluid film or rolling element bearings. Magnetic bearings have several advantages and disadvantages as summarized in Table 1. Two primary advantages of magnetic bearings are the very low power consumption and very long life. Because there is no contact between the rotor and stator, there is no wear. Where fluid film bearings have high friction losses due to the oil shearing effects, magnetic bearing losses are due to some low level air drag, eddy currents, and hysteresis. Also, the losses associated with oil pumps, filters, and piping are much greater than the power associated with controls and power amplifiers. Overall, magnetic bearings normally have an order of magnitude lower power consumption than oil film bearings. Magnetic bearings commonly have lower power consumption than rolling element bearings. Also, rolling element bearings have finite life and DN limits. Because of the noncontact nature of magnetic bearings, they have much longer expected life and higher DN limitations. Other advantages of magnetic bearings are related to reduced dependence on environmental conditions. Magnetic bearings do not require oil lubrication so they are well suited to applications such as canned pumps, turbomolecular vacuum pumps, turboexpanders, and centrifuges where oil cannot be employed. They can operate at much higher temperatures or at much lower temperatures than oil-lubricated bearings. A recent study of aircraft gas turbine engines indicates that the elimination of the oil supply and associated components with magnetic bearings could reduce the engine weight by approximately one fourth. The first magnetic bearing in the U.S. to support a high speed centrifuge rotor axially was constructed at the University of Virginia in 1935 to 1939.1–2 Other magnetic bearings were constructed in Germany at about the same time. As reliable electronic components became available at a reasonable cost, magnetic bearing applications were reported by academic researchers3 and industrial firms.4 Applications have extended to compressors, pumps, turboexpanders, gas turbines, turbomolecular vacuum pumps, X-ray tubes, and many others.5 A magnetic bearing system consists of four basic components: (1) magnetic actuator, (2) electronic control, (3) power amplifier, and (4) shaft position sensor. In many ways, magnetic bearing components resemble electric motors with the basic magnetic actuator being constructed of soft ferromagnetic material electromagnetically activated by a coil of wire. As shown in Figure 1, the stator component is normally composed of horseshoe-shaped magnets and the rotor component is a piece of magnetic material to complete the magnetic path. An air gap separates the nonmoving and moving parts of the bearing. To keep the rotor from contacting the stator, a position sensor signal is used as input for an electronic control circuit to adjust the coil currents. If the rotor is too close to the stator, the coil current is reduced. Alternatively, if the rotor is too far away from the stator, the coil current is increased. BASIC ACTUATOR THEORY
Magnetic bearing design is performed in two phases: (1) the initial design magnetic circuit model and (2) finite element detailed analysis. Only the first will be discussed in this chapter
0-8493-3903-0/94/$0.00 + $0.50 Copyright © 1994 CRC Press, LLC © 1994 by CRC Press, Inc.
577
578
CRC Handbook of Lubrication and Tribology
to illustrate the calculations used.6,7 The theoretical model of the thrust bearing used in this chapter employs a simple magnetic circuit model of the thrust or radial bearing to illustrate the principles involved in magnetic bearing design. It is essentially a one-dimensional flux model taken along the centerline of the magnetic material and in the air gaps. This is then verified with a magnetic finite element analysis if necessary. Finally, test measurements are made on the actual bearing to verify the load capacity, etc. Several assumptions are made in the following: (1) flux levels are always below saturation level, (2) shaft motions are small compared to the steady-state air gap dimension, (3) the flux distribution is relatively uniform in stator cross sections, and (4) leakage is small. In actual magnetic bearing analysis, some or all of the above assumptions may be violated by a particular bearing design. This does not mean that the bearing will not operate, but it does mean that a more complex finite element analysis must be carried out. The magnetic part of the circuit, illustrated in Figure 1, is constructed of ordinary magnetic material such as silicon steel or higher saturation level magnetic materials such as Vanadium Permendur. The air gap has thickness g and area A,. Magnetic flux is produced in each horseshoe-shaped section of the actuator by a coil of N turns of wire with a current i flowing through it. A power amplifier produces the desired current in the coil. The flux path of length L goes through the horseshoe, through the air gaps at the end of each pole face of the horseshoe magnet, and through the rotor. Ampere’s circuital law indicates that the magnetic field intensity H induced by N wires carrying current i wrapped around a closed magnetic path of length L is given by This assumes that the magnetic field intensity direction is parallel to the magnetic path and that the current in the wire is perpendicular to the magnetic path. The quantity Ni is called the magnetomotive force (MMF). The magnetic flux φ in the circuit equals the flux density B times the pole face area Ag, which is also the area of one air gap in the magnetic circuit. In magnetic circuits, most of the magnetic resistance (called the reluctance) resides in the air gaps. Air and other nonferrous materials have nearly the same magnetic properties as free Copyright © 1994 CRC Press, LLC
579
Volume III
FIGURE 1. Basic double-acting magnetic actuator geometry.
space. The flux density in such materials is related to the magnetic field intensity by the linear relation (valid for many nonmagnetic materials) where the permeability of free space (air) is Magnetic bearings are normally constructed of a ferrous magnetic material and the air gaps made as small as practical to minimize the required magnetomotive force. Nearly always, the magnetic flux in magnetic bearings is determined by the gap and the reluctance of the magnet iron can be neglected compared to that in the air gap. The typical ferrous magnetic material employed in a magnetic actuator has a magnetization curve, plotted as B vs. H, as illustrated in Figure 2. The B-H curve is roughly linear for much of the range of B. The slope of this curve in the linear range is called the permeability of the material, µ Often, this is expressed as the product of the permeability of free space, µo, times a relative permeability for the material, µr The B-H relation is Silicon steel (often called soft magnet iron) has a relative permeability of 1000 to 5000. Table 2 gives typical values for various magnetic materials that are used for magnetic bearing construction. At higher values of B, the B-H curve is no longer linear. The level of flux, which can be produced in the magnetic material as the magnetic field intensity H increases, approaches a point where the B-H curve deviates significantly from the linear value. The knee of the curve is called the saturation point. For silicon steel, this typically occurs in the range of 1.5 to 1.7 tesla (1 tesla = 10,000 gauss) for practical magnetic bearing applications. With the advanced magnetic materials such as Vanadium Permendur, this value may be as high as 2.2 to 2.4 tesla. When the bearing operation drives the material to that point, it acts as if it has an air core. The required magnetomotive force is then quite high and not economical. The flux density B in each air gap is Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
580
FIGURE 2. Magnetic flux density B vs. magnetic field intensity H for silicon iron.
as induced by N total turns of wire encircling the magnetic circuit. This flux expression is linearly proportional to the magnetomotive force Ni and inversely proportional to the gap g. The force Fg, per air gap, which attracts the rotor to the stator and permits the actuator to act as a bearing, is given by
There are actually two air gaps in the magnetic bearing circuit so the total force is twice this value
The force is proportional to the square of Ni and inversely proportional to the square of gap g. It would appear from this result that magnetic bearings are very nonlinear devices. However, a closer examination of a double-acting bearing later in this section reveals a different result. Actual magnetic circuits have effects such as fringing and leakage that are not modeled in the previous ideal formulas. These effects are discussed in more detail at the end of this section. A derating factor ε may be conveniently used to give more accurate results including fringing and leakage effects. Then, the force Equation 8 becomes:
Copyright © 1994 CRC Press, LLC
Volume III
581
The derating factor is usually taken as 0.9 for thrust bearings and is closer to 0.8 for radial bearings. The difference primarily reflects higher leakage effects in the radial bearing geometry. A detailed finite element magnetic field solution will give more accurate results, but the derating factor is sufficient for the basic magnetic circuit analysis presented here. Actual operation of the magnetic bearing involves superposition of two fluxes: a bias flux and a perturbation flux, as shown in Figure 2. The bias flux density Bb is a steady state flux level induced in the air gap by a bias ib (steady state) current in the coil. The perturbation flux density Bp is a time-varying control flux density developed by the perturbation (control) current ip in the coil. The total fluxes and currents in the coils are From Equation 6, the two respective fluxes are Usually, the bias flux level Bb is set at about one half of the magnetic saturation level shown in Figure 2, allowing for relatively large perturbation flux levels up and down from the bias level. The associated force expression of Equation 8 becomes Electromagnetic forces are only attractive so actuators must be placed on both sides of the moving components in a double-acting arrangement, as illustrated in Figure 1. The net force FN is given by Assuming that the gap areas are the same in both actuators and substituting Equation 10 gives 10
Copyright © 1994 CRC Press, LLC
582
This can be expanded and then simplified to give
CRC Handbook of Lubrication and Tribology
This is the net force expression in terms of either flux or magnetomotive force. Since the bias flux density Bb is constant, the net force FN is directly proportional to the perturbation flux Bp, rather than proportional to the square, as in the single-acting force relation—Equation 8. If the rotor position is steady with respect to time and centered (equal air gap on both sides), the net force exerted by the double-acting magnetic actuator is zero. However, if the perturbation current is sinusoidal, the net rms force will be
The maximum force capability of magnetic bearings occurs when the magnetic material is saturated and will not develop any more flux. Then the maximum force of the double-acting magnetic bearing will be obtained with one side reduced to zero and the other at maximum flux. The maximum force expression becomes
This is the force limit on magnetic bearings. Magnetic bearings will not produce force beyond this level without a very large magnetomotive force to drive the flux level higher. The pressure loading is a useful quantity obtained by dividing the maximum force by the two pole face areas with the result
Using a saturation flux of 1.6 tesla, the pressure loading has an approximate value of 9.2 x 105 N m-2 (130 lb in. --2) based upon the actual air gap area. This provides a rule-of-thumb estimate of the maximum load capacity which can be delivered. In practice, spacing for coils, tip gaps, and other factors reduce this significantly. Magnetic bearings have a backup bearing system which acts when the maximum force limit is exceeded. Typically, the backup bearing clearance is chosen as one half of the magnetic bearing geometrical clearance. Additional maximum force limits may arise from inadequate power amplifier response rate, as discussed later. Magnetic actuator forces change with both current and air gap thickness. The ratio of a change in force to a change in coil current Ki, called the current gain, is the more important factor for magnetic bearings. Alternatively, the ratio of a change in force to a change in air gap thickness Kx (corresponding to a change in rotor position) is called the position stiffness. The current gain within the linear range, for all four air gaps in a double acting bearing, is defined as
where the derivative is evaluated at ip = 0 and the rotor is centered so that x = 0. The current gain is positive for a magnetic bearing because an increased current produces an increased force which tends to center the rotor force on the moving component. Copyright © 1994 CRC Press, LLC
Volume III
583
The second term on the right side of Equation 19, the linearized expression for the current gain, is independent of the perturbation current but linearly related to the bias current. Thus the magnetic actuator should not be designed with a very low bias current to avoid poor response when a change in force is required. If ib, is one half of the saturation value, the dynamic range of the actuator is a maximum. When the total current on the unloaded side of the bearing is zero (ib - ip = 0), that side of the bearing is turned off; at the same time the loaded side of the bearing is approaching saturation. The next parameter is the position stiffness. The gap thicknesses on either side of the bearing may be written as where the steady state gap thickness is go and the rotor position is x, measured from the centered position. The position stiffness, for all four air gaps in a double-acting bearing, is defined as
where the derivative is evaluated with x = 0 and ip = 0. The position stiffness is negative. As the moving component moves closer to one side, the force increases tending to pull it further in that same direction, unlike a mechanical spring which tends to return it to the center. Since the shaft position change x is typically small compared to the gap dimension g, the linearized expression given on the right of Equation 21 is valid over the range of operation of most magnetic bearings. With these two parameters, the net force is given by
The next section discusses the control circuit which adjusts the current term in this equation. The rate at which the bearing must change force, called the force slew rate, depends upon the rate of change of applied forces on the rotor and must be determined in the rotor/bearing dynamic design process. For a sinusoidal force, this is equal to the net peak force times the frequency of the applied force. The required time rate of change of current, known as the current slew rate, to produce this rate of change of force is Since this required current slew rate is inversely proportional to the current gain, a magnetic bearing with low Ki must have a high voltage power supply to drive it at a high current slew rate.
SENSORS, CONTROLLERS, AND POWER AMPLIFIERS
The electronic circuit which controls the current in the stator coils has three components: sensor, controller, and power amplifier.8,9 A block diagram is shown in Figure 3. It determines coil current in the actuator based upon the rotor (either thrust collar or radial shaft collar) Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
584
FIGURE 3. Electronic feedback circuit block diagram.
position. The general equation (employing Laplace transforms where s is the complex frequency) for the control circuit is where the overall controller transfer function is G(s). If there is no feedback controller employed, the negative position stiffness Kx of the actuator from Equation 21 shows that the actuator is unstable. The primary purpose of the feedback control is to stabilize the rotor and keep it centered. Position sensors used to continuously monitor rotor position for magnetic bearings include eddy current sensors, induction sensors, optical sensors, capacitance sensors, and others. Most commonly, induction sensors of construction similar to the bearings themselves are employed, as shown in Figure 4. The changes in the reluctance across the magnetic gap as detected by the position sensor indicate the opening or closing of the gap. Optical sensors can be used in a vacuum or other clean application. The sensor has a small output voltage proportional to the shaft position. The equation is where Ks is the sensor gain. There is a high frequency rolloff in the sensor amplifier so the transfer function representing the sensor system is
where the denominator terms indicate the rolloff time constant times the frequency. Sensors can be configured in two basically different ways: (1) point sensors and (2) distributed sensors. Point sensors sense the motion of the shaft in a single coordinate direction. Typically, one is used in a thrust bearing and two are used in a radial bearing, at 90 degrees relative to one another. A minimum of one sensor is required per coordinate direction. Copyright © 1994 CRC Press, LLC
585
Volume III
FIGURE 4. Distributed position sensor geometry.
Distributed sensors have the same sort of geometric configuration as the bearing itself, as in Figure 4. For sensing shaft motions, the sensor can be placed anywhere along the shaft unless thermal growth might cause a problem in setting the thrust bearing air gap. In a radial magnetic bearing, the radial sensor should be placed as close as physically possible to the bearing location. If the two components are too far apart, control problems, called noncollocation controls problems, may arise. A particular problem arises when a shaft node point, associated with a particular rotor natural frequency, is located between the sensor and bearing. Other sensors for magnetic bearings may measure the magnetic flux or current. A Hall effect sensor may be employed to measure the flux directly in the air gap or tip coils may be used to indirectly measure the flux in the pole face. Current sensors are employed, particularly as part of the current amplifier. However, these are feedback signals normally used in addition to, rather than in place of, position feedback to control magnetic bearings. The sensor input voltage to the controller output voltage transfer function is
The basic control algorithm for many magnetic bearings is a PID (proportional-integralderivative) control written as
where Kp is the proportional gain constant (similar to a stiffness term), Kic is the integral gain constant (associated with low frequency centering of the shaft), and Kd is the derivative gain constant (similar to a damping term). If the shaft is off center due to some static force acting on it, such as gravity sag, the control loop keeps on increasing the magnetic force to return it to the desired position. At low frequency, the PID phase lag is 90 degrees. At high frequency, the PID phase lead is 90 degrees as needed to stabilize the magnetic bearing-control loop. Figure 5 shows a plot of the ideal PID controller amplitude and phase angle vs. frequency. The frequency indicated is the vibration frequency of some excitation rather than the shaft rotating frequency. It is impossible to construct an electronic circuit (PID) with infinite gain at infinite frequency. There is always some rolloff at high frequency, as illustrated in Figure 5. Typically, this may be modeled as some second order denominator term of the form Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
586
FIGURE 5. Magnitude and phase diagram of ideal and real PID controller.
where bc1 and bc2 are constants determined from the controller circuit design. Generally, this high frequency rolloff is placed so that the magnetic bearing frequencies which need to be controlled, such as machine operating speed, are below the rolloff frequency but higher frequencies, such as high shaft natural frequencies, are not controlled. In the intermediate frequency range, problems may arise which must be dealt with by more advanced control techniques. The output from the control circuit is typically a small voltage proportional to the desired current required for the bearing coils. The current is usually rather large (1 to 10 amps), on the order of amps, so a power amplifier (and possibly a preamplifier) is required for each bearing coil. The transfer function of the amplifier is
where Ka, is the amplifier gain and Ta is the time constant of the amplifier high frequency rolloff characteristic. Copyright © 1994 CRC Press, LLC
587
Volume III
Usually, there will be one amplifier per coil (or per quadrant in a radial bearing) in the magnetic actuator. An amplifier that produces the desired current which is proportional to the controller output voltage is called a transconductance amplifier. A linear amplifier can be employed, but it has rather low efficiency, on the order of 5 to 10%. Most magnetic bearings employ switching amplifiers, which operate in an alternating on-off mode, to produce the required current by varying the period of the on-off cycle. Efficiencies are in the 90 to 95% range. The amplifier must be designed to cope with the relatively high inductance of the actuator coils. In order to obtain the desired current out of the amplifier, the output current is sensed by a shunt resistor or other nonintrusive device. Then an inner feedback circuit is placed around the amplifier to produce the desired current output. As noted earlier, one important limiting factor on magnetic bearing performance is the bearing slew rate required to counter the dynamic forces applied to the rotor. The power supply voltage is given by where R is the coil resistance and L is the coil inductance. Usually the coil inductance is large and the resistance term can be neglected. Thus, the current slew rate equals the amplifier voltage divided by the coil inductance. If the amplifier does not have a high enough supply voltage, the magnetic bearing will be unable to respond quickly enough to rotating machinery forces such as surge in compressors and part flow conditions in pumps. The overall transfer function for the sensor, controller, and power amplifier is
and the resulting transfer function is
While this is a typical magnetic bearing control circuit design, there are many possible variations. More advanced control systems are also employed using state space methods. The following net force expression, from Equation 22, governs the bearing static and dynamic operation
INDUSTRIAL CANNED MOTOR PUMP APPLICATION
Canned pumps, used in chemical and petrochemical industrial plants, have a rotating component (motor rotor plus pump impeller) which is surrounded by the working fluid, such as water, oil, sulfuric acid, or hydrochloric acid. Figure 6 shows a diagram of an industrial canned motor pump supported in magnetic bearings.10 The stator of the pump is canned, typically with stainless steel or similar metal, pierced only by the inlet and exit pipes. This exterior canning prevents the working fluid from leaking into the atmosphere and avoids the need for a mechanical seal placed at the point where the shaft exits from the pump to its drive motor. For many years, industrial firms have desired to find an alternative pump with no leakage and long life. One such alternative is the magnetic bearing supported canned pump. Copyright © 1994 CRC Press, LLC
588
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
FIGURE: 6 Canned pump retrofitted with magnetic bearings
589
Volume III
While common canned pumps have offered a potential solution to the emission problem, their sleeve bearings are lubricated with the process fluid, which normally is a poor lubricant. As a result, the bearings often wear or overheat and fail. A major chemical company found that the average life of existing canned pumps for a specific application was 11 months in their chemical plants. In addition to avoiding leakage and increased life, magnetic bearings offer further benefits of diagnostic capability and vibration control. Pump vibration amplitude is automatically monitored by the bearing sensors, and the bearing coil currents continuously monitor pump performance and can be used to determine when problems have occurred. To evaluate industrial performance, two pumps were retrofitted with magnetic bearings. The first pump was a laboratory prototype10,13 and the second is currently in field testing in an industrial chemical plant.14,15 The first pump was a conventional centrifugal overhung single stage 20-Hp pump with a design flow rate of 300 gal/min in water at 120 ft of head. Figure 6 shows a diagram of the magnetic bearing supported pump. The overall length of the pump was 21 inches and the motor diameter was 8 in. Both the stator and rotor of the three-phase synchronous motor were canned with stainless steel cans. A return flow line cooled the motor. New bearing support sections were constructed to accommodate the magnetic bearings and sensors.11 Two single-acting thrust bearings, one on each end of the motor rotor and two radial magnetic bearings were placed in approximately the same locations as the original sleeve bearings. Five standard industrial eddy current position sensors were used, one for each coordinate axis, to provide the feedback signal for the magnetic bearings.
THRUST BEARINGS
The basic thrust bearing geometry for the canned pump consists of an electromagnetic stator and a thrust collar placed on the rotor. The stator and rotor are separated by a nonmagnetic gap to allow for rotation without physical contact. Figure 7 shows an exploded view of the stator, shaft, and thrust collar. The stator is composed of an inner and outer toroid connected by a common base. The inner and outer toroids and base may be constructed of one piece or separate pieces assembled together. The thrust collar is attached to the shaft. In many applications, the thrust bearing is double acting with the thrust collar between two stators, as shown in Figure 8. A coil of wire between the inner and outer toroids produces the magnetomotive force to drive the bearing. As an example, consider the magnetic thrust bearing employed in the prototype industrial canned pump as shown in Figure 6: N = 576 turns and i = 1.5 amps, where #28 wire was employed. This produces an magnetomotive force of MMF = Ni = 864 ampturns. The magnetic flux path goes through the outer stator toroid, through the outer air gap, through the thrust collar, back through the inner air gap, and returns to the stator inner toroid. The thrust face outer and inner gap areas are given by:
where the diameters for the canned pump application are shown in Figure 9 as
Copyright © 1994 CRC Press, LLC
590
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
FIGURE 7 Exploded view of single-acting magnetic thrust bearing.
591
Volume III
FIGURE 8. Double -acting thrust bearing geometry.
FIGURE 9. Diameters and lengths of thrust bearing.
In this design, the cross sectional area of the flux path is made the same, as closely as possible, over the length of the magnetic circuit. The outer and inner toroid areas are equal, with value The other thrust bearing dimensions are
The flux path cross-sectional area in the end and thrust disk is close to that in the toroids. Copyright © 1994 CRC Press, LLC
592
CRC Handbook of Lubrication and Tribology
FIGURE 10. EI thrust bearing design.
It is important to provide a good flux path to minimize leakage from the magnetic components. Usually, there are no alternating magnetic fields that generate eddy currents in the thrust collar so that it does not normally have to be laminated (see a later section for more detailed discussion). Magnetic thrust bearings can also be configured with two coils, as shown in Figure 10. The magnetic material has three face areas. The flux path goes around each of the two coils into the three thrust face areas so that the middle thrust face must have twice the cross sectional (pole face) area as either of the other two. It is usually called an “E” or “EI” design. The air gap g was chosen to be 0.508 mm (0.020 in.) for the canned motor pump thrust bearing. Equation 5 gives the flux density B = 1.068 tesla (10,680 gauss) at the nominal operating point. At this point, the flux level should be compared to the saturation flux of the material. In the case of the canned motor pump example, the magnetic bearings were constructed from silicon steel. For this material, the saturation level is 1.2 to 1.6 tesla or well above the operating flux level. The nominal operating ideal force which can be taken by the thrust bearing is calculated from Equation 7 as 2,020 N (456 lb). This assumes that only one side of the double-acting thrust bearing is on while the other side is off. Using a typical thrust bearing derating factor of 0.9, the operating actual force is FN = 1818 N (410 lb). If the current capability of the coil and associated electronic supply system are sufficient to push the flux density up to 1.6 tesla, the peak force which can be taken by this single-sided thrust bearing is Fmax = 2,720 N (612 lb). The current gain formula is given by Equation 19. For the canned motor pump the numerical value is Ki = 2,430 N/amp (544 lb/amp). The position stiffness is obtained from Copyright © 1994 CRC Press, LLC
593
Volume III
FIGURE 11. Planar radial magnetic bearing.
FIGURE 12. Homopolar radial magnetic bearing geometry.
Equation 21. Again taking the canned motor pump example, Kx = -7.15 x 106 N/m (-41,200 lb/in.).
RADIAL BEARINGS
In the radial bearing configuration, several stator magnets are arranged around the shaft so that they can exert a radial force on the shaft. Two basic configurations are typical: (1) planar flux path bearings, and (2) axial flux path bearings. The former type is commonly referred to as a planar radial bearing because the flux paths are essentially confined to a plane perpendicular to the axis of shaft rotation, as illustrated by Figure 11. The latter type is sometimes called a homopolar bearing and is depicted in Figure 12. This name arises because the stator geometry generates flux paths which circulate axially along the shaft axis from one stator plane to another so that all of the poles in a given stator plane have the same magnetic polarization. At present, the planar configuration is most commonly used in commercial applications, although recently much attention has been given to some potential advantages of homopolar bearings particularly in high-speed applications. Planar radial bearings have a number of magnetic poles placed around the shaft as shown in Figure 11 which illustrates an eight-pole geometry. It is common to organize the poles into sets occupying each of four quadrants in the stator. If the number of poles is an even multiple of four, then it is fairly simple to control the currents in each of the quadrants so mat they operate as four essentially independent magnetic circuits. While this type of operation is not necessary, it considerably simplifies the control and is conceptually easier to understand than the more general interdependent schemes. Copyright © 1994 CRC Press, LLC
594
CRC Handbook of Lubrication and Tribology
With this quadrantal organization, the coils controlling the fluxes in the legs within a particular quadrant can be wound in series. This permits all of the coils in each quadrant to be controlled by a single power amplifier. Further, the forces generated by the bearing are segregated along two perpendicular axes, generally referred to as the x- and the y-axes. Consequently, the two directions can be controlled independently by measuring the shaft motion along each axis separately. The measured motion is supplied to a controller of some simple configuration such as PID. The controller, in turn, sets the perturbation currents in opposing quadrants along each separate axis. In this manner, each radial bearing requires two independent position sensors, two controllers, and four power amplifiers. The force generated by each individual pole face of the magnetic stator is determined by the flux density in the air gap and the area of the air gap:
where Qi is the angle between the nominal x-axis and the centerline of the ith pole leg. If the individual flux densities are known, then the total force generated by the bearing is simply the algebraic sum of the x- and y- components at each pole face:
In each air gap, the magnetic flux density varies with coil current and shaft position, as shown in Figure 11. The relationship is determined by summing the magnetomotances around closed loops in the stator-rotor circuit and accounting for all of the coil currents linked by the selected loops. Ignoring iron reluctance, this produces equations of the form
The coil currents are determined by the controllers and power amplifiers, while the shaft position affects the air gap lengths according to where go is the nominal air gap, and x and y specify the shaft deflection from centered position. Assume that the bearing has a planar configuration, uses eight evenly spaced pole pieces, that each pole has the same gap area and is wound with the same number of coil turns. Further, the stator will be organized into four quadrants with two legs per quadrant and the coils in each quadrant are wired in series. The pole angles for such a geometry are given by The assumption of series wiring in each quadrant produces Since opposing quadrants are controlled by the same controller which simply introduces a symmetric perturbation in coil currents, we have Copyright © 1994 CRC Press, LLC
Volume III
595
The resulting equations for bearing force are fairly complicated. However, most of the character of the relationship can be seen by examining the Taylor’s series expansions:
and
where each term is evaluated at x = 0, y = 0, ipx, = ip,x,o’, and i>p,y = ip,y,o. The first term in each equation, the nominal bearing force, is equal to the static load which must be supported by the bearing:
The last two terms, which represent first order crosscoupling between axes, are also zero. The remaining elements are the previously mentioned current gain, K,i and position stiffness, Kx:
where it is easily verified that
and
Note that these expressions are only valid as long as the perturbation currents are less than the bias current. It is standard practice in bearings of this design to limit the total current in each coil (sum of the bias and the perturbation currents) to be positive. The terms KI, K, and Ky, can be used to model the actuator when constructing a stability or linear forced response model for the overall rotor-bearing system. Details of assembling such a model are provided in Maslen and Bielk.18 If we define me current required to achieve saturation flux density as Lsat and the force actually achieved at saturation as Fsat then the position stiffness can be rewritten as
Typically, the bias will be chosen so that Ib + ip,x,o = 0.5 Isat and Ib + i>p,y,o =0.5 Isat which implies that Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
596
This provides a quick formula for estimating the position stiffness. While these latter results have been presented for the specific eight-pole geometry with coils wound in series in each quadrant, the properties can be derived for other arrangements following the development given above. It is important to note that the negative values of Kx and Ky imply the well-documented instability of magnetic bearings without feedback. The magnitude of this effect is essentially inversely proportional to the nominal air gap length. This suggests that the destabilizing effect can be reduced by maximizing the air gap length. On the other hand, the current gain is a direct measure of effectiveness of the actuator, and it is inversely proportional to the square of the air gap length. In general, the air gap length should be made as small as is practical and the bias current level should then be selected to place the bias flux density somewhere between one quarter and one half of the saturation flux density. This range reflects a weak tradeoff between thermal considerations and dynamic performance. Radial planar magnetic bearings were designed and installed in the canned motor pump described earlier. They have the dimensions and other properties given in Table 3.
MAGNETIC LOSSES AND OTHER PROPERTIES
Fringing is the tendency of the magnetic field not to go straight across an air gap but to bulge out to the sides. Leakage is the tendency of some of the magnetic flux to go along undesired pathways rather than follow the desired horseshoe-shaped magnetic material path laid out for it. Reduction in the actual force delivered is modeled in an approximate way using the derating factor ε discussed earlier. A detailed analysis of their effects can be obtained by a finite element analysis of the bearing. Two additional effects in magnetic materials influence bearing performance: (1) eddy currents and (2) hysteresis. Eddy currents are induced in the magnetic materials of the stator and rotor by rapidly changing magnetic fields. These induced currents produce two effects. The first is heat generated because the material is resistive; power is dissipated in the form of heat. This translates into drag on the shaft and power losses in the bearing. Second, the induced currents represent back magnetomotance which reduces the H value generated by the wire coils on the stator, and large eddy currents can substantially reduce the flux density produced by the bearing. Eddy currents are generated in the rotor component of radial bearings because the magnetic field alternates from north to south pole in the rotor as the shaft turns, inducing eddy currents primarily in the direction of the shaft axial direction. Laminating the rotor component with thin layers of magnetic material separated by insulation greatly reduces the axially induced eddy currents. The stator should also be laminated when coil currents are expected to change rapidly for some reason, such as when rapidly changing dynamic bearing loads are encountered or when switching amplifiers are used. When a magnetic field intensity H is applied to a magnetic material, a B-H curve such as indicated in Figure 13 is followed, such as from h-j-b. However, when H is removed, the BH curve followed is slightly different, from b-c-BR. Generally, when H is applied to a magnetic material with zero flux and then removed, the flux level does not return to zero. The degree of this hysteresis is dictated by the material composition and the heat treatment. Traditional silicon steel materials have relatively low hysteresis losses. Newer, high saturation density materials tend to exhibit stronger hysteresis effects. A major hysteresis loop goes from e-f-g-h-j-b and returns along b-c-d-e. The area between these curves is the energy lost due to hysteresis. However, an air gap in the magnetic circuit Copyright © 1994 CRC Press, LLC
Volume III 597
Copyright © 1994 CRC Press, LLC
CRC Handbook of Lubrication and Tribology
598
FIGURE 13. Hysteresis plot for silicon steel.
greatly reduces the area covered by the loop and, thus, the hysteresis energy loss. Further, the actual loop covered by the stator of magnetic bearings remains in the first quadrant where both B and H are positive. Thus the hysteresis curve followed is more properly illustrated by the minor loop shown in Figure 13. Hysteresis produces several detrimental effects in magnetic bearings. The first is a power loss due to the increased coil currents required to generate a given dynamic bearing force level as the flux changes with time. Another is the reduction of bearing damping which can be obtained with a given bearing geometry and power supply. Magnetic properties of materials are affected by both operating temperature and stress. As the operating temperature increases, the flux level produced by a magnetomotive force decreases. Thermal energy tends to disrupt the required organized orientation of the material’s magnetic domains. This degradation is manifested both as a reduction in relative permeability and in saturation density. The latter effect is typically more significant, with reductions of as high as 50% at temperatures of the order of 800°F. For annealed materials, the loss in magnetic performance is generally reversible, but for materials which have been enhanced through heat treatment, elevated temperatures can produce irreversible performance degradation. Permanent magnets are particularly sensitive to temperature. This sensitivity is characterized by the Curie temperature (the temperature at which the permanent magnetism is lost). For conventional alnicos and for high-energy product materials such as neodymiumboron alloys, the Curie temperature is on the order of 320°C. In other materials, such as samarium-cobalt, it may be as high as 400°C. Magnetic materials are also sensitive to stress levels. Generally, magnetic properties increase with increased tensile stress and decrease with increased compressive stress. The stress in high-speed rotor laminations is usually tensile so this actually enhances magnetic bearing performance. While rotor dynamics is an important topic for magnetic bearing applications, space is not available in this chapter to discuss it in any detail. Many rigid rotors are supported in magnetic bearings and the dynamic effects must be accounted for, but that is not too difficult. Magnetic bearings are noncontacting and stiffness/damping properties are adjustable within limits, so Copyright © 1994 CRC Press, LLC
599
Volume III
operation is generally quite smooth compared to rotors supported in rolling element bearings. Flexible rotor dynamic effects in magnetically supported rotors is a rather complex topic but very important to magnetic bearing design.16,18 The electronic control system in magnetic bearings allows for control of vibrations. However, the softness of magnetic bearings relative to fluid film bearings and the additional weight of the rotor laminations may lower shaft critical speeds. This may cause some rotor sensitivity to fluid excitations in compressors or turbines. On the other hand, magnetic bearings offer the option of rotating magnetic fields which automatically balance the rotor during machine operation.
ACKNOWLEDGMENTS
The authors would like to thank the Virginia Center for Innovative Technology, Kingsbury Inc., Exxon Chemical Co., Goulds Pump Co., NASA Lewis Research Center, and the U.S. Army Research Center for partial funding of the work reported in this chapter.
A B D F g H I I K, L M MMF N R x ε µ φ ω
= = = = = = = = = = = = = = = = = = =
NOMENCLATURE
Air gap area (m2) Flux density (tesla = Wb/m2) diameter (m) Force (N) Length of air gap (m) Magnetic field intensity (amp-turns/m) Current (amps) QCurrent gain (N/amp) Position stiffness (N/m) Length of magnetic circuit (m) Mass (kg) Magnetomotive force (amp-turns) Number of turns (turns) Reluctance (amp-turns/Wb) Position (m) Fringing and leakage derating factor Permeability (H/m) Magnetic flux (Wb = Nm/amp) Frequency (rad/sec)
REFERENCES
1. Beams, J. W. and Blade, S. A., Electrically driven magnetically-supported vacuum type ultracentrifuge, Rev. Sri. Instrum., 10, 59, 1939. 2. Allaire, P. E., Humphris, R. R., and Lewis, D. W., Professor Jesse W. Beams and the first practical magnetic suspension, in Proc. Int. Symp. Magnetic Suspension Technol., NASA Conf. Pub. 3152, Langley Research Center, 1991. 3. Schweitzer, G., Characteristics of a magnetic rotor bearing for active vibration control, Proc. Inst. Mech. Eng., Pap. No. C239/76, 1976. 4. Haberman, H. and Laird, G., Practical magnetic bearings, IEEE Spectrum, September 1979, 26. 5. Schweitzer, G. and Ulbrich, H., Magnetic Bearings—a novel type of suspension, Institution of Mechanical Engineers, 1980.
Copyright © 1994 CRC Press, LLC
600
CRC Handbook of Lubrication and Tribology
6. Allaire, P. E., Mikula, A., Banerjee, A., and Lewis, D. W., Design and test of a magnetic thrust bearing, J. Franklin Institute, 326(6), December 1989. 7. Allaire, P. E., Magnetic bearing design, in Proc. Conf. Magnetic Bearings and Dry Gas Seals, Washington, DC, March 13 to 15, 1991. 8. Humphris, R. R., Kelm, R. D., Lewis, D. W., and Allaire, P. E., Effect of control algorithms on magnetic bearing properties, J. Eng. Gas Turbines Power, 108, 624, 1986. 9. Williams, R. D., Keith, F. J., and Allaire, P. E., A comparison of analog and digital controls for rotor dynamic vibration reduction through active magnetic bearings, J. Eng. for Gas Turbines Power, Trans. ASME, 113(4), 535, 1991. 10. Allaire, P., Imlach, J., McDonald, J., Humphris, R., Lewis, D., Banerjee, B., Blair, B., Claydon, J., and Flack, R., Design, construction and test of a magnetic bearing in an industrial canned motor pump, in Proc. Texas A&M Pump Symp., Houston, TX, May 1989. 11. Imlach, J., Humphris, R., Blair, B., Allaire, P., and Lewis, D., Testing of a magnetic bearing equipped canned motor pump for installation in the field, in 2nd Int. Symp. Magnetic Bearings, Tokyo, July 1990, 39. 12. Humphris, R. R., Lewis, D. W., Allaire, P. E., Blair, B. J., and Imlach, J., System diagnostics through magnetic bearings applications to a canned motor pump, REVOLVE ‘89, Nova Corp., Calgary, Canada, September 26 to 28, 1989. 13. Blair, B. J., Humphris, R. R., Allaire, P. E., Lewis, D. W., and Barrett, L. E., A canned pump with magnetic bearings for industrial use—laboratory testing, in 25th Intersoc. Energy Conversion Eng. Conf., Reno, Nevada, August 1990. 14. Imlach, J., Blair, B. J., and Allaire, P. E., Measured and predicted force and stiffness characteristics of industrial magnetic bearings, J. Tribol. Trans. ASME, 113(4), 535, 1991. 15. Imlach, J., Blair, B. J., and Allaire, P. E., Measured and predicted force and stiffness characteristics of industrial magnetic bearings, J. Tribol. Trans. ASME, 113(4), 784, 1991. 16. Allaire, P. E., Lewis, D. W., and Knight, J. D., Active vibration control of a single mass rotor on flexible supports, J. Franklin Inst., 315(3), 211, 1983. 17. Maslen, E. H., Allaire, P. E., Scott, M. A., and Hermann, P., Magnetic bearing design for a high speed rotor, in Proc. 1st Int. Conf. Magnetic Bearings, Springer-Verlag, New York, 1988, 137. 18 Maslen, E. H. and Bielk, J. R., A stability model for flexible rotors with magnetic bearings, J. Dyn. Syst. Meas. Control, Trans. ASME, 114, 172, 1992.
Copyright © 1994 CRC Press, LLC