Synthesis Gas Combustion: Fundamentals and Applications

SYNTHESIS GAS COMBUSTION Fundamentals and Applications

SYNTHESIS GAS COMBUSTION Fundamentals and Applications

Edited by

Tim C. Lieuwen Vigor Yang Richard Yetter

Boca Raton London New York

CRC Press is an imprint of the Taylor & Francis Group, an informa business

CRC Press Taylor & Francis Group 6000 Broken Sound Parkway NW, Suite 300 Boca Raton, FL 33487-2742 © 2010 by Taylor and Francis Group, LLC CRC Press is an imprint of Taylor & Francis Group, an Informa business No claim to original U.S. Government works Printed in the United States of America on acid-free paper 10 9 8 7 6 5 4 3 2 1 International Standard Book Number: 978-1-4200-8534-1 (Hardback) This book contains information obtained from authentic and highly regarded sources. Reasonable efforts have been made to publish reliable data and information, but the author and publisher cannot assume responsibility for the validity of all materials or the consequences of their use. The authors and publishers have attempted to trace the copyright holders of all material reproduced in this publication and apologize to copyright holders if permission to publish in this form has not been obtained. If any copyright material has not been acknowledged please write and let us know so we may rectify in any future reprint. Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying, microfilming, and recording, or in any information storage or retrieval system, without written permission from the publishers. For permission to photocopy or use material electronically from this work, please access www.copyright. com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and registration for a variety of users. For organizations that have been granted a photocopy license by the CCC, a separate system of payment has been arranged. Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. Library of Congress Cataloging-in-Publication Data Synthesis gas combustion : fundamentals and applications / editors, Tim Lieuwen, Vigor Yang, Richard Yetter. p. cm. “A CRC title.” Includes bibliographical references and index. ISBN 978-1-4200-8534-1 (hardcover : alk. paper) 1. Synthesis gas--Combustion. 2. Gas as fuel. I. Lieuwen, Timothy C. II. Yang, Vigor. III. Yetter, Richard A., 1952- IV. Title. QD516.S926 2010 665.7’72--dc22 Visit the Taylor & Francis Web site at http://www.taylorandfrancis.com and the CRC Press Web site at http://www.crcpress.com

2009026277

Contents Preface......................................................................................................................vii The Editors.................................................................................................................ix Contributors...............................................................................................................xi Chapter 1. Gasification Technology to Produce Synthesis Gas..............................1 Geo A. Richards and Kent H. Casleton Chapter 2. Syngas Chemical Kinetics and Reaction Mechanisms....................... 29 Marcos Chaos, Michael P. Burke, Yiguang Ju, and Frederick L. Dryer Chapter 3. Laminar Flame Properties of H2/CO Mixtures.................................. 71 Jayaprakash Natarajan and Jerry M. Seitzman Chapter 4. Fundamental Combustion Characteristics of Syngas............................ 99 Guillaume Ribert, Piyush Thakre, Zhe Wang, Richard A. Yetter, and Vigor Yang Chapter 5. Turbulent Combustion Properties of Premixed Syngas.................... 129 Robert K. Cheng Chapter 6. Pollutant Formation and Control....................................................... 169 Kevin J. Whitty, Hongzhi R. Zhang, and Eric G. Eddings Chapter 7. Syngas Utilization............................................................................. 193 Geo A. Richards, Kent H. Casleton, and Nathan T. Weiland Chapter 8. Catalytic Combustion of Syngas....................................................... 223 John Mantzaras Chapter 9. Operability Issues Associated with Steady Flowing Combustors.... 261 Tim Lieuwen, Vincent McDonell, Domenic Santavicca, and Thomas Sattelmayer v

vi

Contents

Chapter 10. Combustion of Syngas in Internal Combustion Engines.................. 289 Melanie K. Fox, Gregory K. Lilik, André L. Boehman, and Olivier Le Corre Chapter 11. Solid Oxide Fuel Cells Using Syngas............................................... 329 Robert J. Kee, Huayang Zhu, and Gregory S. Jackson Index....................................................................................................................... 375

Preface The confluence of concerns about climate change, environmental degradation, and energy supply raises many far-reaching challenges regarding energy sources and use. Power suppliers and consumers are looking for an affordable energy supply that will cause limited environmental impact. This situation has created unprecedented technological challenges and opportunities. The fact that coal is an abundant and indigenous U.S. resource makes it a likely contributor to the future energy mix. Furthermore, both India and China, two of the most rapidly developing economies in the world, are also heavy coal users. Coal currently supplies over 50% of the electric power in the United States. Coal usage, however, has been associated with degradation of air quality, water resources, and habitat. In addition, concerns over the role of carbon dioxide (CO2) emissions in global warming raise additional questions about carbon management with coal usage. Currently, 80% of the CO2 emissions due to electric power production come from coal. The continued use of coal and the need to reduce CO2 emissions will thus require lower cost and more effective approaches to coalbased power generation with reduced emissions. Such “clean coal” energy conversion devices will rely on combustion of gasified coal, referred to as synthesis gas, or syngas. Developing a basic understanding of synthesis gas and hydrogen combustion has relevance to many situations—the most near-term application, however, appears to be coal-based integrated gasification combined cycle (IGCC) technology. Through coal-based IGCC technology, cleaner electric power production with reduced carbon dioxide emissions is possible. Opportunities for cleaner coal use, hydrogen combustion and production, and the potential for CO2 capture are all strong motivators for the development of IGCC technology. IGCC technology produces significantly lower emissions than other coal-based power systems. When IGCC emissions are compared with those of its nearest coalbased competitor—the super critical pulverized coal (SCPC) boiler power plant—we see that IGCC produces 82% less carbon monoxide, 24% less oxides of nitrogen, 71% less sulfur dioxide, 66% less mercury, 97% less fluorides, 90% less sulfuric acid mist, and 58% less particulate matter. For a 600 megawatt (MW) IGCC power plant, this would represent an overall reduction of 6426 tons annually of regulated pollutants, compared to SCPC. Furthermore, system-level–based models indicate that IGCC is the lowest-cost, coal-based approach to CO2 mitigation through CO2 capture and geologic storage. There is also growing interest in the combustion of bio-derived synthesis gas. Signatory countries to the Kyoto protocol must minimize CO2 emissions—this has motivated substantial interest in utilizing bio-derived fuels, such as tree residues, agricultural waste, paper and pulp residue, switchgrass, and other biomass, which emit near zero net CO2 emissions. vii

viii

Preface

There are other benefits to gasification technology that go beyond reduced emissions and more effective CO2 capture. These include opportunities for hydrogen and liquid fuel production, the use of biomass feedstocks, and compatibility with fuel cells and other advanced power generation technologies. For the above reasons, there is significant interest in better understanding the combustion characteristics of coal and bio-derived synthesis gases. This book presents the current understanding of syngas combustion by summarizing and compiling it into a single volume. The book is divided into two sections, focusing on combustion fundamentals and application/utilization-specific issues. The first section begins with a chapter on syngas production, detailing the technical issues and trade-offs that influence fuel composition. The remaining chapters in this section then treat syngas combustion fundamentals, such as chemical kinetics, laminar and turbulent flame properties, and emissions. The second section details application-specific issues associated with syngas usage in fuel cells, internal combustion engines, and steady-flowing combustion devices, such as gas turbines or boilers. It begins with an overview chapter, with the subsequent chapters focusing on different energy utilization devices. Publication of this book was made possible through the substantial contributions of a number of individuals. We thank the authors for sharing their time and talent in preparing their manuscripts and carefully revising them. The technical drawing and editorial services provided by Xiaodong Chen are gratefully acknowledged. Richard A. Dennis National Energy Technology Laboratory U.S. Department of Energy Timothy C. Lieuwen Georgia Institute of Technology Vigor Yang Georgia Institute of Technology Richard A. Yetter Pennsylvania State University

The Editors Tim Lieuwen, Ph.D., P.E., is an associate professor in aerospace engineering at Georgia Institute of Technology. Dr. Lieuwen performs research in areas relating to clean combustion technologies and alternative fuels. He is an associate editor of the Journal of Propulsion and Power, Combustion Science and Technology, and the Proceedings of the Combustion Institute. He is also on the editorial review board of the American Institute of Aeronautics and Astronautics (AIAA) Publication Committee. Dr. Lieuwen has been the recipient of a variety of teaching awards, best paper awards, and the AIAA Lawrence Sperry Award.

Vigor Yang is William R. T. Oakes Professor and Chair of the School of Aerospace Engineering at Georgia Institute of Technology. He also serves as the editor-in-chief of the Journal of Propulsion and Power. He is the author and editor of several books on propulsion and combustion.

Richard Yetter is professor of mechanical engineering at Pennsylvania State University. He has conducted research in the fields of combustion and propulsion for more than 30 years. He is coauthor of the 4th Edition of Combustion and serves as editor-in-chief for Combustion Science and Technology.

ix

Contributors André L. Boehman Department of Energy and Mineral Engineering Pennsylvania State University University Park, Pennsylvania

Melanie K. Fox Department of Energy and Mineral Engineering Pennsylvania State University University Park, Pennsylvania

Michael P. Burke Department of Mechanical and Aerospace Engineering Princeton University Princeton, New Jersey

Gregory S. Jackson Department of Mechanical Engineering University of Maryland College Park, Maryland

Kent H. Casleton National Energy Technology Laboratory U.S. Department of Energy Morgantown, West Virginia Marcos Chaos Department of Mechanical and Aerospace Engineering Princeton University Princeton, New Jersey Robert K. Cheng Environmental Energy Technology Division Lawrence Berkeley National Laboratory Berkeley, California Frederick L. Dryer Department of Mechanical and Aerospace Engineering Princeton University Princeton, New Jersey Eric G. Eddings Department of Chemical Engineering Institute for Clean and Secure Energy University of Utah Salt Lake City, Utah

Yiguang Ju Department of Mechanical and Aerospace Engineering Princeton University Princeton, New Jersey Robert J. Kee Engineering Division Colorado School of Mines Golden, Colorado Olivier Le Corre Department of Energetics and Environmental Engineering École des Mines de Nantes Nantes, France Gregory K. Lilik Department of Energy and Mineral Engineering Pennsylvania State University University Park, Pennsylvania John Mantzaras Paul Scherrer Institute Combustion Research Villigen PSI, Switzerland xi

xii

Vincent McDonell UCI Combustion Laboratory University of California (UCI) Irvine, California Jayaprakash Natarajan Georgia Institute of Technology Atlanta, Georgia Guillaume Ribert CORIA–CNRS 6614–INSA de Rouen Campus du Madrillet St. Etienne du Rouvray, France Geo A. Richards National Energy Technology Laboratory U.S. Department of Energy Morgantown, West Virginia Domenic Santavicca Department of Mechanical and Nuclear Engineering Pennsylvania State University University Park, Pennsylvania

Contributors

Piyush Thakre Department of Mechanical and Nuclear Engineering Pennsylvania State University University Park, Pennsylvania Zhe Wang Department of Mechanical and Nuclear Engineering Pennsylvania State University University Park, Pennsylvania Nathan T. Weiland National Energy Technology Laboratory West Virginia University Morgantown, West Virginia Kevin J. Whitty Department of Chemical Engineering Institute for Clean and Secure Energy University of Utah Salt Lake City, Utah

Thomas Sattelmayer Lehrstuhl für Thermodynamik TU–München Garching, Germany

Hongzhi R. Zhang Department of Chemical Engineering Institute for Clean and Secure Energy University of Utah Salt Lake City, Utah

Jerry M. Seitzman School of Aerospace Engineering Georgia Institute of Technology Atlanta, Georgia

Huayang Zhu Engineering Division Colorado School of Mines Golden, Colorado

Technology 1 Gasification to Produce Synthesis Gas Geo A. Richards and Kent H. Casleton Contents 1.1 Overview............................................................................................................1 1.2 Feedstock Properties..........................................................................................3 1.2.1 Reactivity...............................................................................................4 1.2.2 Ash Composition and Properties...........................................................6 1.2.3 Feedstock Preparation...........................................................................7 1.3 Gasifiers.............................................................................................................8 1.3.1 Moving Bed Gasification.......................................................................9 1.3.2 Fluid Bed Gasification......................................................................... 10 1.3.2.1 Bubbling Bed........................................................................ 11 1.3.2.2 Circulating Fluid Bed........................................................... 12 1.3.2.3 Transport Reactor................................................................. 13 1.3.3 Entrained Flow, Slagging Gasifiers..................................................... 13 1.3.4 Comparison of Gasifier Types and Approaches.................................. 15 1.3.5 Syngas Thermal Management............................................................. 16 1.4 Syngas Purification.......................................................................................... 18 1.4.1 Cold Gas Cleanup................................................................................20 1.4.1.1 Acid Gas Scrubbing.............................................................. 22 1.4.1.2 Tars and Carbonyls............................................................... 23 1.4.2 Warm Gas Cleanup.............................................................................. 23 1.5 Conclusions......................................................................................................24 References.................................................................................................................25

1.1 Overview This chapter presents an overview of the processes that are used to create synthetic gas (syngas) for applications described in subsequent chapters. The production of syngas has a long history, dating back to the 1800s. Before the wide availability of electricity and natural gas, “town gas” was generated by coal gasification, and distributed for street lighting, residential cooking, and industrial heating. Some interesting historical mile­stones for gasification are shown in Table 1.1. While electricity and natural gas have supplanted syngas in these instances, growing demand for energy and chemicals has renewed interest in syngas technology. There are two major reasons: First, syngas can be generated from multiple solid fuel feedstocks, providing greater 1

2

Synthesis Gas Combustion: Fundamentals and Applications

Table 1.1 Some Milestones in the Early History of Gasification 1804: Coal gas first patented for lighting. 1813: Westminster Bridge (London) illuminated with “town gas” lights on New Year’s Eve using wooden pipes for gas delivery. 1816: Baltimore, Maryland, becomes the first U.S. city to light streets with town gas. 1800s: Town gas lighting in factories replaces candles and lanterns, making the night shift possible and enabling the Industrial Age. Source: NETL (2008).

opportunity to use low-cost and renewable fuels. Second, syngas itself can be flexibly used for power genera­tion, fuel production, or chemical manufacturing. In addition, the carbon associated with the feedstock can be converted to CO2 and separated for geologic sequestration. Historically, many approaches to gasification have been developed, and some have adopted specific terminology that is still used occasionally. Some of these are listed in Table 1.2. Throughout this chapter, the term syngas will be used to describe gasification products from any process. Syngas is ideally a mixture of hydrogen and carbon monoxide produced by gasifying a solid fuel feedstock (such as coal or biomass). Figure 1.1 shows a schematic repre­sentation of the general process of converting a solid fuel to synthesis gas. The process is usually carried out by using the heat from carbon oxidation to sustain the gasification reaction. The solid fuel is mixed with the oxidant (air or oxygen) to gasify the fuel. Water or steam is added to control the reaction temperature, and participates in some of the gasification reactions shown in Figure 1.1. In actual practice, the details of this con­version are complicated, and multiple products aside from CO and H2 may be included in the syngas, such as CH4 and tars. Tars are complex mixtures of hydrocarbon materials that can condense on downstream equipment if not removed Table 1.2 Historical Terminology for Various Types of Syngas Town Gas—Syngas that was generated from coal and distributed principally for lighting in the late 1800s. Depending on the gasification approach, higher hydrocarbons could be added to create a yellow flame for illumination (termed carbureted). Water Gas—Syngas produced by reacting hot coke with steam, producing nearly equal volumes of CO and hydrogen. Producer Gas—Syngas produced by reacting humid air with coke, resulting in syngas with significant nitrogen diluent. Blast Furnace Gas—The product gas from blast furnaces where coke is used to reduce iron oxide to iron. The resulting gas is mostly nitrogen-diluted CO, because air is used to oxidize the coke. Source: Shadle et al. (2002).

Gasification Technology to Produce Synthesis Gas

3

Major Gasification Reactions

Oxygen or Air

Gasification with Oxygen CO –111 MJ/kmol C + 1/2 O2

Combustion with Oxygen CO2 –283 MJ/kmol CO + 1/2 O2 Gasification with Carbon Dioxide 2CO + 172 MJ/kmol C + CO2 Gasification with Steam C + H2O CO + H2 + 131 MJ/kmol Gasification with Hydrogen C + 2H2 CH4 + –75 MJ/kmol Water–Gas Shift CO + H2O H2 + CO2 –41 MJ/kmol

Fuel (coal, biomass, petcoke, waste…)

Water (Steam)

Syngas

Ash/Slag

Figure 1.1  Gasification process and major reactions.

or further processed and utilized. Impurities in the solid feedstock (compounds of sulfur, nitrogen, chlorine, and others) will produce impurity species that need to be removed from the syngas. Solid ash residue, the noncombustible material that is primarily the inorganic component of the fuel feedstock, must likewise be separated. The syngas is usually cooled to allow impurity removal, providing sensible heat that can be used to raise steam for power generation, or chemical processes. To understand the various process steps involved in producing clean syngas, this chapter is divided into three sections: properties of the fuel feedstock, approach to gasifica­tion, and gas treatment and conditioning.

1.2 Feedstock Properties This chapter emphasizes solid fuel gasification. Coal is the predominant source of gasifier feedstock, supplying 55% of syngas worldwide in 2007 (Gasification World Database, 2007). Although not discussed in this chapter, petroleum and natural gas can also be used to create syngas, supplying 33% and 12%, respectively (Gasification World Database, 2007). In many aspects, liquid and gaseous feedstocks are simpler to gasify than solid fuels. Heavy petroleum liquids may have ash residues and impurities that require consideration, as with solid fuels, but in general are easier to feed into the gasifier. More discussion on gas and liquid fuel gasification can be found in Higman and van der Burgt (2003). The solid fuel feedstock properties are a major factor in deciding what type of gasifier is best suited to a given application. There is a significant body of literature on solid fuel properties; the discussion here will focus on just those properties that

4

Synthesis Gas Combustion: Fundamentals and Applications

are relevant to explain why some gasifier designs are a better (or worse) match to a given feedstock. Properties of some example feedstocks are discussed; other types of feedstock can be found in the literature data, or recorded using standard methods. The American Society for Testing and Materials (ASTM) has established test methods that can be used to measure properties of interest. Details can be found in the ASTM literature. The major properties of interest to evaluate a given fuel are included in the proximate analysis. The proximate analysis for coal is carried out following standard procedures as found in ASTM D3172. An example of the proximate analysis for several coals, biomass, and some petcoke is presented in Table 1.3. The properties of biomass are not necessarily measured in the same manner as coal. Jenkins et al. (1998) describe the different ASTM methods associated with biomass, and present data for many different types of fuels. Comparing the different coals in Table 1.3, it is important to note that the moisture and ash content of the coal are widely variable. For the coals listed, notice, for example, that up to 44% of the coal feedstock can be ash and water that do not contribute to the syngas product (see the lignite entry). Aside from simply diluting the usable feedstock, these constituents rob the syngas product in a second manner. They must be heated to the gasification temperature, yet do not contribute to usable product. For example, if coal with 10% ash is gasified at 1400°C, the ash must be heated to 1400°C by using heat generated during oxidation of carbon. Similar comments apply to the coal moisture, although water can participate in some desirable chemistry, such as hydrogen production. The quantity of heat supplied to the ash and moisture is not available to drive endothermic gasification reactions. This means that more carbon must be “burned” simply to sustain gasification temperatures, thereby lowering the overall process efficiency. The situation is exacerbated by gasifiers that operate at higher temperatures, and explains why moist high-ash coals, particularly when prepared with a water slurry for pumping, are not a good match for high-temperature entrained gasifiers, discussed later. When reporting the performance of gasifiers, it is common to describe the syngas properties per mass of feedstock. Because ash and moisture can represent a significant part of the feedstock, it is important to distinguish performance using coal “as received (ar),” “moisture free (mf),” or “moisture and ash free (maf).” Sometimes maf is designated “dry ash free (daf).” The actual values of moisture and ash content are measured according to ASTM D3302 and D3172 for coal, or E871, E830, and D1102 for biomass. Table 1.3 also includes the ultimate analysis and some ash fusion temperatures, discussed later. The ultimate analysis lists the feedstock composition in terms of chemical elements, and is especially useful to determine the balance of oxygen (and possibly steam) required to produce syngas.

1.2.1 Reactivity In addition to the properties listed in Table 1.3, the reactivity of a feedstock is an important consideration for gasification. The rate of carbon conversion, R = dC/dt, is a measure of reactivity. Depending on the temperature, the rate can be controlled by

33.3 23.4   5.2 17.6   7.7 n/r n/r n/r n/r n/r n/r   6.0   2.2

43.6 40.8 40.2 44.2 6.4 72.9 76 87.1 85.6 76.7 84.8 9.1 5.1

Volatile Matter % (dry) 45.3 54 50.7 45 83.1 24.2 18.7 12.4 12.0 14.4 12.5 89.8 93.6

Fixed Carbon % (dry) 11.1 5.2 9.1 10.8 10.5 2.9 5.3 0.5 2.44 9.0 2.7 1.1 1.35

Ash % (dry) 63.3 72.0 74.0 69.0 83.7 53.4 49.7 48.3 48.6 46.7 50.2 88.7 86.3

C 4.5 5.0 5.1 4.9 1.9 5.6 5.4 6.1 5.9 5.8 6.1 3.6 2.2

H 19.0 16.4 7.9 10.0 2.3 37.9 39.3 45.0 42.8 37.4 40.4 0.0 0.8

O 1.0 1.0 1.6 1.0 0.9 0.1 0.2 0.03 0.2 0.8 0.6 1.8 2.4

N

 Ultimate Analysis % (dry) 

1.2 0.4 2.3 4.3 0.7 0.1 0.1 0.03 0.04 0.2 0.02 4.7 6.9

S 24.7 28.9 30.7 29.0 29.9 18.4 17.5 17.2 19.0 18.1 19.0 33.8a 32.6a

Dry HHV (MJ/kg)

Notes: A = Babcock and Wilcox Co. (2005); B = Tillman (1994); C = Jenkins et al. (1998); D = Bryers (1995); n/r = not reported. a Not reported wet or dry analysis.

Lignite, North Dakota Sub-Bituminous, Montana Bituminous, Pittsburgh #8 Bituminous, Illinois #6 Anthracite Pine bark Oak bark Tan oak Bagasse Switchgrass Poplar Petcoke (delayed) Petcoke (fluid)

Moisture %

Proximate Analysis 

Table 1.3 Properties of Representative Feedstocks Used for Gasification

1,110 1,160 1,216 1,055 n/r 1,194 1,477 1,377 n/r n/r n/r 1,540+ 1,380

Ash Initial Deformation Temperature, Reducing Conditions °C

A A A A A B B B C C C D D

Source

Gasification Technology to Produce Synthesis Gas 5

6

Synthesis Gas Combustion: Fundamentals and Applications

gaseous diffusion to the fuel particle, diffusion in the pores, or the chemical reaction rates themselves. It is difficult to precisely quantify the reactivity without specific measurements because it can be influenced by a host of factors, including catalytic effects from ash constituents, and the temperature history of the char. Reactivity data are often developed from thermogravimet­ric analysis (TGA) measurements. The fuel is heated at a rate that is much slower than in most practical gasifiers. Megaritas et al. (1998) compared measured reactivity from slowly versus rapidly heated coal. The heating rate has a large impact on the measured reactivity, with a higher heating rate producing a more reactive char. Thus, the reported reactivity data should be used only with an awareness of how the reactivity was measured. In spite of the complexity of quantifying reactivity, it is generally true that reactivity is greater for lower-rank coals and biomass than for higher-rank coals:

high reactivity

low reactivity

biomaass, lignites > bituminous > anthracite This particular ordering has practical implications. A relatively unreactive feedstock is difficult to use in a low-temperature gasifier because the carbon conversion is very slow, reducing the gasifier throughput. This is why low-temperature fluid bed systems are usually not considered for anthracite gasification, whereas biomass is very well suited. Petcoke, which is a by-product of petroleum refining, is not very reactive relative to fuels having a greater volatile content. Bryers (1995) discussed different types of petcokes, and presented a table of typical proximate and ultimate analyses, along with ash properties. The reactivity data from TGA show that the reactivity is at the low end of the bituminous range. Salvador et al. (2003) suggested that the reactivity of petcoke is enhanced by catalytic effects of vanadium, which is usually present in petcoke.

1.2.2 Ash Composition and Properties The composition of the ash influences how the ash can be handled during gas­ ification and subsequent gas processing. For example, in fluid or fixed-bed gasifiers, the feedstock is converted in a reacting bed that is designed to handle the feedstock as a dry solid. The bed temperature must be high enough to convert the carbon, but avoid melting the ash. In contrast, high-temperature “slagging” gasifiers are designed to operate at temperatures where the ash is molten, and can flow as a liquid slag. In this approach, the slag viscosity becomes an important factor, and the gasification temperature must be high enough to insure proper slag flow. The ash properties are most commonly reported with several temperatures that identify how a specified sample behaves following ASTM D1857. As the sample is heated, the sequence of temperatures are reported with the initial deformation, softening, “hemi­spherical” (that is, forming a hemisphere), and fluid states. For slagging, the temperature where the viscosity is less than 250 Poise is also useful. This is

Gasification Technology to Produce Synthesis Gas

7

accepted as the minimum viscosity to allow slag to flow (Babcock and Wilcox Co., 2005). It is possible to use limestone to “flux” the slag, lowering the temperature needed for adequate slag flow. This approach allows using feedstocks with high melting temperature ash, but at a reduced gasification temperature. Aside from the viscosity and melting temperature, the ash chemical composition is an important consideration. Biomass ash is typically high in alkali, notably potassium, which can attack the refractory used to line some high-temperature gasifiers. This can limit the quantity and type of biomass that can be used in refractory-lined gasifiers, and again emphasizes the need to consider carefully the properties of the fuel and ash when evaluating an application in a specific gasifier. Ash and refractory interactions are discussed in more detail by Bennett et al. (2007, 2008). Table 1.3 lists initial deformation temperatures for the feedstocks under reducing conditions (e.g., for gasification). It is important to emphasize that the characteristic ash temperatures may be considerably different between oxidizing and reducing conditions, sometimes by more than a 100°C. This is because of the different melting characteristics of oxides versus their parent (unreduced) compounds. As a result, ash behavior studied in combustion applications may not be relevant to gasification situations. The characteristic temperatures are strongly influenced by composition, and care must be taken to fully characterize the behavior of ash properties associated with specific fuels. It is not possible to generalize any trend in ash behavior between biomass, coal, or petcoke from Table 1.3, except to note that the ash initial deformation temperatures cover a wide range.

1.2.3 Feedstock Preparation The gasification behavior can be influenced by how the feed is prepared for gasification. Relatively unreactive feedstock could (in principle) be ground fine enough to operate in lower-temperature systems, but the finer particles may not be compatible with fluid bed systems, for example. Likewise, the gasification and feeding behavior may be influenced by drying the feedstock, so that feed preparation should be considered when evaluating how to use a given feedstock. As discussed later, some gasifiers will not operate with coal fines and require removing the fines from the feedstock. Feedstock preparation includes size reduction and drying as needed to meet the gasifier requirements. Drying the coal with low-grade waste heat avoids diluting the high-grade heat from exothermic reactions needed to sustain the endothermic gasification reactions. Drying is also helpful to ensure good feed handling. Takematsu and Maude (1991) review various studies showing the efficiency advantage of drying coal prior to integrated gasification combined cycle (IGCC) applications. In one case, the net IGCC power plant efficiency was projected to rise by more than 1.5% when using dry coal feed versus adding water to create a coal slurry for easy coal feeding (discussed below). Takematsu and Maude (1991) also report the change in efficiency to create liquid fuel from coal when using coal dried to 2% moisture versus as received (16.5% moisture). Even though some coal energy was spent drying the coal, the conversion efficiency increased with drying. Adding water to create a coal slurry for feeding produced a significant loss in efficiency. The details of this analysis were presented by Vogt et al. (1984).

8

Synthesis Gas Combustion: Fundamentals and Applications

Size reduction requirements can range from 50 mm to finely ground particles (~100 microns) for different gasifiers, described in Section 1.3. After proper sizing, the feedstock must be supplied at a controlled rate and conveyed into the gasifier. This can be very challenging for high-pressure gasifiers. Lockhopper arrangements can be used to batch-feed the coal into a high-pressure feed system, but this introduces an awkward cyclic valve arrangement into an otherwise steady-state process. These valve systems require significant maintenance as well as purge and conveyance gas (typically nitrogen) to operate. The conveyance gas is not insignificant because at high pressure, gas density and flow required for dilute-phase solids transport will compromise the syngas purity. Where pulverized coal is used, steady feed coal “pumps” have recently been developed (Aldred and Saunders, 2003, 2005) that are expected to simplify the process of feeding dry solids. Alternatively, the coal can be suspended in water slurry having as much as ~70% solids loading by mass. This allows easy pumping of the coal, but the injected water adds a significant energy penalty, discussed above. Because of growing interest in biomass as a feedstock, a few points are discussed relative just to biomass. At the present writing, there is no standard method to prepare and feed biomass for pressurized gasifiers. Biomass preparation techniques for atmospheric pressure boilers are still developing (Werthera et  al., 2000; Zulfiqar et  al., 2006), such that pressurized feeding for gasification presents added challenges. Wood biomass (e.g., trees) requires considerable grinding energy to reduce the wood to small particles (submillimeter), which are required in entrained gasifiers. Likewise, fibrous biomass (such as switchgrass) is very difficult to shred and grind into easily fed particles. For high-pressure entrained flow gasifiers, Bergman et al. (2005) evaluate the use of a low-temperature (200 to 300°C) heat treatment called torrefaction to simplify grinding the biomass. Without torrefaction, the electrical power required to mill woody biomass to 0.2 mm size is approximately 12 to 16% of the electrical energy produced by the biomass (assuming a 40% efficient biomass to electricity). With torrefaction, this grinding energy could drop to as low as 1.5%, but this is offset by the need to supply heat, albeit at low temperature. Further study is recommended on optimizing the methods to prepare and feed biomass and coal mixtures in high-pressure gasifiers. It is interesting to note that in some laboratory studies of mixed coal/biomass gasification (McClendon et al., 2004) the addition of fine sawdust actually improved the feeding reliability for coal injection. Thus, there may be a beneficial synergy when combining coal and particular forms of biomass, as expected from fundamental studies (Zulfiqar et al., 2006).

1.3 Gasifiers Detailed descriptions of gasifier types can be found in various references (Probstein and Hicks, 1990; Simbeck et al., 1993). Operating data and novel gasifiers studied prior to the 1980s are found in some early reports (Handbook of Gasifiers and Gas Treatment Systems, 1976; Hendrickson, 1975). Novel approaches to gasification such as catalytic, molten salt, plasma, or indirectly heated systems are described elsewhere (Handbook of Gasifiers and Gas Treatment Systems, 1976;

9

Gasification Technology to Produce Synthesis Gas

Hendrickson, 1975; Heinemann and Somorjai, 1994; Dighe and Lazzara, 2002; Paisley and Overend, 2002). In this chapter, only the major gasifier configurations in use today are described. The main gasifier types are moving bed, fluid bed, and entrained flow systems.

1.3.1 Moving Bed Gasification Figure 1.2 shows the general layout of the moving bed gasifier. The gasifier can be arranged in updraft (countercurrent) and downdraft (co-current) arrangements. In the downdraft arrangement, coal and oxidants enter together, so that the highest temperatures occur at the exit, producing a relatively high-temperature exhaust stream (~700°C) (Reed and Gauer, 2001). This is shown schematically by the temperature profile in Figure 1.2. The high temperatures help complete gasification reactions, so that relatively little tar exists in the products. This type of gasifier has been used extensively in small-scale applications, including vehicle applications during World War II (Reed and Gauer, 2001). Compared to the counterflow arrangement, the disadvantage Coal

Product gas Updraft Bed

Temp Oxygen or air

Oxygen or air

Ash Coal

Steam

Steam

Downdraft Bed Product gas Ash

Figure 1.2  Moving bed gasifiers.

Temp

10

Synthesis Gas Combustion: Fundamentals and Applications

of the co-flow arrangement is that the higher sensible energy of the product gas must be integrated in downstream processes to achieve comparable overall efficiency. For the updraft (counterflow) arrangement, coal enters the reactor at the top through a lockhopper arrangement. The coal typically enters via a rotating distributor in a bed that is supplied with oxygen and steam from the bottom. The counterflow arrangement allows high-temperature combustion reactions to consume the coal char at the bottom of the bed. Hot products flow upward to gasify the coal and volatiles in the upper region of the bed. Thus, as shown in Figure 1.2, the temperature profile has a bulge in the middle of the bed. Ash drops out of the bottom of the bed to an ash handling system. The advantage of this counterflow arrangement is that the gases exiting the bed are cooled by the incoming fuel, so that the gas exit temperature is typically between 425 and 650°C, but may be even lower (250 to 500°C), depending on the coal type (Rudolph, 1984). This makes efficient use of the thermal energy released by oxidizing some of the solid carbon. However, because of these low temperatures, coal tar and some oxygenated compounds are formed in the syngas. These compounds can represent a significant fraction of the input fuel energy (Higman and van der Burgt, 2003) and should be used in the process downstream for efficient operation. The moving bed design can be used with multiple coal types. However, care must be taken when using “swelling” coals, which can essentially plug the bed. Coal swelling can be characterized on scales defined, for example, in ASTM D720-91. Where high-swelling coals are used, the distributor can be arranged to stir the bed (see Rudolph, 1984). For moving bed operation, coal fines (less than 3mm diameter coal) should be avoided to prevent the bed from agglomerating. The necessity to avoid fines adds process complexity to the plant configuration. In existing designs, the coal fines can be separated from the fuel stream, and supplied to an auxiliary boiler to raise steam, or some of the fines can be bound into “briquettes” using coal tars as a binder (Higman and van der Burgt, 2003). The counterflow moving bed configuration is the type of gasifier widely used to produce synthesis gas for the Fischer-Tropsch process for creating liquid fuel from synthesis gas (Gasification World Database, 2007). A slagging version of the moving bed has been developed, and operates in the same manner as described above, except that steam injection is reduced. This allows the bed peak temperatures to rise high enough to melt the ash and produces an easily handled slag product (Higman and van der Burgt, 2003).

1.3.2 Fluid Bed Gasification Within the category of fluid bed gasifiers, there are subcategories distinguished by the bulk velocity of gas moving in the bed. Each of these processes is described below and shown schematically in Figure 1.3. The higher gas velocities of circulating and transport gasifiers produce greater heat transfer to the fuel particles. This increases the heating rate, and as discussed in Section 1.2, rapid heating produces a more reactive char (Megaritis et al., 1998), and is favored for carbon conversion to syngas. The different types of fluid bed gasifiers are described in the following subsections referring to Figure 1.3.

11

Gasification Technology to Produce Synthesis Gas Coal

Product gas

Bubbling

Steam

Ash

Oxygen or air Product gas

Coal

Entrained solids

Cyclone

Circulating

Steam

Ash

Oxygen or air Product gas

Transport Coal

Steam

Oxygen or air

Figure 1.3  Fluid bed gasifiers.

1.3.2.1 Bubbling Bed At relatively low velocity (less than 5 m/s), the bed is characterized by discrete bubbles of gas, much like a boiling fluid. The region above the bed (freeboard) is clearly distinct from the bed. The fluid motion of the bed approximates a continuously stirred reactor and provides a homogenous temperature environment for coal reaction. The bed itself is mostly ash, and the intent is that the carbon leaves the bed as syngas, although some ash with carbon and unburned fines are conveyed up into the freeboard region. A key to operating is avoiding temperature conditions where the ash will soften and form agglomerates. This means that the bed temperature must be maintained well below the ash fusion temperature, which may range from 950 to 1100°C for coal, and

12

Synthesis Gas Combustion: Fundamentals and Applications

can be even lower for some types of biomass ash. Thus, control of the bed conditions and knowledge of the ash properties are design considerations for these systems. The relatively low gasification temperature requires less oxygen than for entrained flow gasifiers. However, the stirred-reactor mixing ensures that some of the extracted ash will contain unburned carbon. Higman and van der Burgt (2003) state that the best atmospheric pressure fluid bed gasifiers achieve only 97% carbon conversion, but that pressurized biomass fluid bed gasifiers can achieve 99% efficiency. Because fluid bed gasifiers operate at low temperature, and the bed is made of ash, they can handle fuels with high reactivity and high levels of ash content—which describes many biomass fuels and low-rank coals. For less reactive coals, the lower carbon conversion is a disadvantage. In some cases, the ash can contain 20% of the original fuel carbon, and a subsequent boiler is added to burn out the remaining carbon (Higman and van der Burgt, 2003). To improve carbon conversion, and increase throughput, additional oxidant and steam can be added to the freeboard region, raising the temperature of the product stream to provide better conversion of the fines and unburned carbon carried into this region. The higher temperatures also reduce the tars remaining in the syngas. A recycle loop is used to further oxidize unburned carbon. Supp (1990) provides some data on a commercial process using this approach, and notes that ash leaving the bottom of the bed can still carry 4% of the carbon input and may be sent to an auxiliary boiler. There are numerous small-scale commercial vendors of bubbling fluid bed gasifiers. A report by Ciferno and Marano (2002) considered the range of possible gasifiers for biomass applications, and noted that there were multiple bubbling fluid bed gasifiers being used for biomass applications. From the report, bubbling fluid bed reactors have been developed or demonstrated over the following range of conditions: • 4.5 to 181 metric tons per day input of biomass • Both air and oxygen input • Pressures from 1 to 35 bar(s) 1.3.2.2 Circulating Fluid Bed Compared to the bubbling bed approach, the circulating fluid bed can provide greater carbon conversion and reduced tar formation. As shown in Figure  1.3, raising the fluid­ization gas velocity (typically 5 to 8 m/s) entrains smaller particles, which are converted above the bed or separated in a cyclone for return to the bed (that is, circulat­ing bed). The larger coal particles remain in the bed where they are oxidized. The higher velocity increases the coal heating rate, which lowers tar production. The higher gas velocity can accommodate greater throughput than a bubbling bed system. Circulating fluid bed (CFB) systems are being used for multiple biomass applications. In many installations, CFB gasifiers are atmospheric pressure and use air as an oxidant. This approach is relatively simple, and where the fuel is used for process heat (combustion) applications, it is very effective. Commercial literature indicates

Gasification Technology to Produce Synthesis Gas

13

multiple applications of CFB gasifiers to provide fuel gas to ambient pressure combustors used in lime kilns, or boilers (Palonen et al., 2005). A pressurized version of a CFB gasifier has been used for a biomass-fueled IGCC demonstration in Värnamo, Sweden. The concept was technically successful, but was not profitable enough for continued operation in the year 2000 (see Reed and Gauer, 2001, or more recently, the Chrisgas project, http://www.chrisgas.com). CFB gasifiers have been evaluated as a source of biomass syngas for liquid fuel production by Boerrigter et al. (2003). These authors suggest (but do not quantitatively demonstrate) that biomass syngas plants will likely be plants with pressurized, oxygen-blown gasifiers. The relative merits of pressurized (versus atmospheric) gasifiers for biomass applications are discussed by Bridgewater (1995). 1.3.2.3 Transport Reactor At sufficiently high gas velocity (~15 m/s), all of the bed material can be deliberately transported up the reactor by the gas flow (Figure 1.3). The particulate matter is separated from the gas and then collects in a “standpipe” before being entrained again in the riser reactor for another cycle around the reactor. The circulating solid, including the ash, provides useful thermal ballast that recycles heat back into the incoming reaction stream. Compared to circulating fluid bed, the transport reactor has yet again greater throughput, and the heating rates for coal particles are higher than in other fluid beds. The rapid coal particle heating serves to rapidly evolve the coal volatile matter, reducing the tendency to crack the volatiles and form tars. For coal applications, a development reactor has been built at Wilsonville, Alabama (Leonard, 2007) and has demonstrated gasification of low-rank coal at up to 2500 kg/h (Nelson et al., 2003). Most experience to date has been on low-rank coals, but less reactive bituminous coal has also been gasified with a lower carbon conversion (Wallace et al., 2006).

1.3.3 Entrained Flow, Slagging Gasifiers Unlike moving bed or fluid bed gasifiers, entrained flow gasifiers are designed to operate at temperatures high enough to melt the coal ash, usually above 1250°C. This is achieved by using more oxygen to achieve higher temperatures in the gasifier. Compared to other approaches, this uses more of the fuel energy for gasification, and lowers the overall efficiency of converting the feedstock to syngas. The lower efficiency is offset by some significant advantages. The higher temperatures usually ensure that the product gas is free of tars, and the carbon conversion is very high. Virtually any type of solid fuel can be used in these gasifiers, as long as it is properly ground for the feed system, usually less than 100 microns in size. In some designs, the slag is easily disposed, or ground to saleable product (Amick and Dowd, 2001; Geertsema et al., 2002). These advantages explain why many new IGGC plants have been proposed to use entrained flow systems. Figure 1.4 shows simple schematics of different types of entrained flow gasifiers. In the single-stage system at the top, all the coal and oxidant enter at one end of the

14

Synthesis Gas Combustion: Fundamentals and Applications Oxygen or air

Coal

Steam Product gas

Downflow

Single Stage

Upflow

Coal Steam Oxygen or air

Product gas Slag Oxygen or air

Coal 1st stage

Slag Steam Product gas

Downflow Two Stage

Upflow Coal 2nd stage

Coal 2nd stage

Coal 1st stage Steam Oxygen or air

Product gas Sl ag

Slag

Figure 1.4  Entrained flow gasifiers.

gasifier, and the heat released by combustion serves to gasify the coal. The flow can be arranged up or down, with the slag flowing out the bottom. Sufficient residence time is needed to completely oxidize the coal carbon to CO, typically requiring less than 10 s residence time. In the lower part of Figure 1.4, the two-stage gasifier uses the products of the first gasification zone to gasify coal injected in the second stage. Again, the process can be arranged with up- or downflow configuration. Endothermic gasification reactions in the second stage serve to lower the exit temperature compared to a single-stage design. The result is a lower oxygen demand per mass of coal, and a higher efficiency conversion to syngas fuel. If the second stage is operated with excess coal, some tars and hydrocarbons can be formed, but this can be avoided with proper design and operation.

Gasification Technology to Produce Synthesis Gas

15

1.3.4 Comparison of Gasifier Types and Approaches The discussion thus far has not specifically described the trade-offs between oxygenblown and air-blown gasification, or between high-pressure and ambient-pressure gasification. In principle, the gasifiers described above could operate with air or oxygen and at any pressure level. In practice, the choice to use oxygen, or high pressure, depends on the overall plant economics and energy needs. Pressurized operation increases the throughput for a given size reactor, and is especially favored when the syngas is to be used in a pressurized application such as a gas turbine, or highpressure chemical synthesis. Likewise, oxygen gasification avoids diluting the syngas with nitrogen from air, making it suitable for many chemical applications that require undiluted syngas. However, oxygen generation typically requires considerable energy—1.3 MJ/kg oxygen is a typical value of the electrical energy required by a cryogenic oxygen plant (from Simbeck et al., 1993) for oxygen supplied at pressure. This energy requirement may be 5 to 7% of the electrical power produced by an IGCC power plant (Higman and van  der Burgt, 2003) or even greater (Jaeger, 2007). For some applications, the cost of the oxygen supply may not be justified and air-blown gasification can be used. For example, a recent IGCC plant in Japan uses a two-stage air-blown gasifier with a very small air separation unit to provide small amounts of nitrogen for conveying the feedstock and slightly enriching the gasifier oxygen level (“Mitsubishi,” 2007). Table 1.4 summarizes the features of the different gasifiers discussed above. The table listings are typical, but it should be recognized that different fuel types and operating conditions can change these listings. For example, high-moisture feedstocks may operate at temperatures below the range listed if they have adequate reactivity. The throughput and residence times are useful to compare each style of reactor, but the actual fuel throughput is a strong function of the specific feedstock and operating conditions. The chart shows the advantage of the higher-temperature gasifier: the throughput is much greater for a given reactor cross section, and essentially any fuel that is suitably ground can be used. As already explained, the dis­ advantage of the higher temperature is that the sensible energy in the product gas needs to be recovered for efficient fuel use, and this is discussed in Section 1.3.5. Table 1.4 lists some cold gas efficiencies. The cold gas efficiency is the ratio of fuel heat content to the syngas heat content at ambient conditions and is a measure of how efficiently fuel energy is converted to syngas energy. For the cases reported, the moving bed is most efficient but has the disadvantage that some of the syngas energy is produced in tars. The listed efficiencies (~85%) support the rule of thumb that approximately 15% of the feedstock heating value is used to convert the feedstock to syngas. The oxygen requirement in Table 1.4 applies to oxygen-blown gasification only and is expressed as the ratio of oxygen to moisture and ash-free coal. The advantage of low-temperature gasification is evident—the oxygen demand for the moving bed design is slightly more than half that of the entrained flow gasifier. Given the high

16

Synthesis Gas Combustion: Fundamentals and Applications

Table 1.4 Comparison of Gasifier Types Moving Bed

Fluid Bed

Entrained Flow

References

Feedstock size

5–50 mm

<5 mm

<100 microns

Preferred feedstock type

Low rank, limited fines, swelling coals must be stirred 425–650 C

Low rank

Any feedstock

Simbeck et al., 1993; Higman and van der Burgt, 2003 Simbeck et al., 1993; Higman and van der Burgt, 2003

900–1,050 C

1,250–1,600 C

100–160

240–320

320–450

60–120 min 84% (dry ash) 88% (slagging) 0.43 (dry ash) 0.53 (slagging)

20–30 min 80.8

<10 s 81

0.71

0.81

Gasifier outlet temperature Throughput (metric tons/day/m2) Residence time Cold gas efficiency (high-rank coal) Oxygen requirement (kg/kg maf coal)

Simbeck et al., 1993; Higman and van der Burgt, 2003 Takematsu and Maude, 1991 Riemert, 1989 Takematsu and Maude, 1991 Takematsu and Maude, 1991

Note: The numbers are typical from the cited sources, but may differ with specific installations and feedstocks.

energy demand for making oxygen, that can be an advantage, but is somewhat offset by lower throughput, tar production, and feedstock limits for moving bed designs.

1.3.5 Syngas Thermal Management While higher gasification temperatures are advantageous to avoid tars and enhance carbon conversion, the product synthesis gas contains appreciable sensible thermal energy that should be used for efficient fuel conversion. Mahagaokar and Doering (1995) discuss the various approaches to recover this energy, typically about 15% of the fuel heating value for entrained flow gasifiers. Figure 1.5 shows three approaches that are used. These methods are typically considered for entrained flow gasifiers, where the higher gasification temperature justifies heat recovery. At the top, water quench can be used to cool the syngas in one step to approximately 250°C. This approach is very effective for cooling, and where water needs to be added to the process (for downstream water gas shift), quenching is not a significant drawback. The quench also serves to freeze molten slag and ash in the water, reducing the solid filtration load. However, the quench does degrade the sensible energy availability that could otherwise be used to raise steam for power cycles such as IGCC. To recover more energy, the gas quench approach (middle, Figure 1.5) recycles cooled syngas to lower the temperature enough so that convection coolers can be used to raise steam. Enough gas is recycled to cool the syngas below the temperature

17

Gasification Technology to Produce Synthesis Gas

Water Quench Coal and O2 Gasifier

Solid filtration

Water ~250°C Slag

Coal and O2

Partial Gas Quench Recycle blower Solid filtration

Gasifier

~250°C

H2O (liq) Steam

~900°C Radiant Syngas Cooler Coal and O2 Solid filtration

Gasifier

~250°C Radiant boiler

H2O (liq)

H2O (liq) Steam

Steam ~90°C

Figure 1.5  Approaches to thermal integration. Temperatures are approximate.

18

Synthesis Gas Combustion: Fundamentals and Applications

where ash adhesion could foul the heat exchangers, usually below 900°C, depending on the ash. Finally, at the bottom of Figure 1.5, a radiant boiler can be used to directly cool the gasifier products from the gasification temperature. This has the potential to raise steam from the syngas at very high temperature, and is currently practiced in one IGCC plant. The radiant cooler measures an impressive 30 m in length and weighs 815 metric tons (Tampa Electric Integrated Gasification Combined-Cycle Project, 2000). The size and complexity of this type of boiler are obviously a trade-off with the value of recovering high-grade thermal energy. The approach selected to manage the syngas thermal energy depends on specifics of the syngas application. For example, when the syngas is used to make chemicals, the benefit of adding syngas coolers may not offset the penalty of size, cost, and complexity. For example, the Fischer-Tropsch process to create liquid fuels from syngas is carried out at 200 to 250°C, so that direct quench to saturation at these conditions can provide water needed for the process, and cools the gas in a straightforward manner.

1.4 Syngas Purification Syngas must usually be cleaned to meet requirements for downstream applications or environmental emission standards. Table 1.5 summarizes the major species that must be considered for cleanup and indicates why some cleanup requirements are established for different applications or environmental reasons. The table does not include quantitative standards for every species, because the actual requirements can differ considerably and must be evaluated in the application of interest. Where the syngas is used in combustion applications, impurities are removed to meet requirements limiting atmospheric emissions such as SOX or NOX. In this instance, the required syngas purity can be estimated from combustion material balances, and making some assumptions about how the syngas species are oxidized in the combustor.* As indicated in the table, the sulfur removal requirement may also be linked indirectly to the NOX emissions where selective catalytic reduction (SCR) is used to scrub the flue gas. This is discussed in further detail in Chapter 2. For particulate removal, it is usually the case that current gas purification techniques control particulate to levels well below any environmental standard because the downstream processes simply cannot tolerate significant levels of fine ash accumulation over extended operation. Where the syngas is used for chemical or fuel production, syngas purity may be dictated by catalyst requirements, which can be very stringent. For example, Higman and van  der Burgt (2003) report that catalysts used to produce liquid fuel by the Fischer-Tropsch (FT) process require syngas with a combined HCl + HBr + HF content of less than 10 parts per billion. This level of gas cleaning is achieved in current practice for FT production, so it is not a deterrent to syngas use, but requires different purification steps than might be needed in power production applications. *

For estimates, it is usually adequate to assume that the impurities like H2S and NH3 are converted directly to SO2 and NOX stack emissions.

19

Gasification Technology to Produce Synthesis Gas

Table 1.5 Syngas Cleanup Requirements Impurity

Turbines

Chemical Production

Particulate

Typical: <1 ppmw in fuel, wide range possible (Stringer, 1989; Wenglarz et al., 1986; Bossart et al., 1990; GE Power Systems, 1999).

Total Sulfur

Environmental, or may be set by SCR catalysts (~10 ppmv H2S fuel gas) (DeBiasi, 2005).

Total Nitrogen Compounds Trace Impurities Halides (HCl, HF)

Alkali (Na, K)

Environmental

Environmental Process requirements often exceed environmental; typical stack range (see notes) PM10 < 0.015 lb/106 BTU (Ratafia-Brown et al., 2002)

See notes for molten carbonate and solid oxide fuel cells below.

Turbines: Stack SO2 range 0.03–0.2 lb/106 BTU, see Ratafia-Brown et al., 2002. Stack NOX limit, typical range 2 ppm–25 ppm.

Set by material corrosion limits; typical range 0.4–0.6 ppm (Takematsu and Maude, 1991). Vendor specs, typical range 0.02–1 ppmw fuel (GE Power Systems, 1999).

Metals (Hg, Cd, Se, V, and others)

Environmental and vendor specs.

Carbonyls Fe(CO)5, Ni(CO)4

Produce in-situ; controlled to avoid turbine material deposits.

Carbon Dioxide

Catalyst and process specific

Fuel Cells

Environmental

Recent IGCC permits: Hg stack emission ~0.5 lb/1012BTU (WePower SCPC and IGCC Information, 2003).

No standards to date.

Fuel cells: For molten carbonate fuel cells (MCFCs), earlier studies and requirements are summarized by Thambimuthu (1993). For solid oxide fuel cells (SOFCs), recent studies have established the required purity for multiple trace species; see papers by Gemmen and Trembly (2006) and Trembly et al. (2007a, 2007b, 2007c). PM10: The particulate matter smaller than 10 microns in diameter.

The common methods of gas cleanup are described next. For virtually all operating syngas plants, gas purification is accomplished with the syngas cooled to ambient temperature, usually referred to as cold gas cleanup. The advantage of cold gas cleanup is that virtually all of the syngas impurities can be removed with available technology operating near (or below) ambient temperature. The disadvantage of cooling is an exergy loss. Both the sensible thermal energy and the latent heat of vaporization (for syngas H2O) are degraded by cooling, especially where significant quantities of syngas steam are condensed during cooling. This is the motivation to develop warm or humid gas cleanup, discussed in Section 1.4.2.

20

Synthesis Gas Combustion: Fundamentals and Applications

1.4.1 Cold Gas Cleanup Figure 1.6 shows the major process steps involved in syngas purification using conventional cold gas cleanup. The figure shows the major process steps in most cleanup systems and gives some approximate temperatures at each step. It is emphasized that there are many possible variants in process arrangement; this is a generalization. Following the schematic in Figure 1.6, the syngas temperature is reduced via process steps described in Section 1.3.5. If quench is used to cool the syngas, much of the slag and particulate will be caught in the water quench. The remaining fly ash can be separated using cyclones or filters, and at the temperatures shown (400°C), most of the alkali materials will condense on the ash (Thambimuthu, 1993). Depending on the gasifier, the separated ash may contain enough carbon that it is worthwhile to recycle it to the gasifier where carbon conversion is completed, and the remaining mineral matter then exits with the slag or ash removal. Following the particulate removal, a subsequent water wash is used to capture the finest particulates and the majority of water-soluble impurities. The bulk of the fuel ammonia, HCN, and HCl will be captured by the scrubbing water, but further purification may occur in downstream water knockout or in the acid gas scrubbers (Section 1.4.1.1). After the wet scrubbing, it may be necessary to “shift” the COS to H2S because some of the sulfur removal schemes are not selective to COS:

COS + H2O → CO2 + H2S

COS hydrolysis

COS can also react with hydrogen in the syngas stream, producing CO and H2S; see Supp (1990) for details about where this reaction is dominant. Where more hydrogen is sought in the syngas, the water gas shift can be conducted at this point:

CO + H2O → CO2 + H2

Water gas shift

This important reaction is a key element in future scenarios where CO2 is removed from the syngas for the purpose of carbon dioxide sequestration. By using water gas shift and CO2 scrubbing (Section 1.4.1.1), a carbon-containing feedstock can be converted to hydrogen fuel. The details of the water gas shift are more involved than shown schematically in Figure  1.6, and can include catalysts suited to both hightemperature (~400°C) and low-temperature operation (~250°C). The shift reaction can occur in multiple stages, with intermediate heat removal; more details can be found in Twigg (1996). In Figure 1.6, the water gas shift is shown ahead of the sulfur removal (sour shift), but it can also be arranged after the sulfur removal (sweet shift), although the latter requires reheating the gas. Continuing to follow the schematic, the temperature is further reduced so that significant water condensation occurs (knockout). The condensing water serves to capture even more water-soluble trace species like ammonia. As shown, the gas can now be sent to a carbon bed where mercury is removed. A description of these beds is given by Klett et al. (2002). The bed can operate for ­multiple months, capturing trace metals before the bed material needs to be exchanged. As an alternative to Figure 1.6, Klett et al. noted that the bed may be placed downstream

Recycled to consume residual carbon

Temperature reduction Possible via: - Heat exchanger - Gas quenching - Water quenching

Particulate flyash

Cyclone separation or filters

400°C

Scrubbing water

Shift steam

COS hydrolysis and/or water gas shift

200–230°C

To water treatment and recycle

Removes in water - Finest particulate - Nitrogen compounds (NH3, HCN) - Halogen (HC2, HF, not 100%) - Alkali

Wet scrubbing

Figure 1.6  Schematic of conventional cold gas cleanup system.

Slag

Gasifier

Gasifier exit temperature

Water knockouts

Mercury

20°C

CO2 H2S

Acid gas scrubber

Typical range (Specifics may vary)

Clean syngas

Gasification Technology to Produce Synthesis Gas 21

22

Synthesis Gas Combustion: Fundamentals and Applications

of the acid gas removal, but the upstream location keeps some impurities out of the acid gas wash, described next. 1.4.1.1 Acid Gas Scrubbing The final cleanup step shown in Figure  1.6 is the acid gas purification (CO2 and H2S). Because both H2S and CO2 will form an acid when contacting water, they are referred to as acid gases, and the techniques to remove them are called acid gas removal. Multiple choices exist for liquid solvents that can remove these species, with distinctions being the level of sulfur removal versus CO2, the operating temperature, and the method of solvent regeneration. Thambimuthu (1993) lists more than 15 commercial processes that could be used to treat coal syngas. With this wide number of options, the advantages for selecting a process depend on the details of the specific plant requirements. Design data and options for these process steps can be found in various papers and reference texts (see Shah and McFarland, 1988; Kohl and Nielson, 1997; and Kriebel et al., 1989). Among these different processes, there are three major classes of solvents: chemical, physical, and mixed, described below. The trade-offs among the various processes and general considerations for selecting a process are discussed by Korens et al. (2002). 1.4.1.1.1 Chemical Solvents These solvents form a reversible chemical bond with the target acid gas. They are regenerated by heating the solvent to a sufficient temperature to reverse the chemical reaction and release the absorbed gas (known as thermal swing regeneration). The most common chemical solvents for H2S (and also effective on CO2) are based on amine compounds dissolved in water. Acid gases are absorbed in a column contactor, and then the “rich” solvent is regenerated with heat that reverses the absorption process. After regeneration, the “lean” solvent is cooled and recycled to the absorber. The energy required to regenerate the solvent is a key parameter in evaluating the performance of the acid gas removal system; typical regeneration energies are reported in Kohl and Nielsen (1997). 1.4.1.1.2 Physical Solvents In contrast to chemical solvents, the physical solvents capture the acid gas by physical absorption. There is no chemical bond; the gas species j at partial pressure Pj is simply dissolved in the liquid to mole fraction xj following Henry’s law:

xj = Pj /Hj   Henry’s law

In this expression, Hj is the Henry’s law coefficient for the species of interest. The solvents are devised to have greater solubility (large 1/Hj) for the captured acid gas, while ideally excluding desired syngas species (H2, CO). The published solubilities of H2S and CO2 are orders of magnitude greater than CO or H2, such that good separation of these gases is readily achieved, with very little syngas lost into the solvent (Kohl and Nielsen, 1997). A complication arises in the solvent regeneration if CO2 and H2S are desired in separate streams. Depending on the solvent, sequential absorption stages can be designed to capitalize on differences in solubility, as

Gasification Technology to Produce Synthesis Gas

23

well as absorption rates, to produce the desired separation. One species of particular interest is COS, which has a solubility between CO2 and H2S, making it difficult to exclude from the CO2 stream. As discussed above, this is why COS hydrolysis is used upstream of the acid gas solvent. Because the physical solvents operate using the gas partial pressure, regeneration is achieved with straightforward depressurization. No additional thermal regeneration energy is necessary, and this can be an advantage compared to chemical solvents. However, depending on the solvent loading, and required circulation rates, the advantages for physical versus chemical solvents are a function of specific installation requirements. One widely used physical solvent is chilled methanol. At temperatures operating between –70 and –30°C, methanol will absorb both H2S and CO2, as well as many other coal gas impurities to very low levels. This explains why chilled methanol is usually used to purify syngas for chemical processes. While the solvent is effective, there is an obvious complication and penalty associated with cooling the syngas to subzero temperature. Another common physical solvent is based on a mixture of dimethyl ethers of polyethylene glycol. This type of solvent can be used at approximately ambient temperatures, avoiding the need to chill the syngas. However, the solvent does not achieve the same level of purification as chilled methanol, and is therefore applicable where modest impurities can be tolerated in the syngas (e.g., combustion applications). 1.4.1.1.3 Mixed Chemical/Physical Solvents Physical and chemical solvents each have advantages and disadvantages in terms of required circulation rates, separation performance, regeneration energy, and process complexity. This has been the motivation for the development of solvents that combine features of both chemical and physical absorption, combining, for example, amines with methanol to raise the temperature of absorption compared to chilled methanol. 1.4.1.2 Tars and Carbonyls Figure 1.6 has shown some of the main process steps in conventional gas cleanup. There are different variations possible depending on the gasification approach. For example, because of their low gasifier exit temperature, moving bed gasifiers will produce appreciable tars. These tars must be recycled or used, and this is done with a solvent wash stream (not shown in Figure 1.6). Details can be found in Rudolph (1984) or Supp (1990). Table 1.5 also lists requirements to avoid carbonyls. Carbonyls are not coal impurities; they can form in the low-temperature section of the syngas processing and can be avoided by process design. Higman and van Burgt (2003) describe the process conditions that must be avoided to prevent carbonyl formation.

1.4.2 Warm Gas Cleanup To maximize the use of syngas thermal energy in power cycles, it is very desirable to conduct the gas purification at high temperatures, usually referred to as warm

24

Synthesis Gas Combustion: Fundamentals and Applications

gas cleanup or humid gas cleanup (because the syngas moisture is not condensed). Some studies suggest that the efficiency of an integrated gasification combined cycle power plant can be increased more than 3% when comparing cold gas to warm gas cleanup (Schlather and Turk, 2007). With the exception of particulate removal, warm gas cleanup is in the research and development stage. Korens et al. (2002) assessed the status of warm gas cleanup in 2002 and noted that particulate control methods for warm gas conditions have been successfully developed, and sulfur removal schemes also appeared promising. Since that time, laboratory tests have demonstrated techniques to remove other impurities (Siriwardane et al., 2007b; Poulston et al., 2007; Granite et  al., 2006), but these are not yet being used commercially. A relatively recent consideration is how to remove CO2 under warm gas conditions. Laboratory studies have evaluated ionic liquid solvents (Heintz et  al., 2005), CO2 membranes (Ilconich et al., 2007), and solid sorbents (Siriwardane et al., 2007a) for the purpose of carbon dioxide removal at warm gas conditions.

1.5 Conclusions This chapter has described the basic concepts for producing clean synthesis gas as a fuel for power production or a precursor for fuel or chemical production. Syngas produced by solid fuel gasification is expected to play an increasingly important role in the future to make greater use of multiple fuel sources and permit direct removal of CO2 from fuel streams. Many different types of solid fuels can be gasified, but key differences in fuel and ash properties must be considered when matching a gasifier to a fuel type and application. Lower-temperature gasification schemes are best suited to fuels with significant ash content because chemical energy must be spent to heat the ash to the gasifier temperature, penalizing the process efficiency. Likewise, the ash behavior must be matched to the process equipment; for example, fluid bed operating temperatures must be maintained below the ash fusion temperature. The reactivity of the fuel is also a consideration, with less reactive fuels requiring higher temperatures or longer residence time for complete conversion. The different types of gasifiers, with their various attributes, were discussed. Trade-offs among the different types of gasifiers involve the operating temperature, the required coal preparation (size reduction), and methods to utilize the thermal energy released during gasification. The choice to operate the gasifier with oxygen or air depends on the application, but oxygen is preferred, especially for chemical production, and pressurized gasification is favored when the syngas is needed for high-pressure applications. A survey of methods to clean the syngas of trace impurities shows that very clean syngas can be produced using a combination of physical separation (for particulates), water scrubbing, solvents, and carbon beds. While existing cleanup methods are effective, they require cooling the gas to ambient or even lower temperatures, degrading the available thermal energy released during gasification. This has been the motivation to develop gas cleaning techniques that can operate at elevated temperatures. In conclusion, gasification technology can convert a wide range of solid fuels into a common syngas product. The syngas can be used for many power and chemical

Gasification Technology to Produce Synthesis Gas

25

production applications discussed in this book. The syngas itself can be converted to pure hydrogen, with CO2 separated for geological sequestration. The combined attributes of fuel flexibility and separation of CO2 explain the growing appeal of gasification for power and chemical production.

References Aldred, D., and Saunders, T. (2003). Continuous injection of solid fuels into advanced combustion and gasification system pressures. Paper presented at 2003 Gasification Technologies Conference, San Francisco. Available at http://www.gasification.org. Aldred, D. L., and Saunders, T. (2005). Achieving continuous injection of solid fuels into advanced combustion system pressures. Final report, DE-FC26-02NT41439. U.S. Department of Energy Project. Amick, P., and Dowd, R. (2001). Environmental performance of IGCC repowering for conventional coal plants. Paper presented at the 2001 Gasification Technology Conference, San Francisco. Available at http://www.gasification.org/. Babcock and Wilcox Co. (2005). Steam 41:9–11. Bennett, J. (2008). Issues in hydrogen production using gasification. In: Materials for the hydrogen economy, ed. R. H. Jones and G. J. Thomas, chap. 1, pp. 1–36. Boca Raton, FL: CRC Press. Bennett, J. P., Kwong, K.-S., and Powell, C. A. (2007). Issues impacting refractory service life in biomass/waste gasification. In Corrosion 2007, Nashville, Tennessee, p. 10. Houston, TX: NACE International. Bergman, P. C. A., Boersma, A. R., Kiel, J. H. A., Prins, M. J., Ptasinski, K. J., and Janssen, F. J. J. G. (2005). Torrefaction for entrained-flow gasification of biomass. ECN-C-05-067, Energy Research Center of the Netherlands (ECN). Available at www.ecn.nl/biomass. Boerrigter, H., den Uil, H., and Calis, H.-P. (2003). Green diesel from biomass via Fisher-Tropsch synthesis: New insights in gas cleaning and process design. In Pyrolysis and gasification of biomass and waste, ed. A. V. Bridgewater, 371–84. Newbury, UK: CPL Press. Bossart, S. J., Cicero, D. C., Zeh, C. M., and Bedick, R. C. (1990). Gas stream cleanup. DOE/ METC-91/0273, U.S. Department of Energy, Morgantown Energy Technology Center. Bridgewater, A. V. (1995). The technical and economic feasibility of biomass gasification for power generation. Fuel 74:631. Bryers, R. W. (1995). Utilization of petroleum and petroleum coke/coal blends as a means of steam raising. Fuel Proc. Tech. 44:121. Chrisgas: Fuels from biomass. http://www.chrisgas.com. Ciferno, J., and Marano, J. (2002). Benchmarking biomass gasification technologies for fuels, chemicals and hydrogen production. NETL report. Available at www.netl.doe.gov. DeBiasi, V. (2005). SCR poses problems but can be used with IGCC to achieve ultra-low NOX. Gas Turbine World, July–August, p. 28. Dighe, S. V., and Lazzara, D. (2002). Westinghouse plasma gasification technology for application to coal gasification and co-fired coal/biomass gasification. Paper presented at the 19th International Pittsburgh Coal Conference, Pittsburgh, PA. Gasification World Database. (2007). Available at www.netl.doe.gov/technologies/coalpower/ gasification/database/database.html. Geertsema, A., Groppo, J., and Price, C. (2002). Demonstration of beneficiation technology for Texaco gasifier slag. Paper presented at the Gasification Technology Conference, San Francisco. Available at http://www.gasification.org/. Gemmen, R. S., and Trembly, J. (2006). On the mechanisms and behavior of coal syngas transport and reaction within the anode of a solid oxide fuel cell. J. Power Sources 161:1084.

26

Synthesis Gas Combustion: Fundamentals and Applications

GE Power Systems. (1999). Specifications for fuel gases for combustion in heavy duty gas turbines. Report GEI 41040G. Revised January 2002. Granite, E. J., Myers, C. R., King, W. P., Stanko, D. C., and Pennline, H. P. (2006). Sorbents for mercury capture from flue gas with application to gasification. Ind. Eng. Chem. Res. 45:4844. Handbook of gasifiers and gas treatment systems. (1976). Final report, Contract E(49-18)-1772, FE-1772-11, U.S. Energy Research and Development Administration. Heinemann, H., and Somorjai, G. A. (1994). Fundamental and exploratory studies of catalytic steam gasification of carbonaceous materials. Final report, U.S. DOE Contract DE-AC03-76SF00098, LBL-35374, UC109. Heintz, Y. J., Lemoine, R. O., Sehabiague, L., Morsi, B. I., Jones, K. L., and Pennline, H. W. (2005). Investigation of perflourinated compounds as physical solvents for selective CO2 capture at elevated pressures and temperatures. Paper presented at 22nd Annual International Pittsburgh Coal Conference, Pittsburgh, PA. Hendrickson, T. A. (1975). Synthetic fuels data handbook. Denver, CO: Cameron Engineers. Higman, C., and van der Burgt, M. (2003). Gasification. Boston, MA: Gulf Professional Publishing (Elsevier). Ilconich, J., Myers, C., Pennline, H., and Luebke, D. (2007). Experimental investigation of the permeability and selectivity of supported ionic liquid membranes for CO2/He separation at temperatures up to 125°C. J. Memb. Sci. 298:41. Jaeger H. (2007). CO2 cloud looms large over IGCC and gasification plant development. Gas Turbine World, November–December, p. 16. Jenkins, B. M., Baxter, L. L., Miles, T. R., Jr., and Miles, T. R. (1998). Combustion properties of biomass. Fuel Proc. Tech. 54:17. Klett, M. G., Maxwell, R. C., and Rutkowski, M. D. (2002). The cost of mercury removal in an IGCC plant. Available at www.netl.doe.gov. Kohl, A., and Nielsen, R. (1997). Gas purification. 5th ed. Houston, TX: Gulf Publishing. Korens, N., Simbeck, D. R., and Wilhelm, D. J. (2002). Process screening analysis of alternative gas treating and sulfur removal for gasification. Final report, U.S. Department of Energy, National Energy Technology Laboratory Task Order 739656-00100. Kriebel, M., Sclicting, H., and Tanz, H. (1989). Gas treating. In Ullman’s encyclopedia of industrial chemistry, sect. 5.4–5.8, pp. 395–429. 5th ed. Weinheim: Wiley-VCH. Leonard, R. (2007). Gasification testing at the power systems development facility. Paper presented at 2007 Gasification Technologies Conference, San Francisco, October 14–17. Available at www.gasification.org. Mahagaokar, V., and Doering, E. L. (1995). High level heat recovery in coal and coke gasification and combined cycles. ASME 95-67-259. McClendon, T. R., Lui, A. P., Pineault, R. L., Beer, S. K., and Richardson, S. W. (2004). High-pressure co-gasification of coal and biomass in a fluidized bed. Biomass and Bioenergy 26:377. Megaritis, A., Messenbock, A.-G., Collot, Y., Zhuo, Y., Dugwell, D. R., and Kandiyoti, R. (1998). Internal consistency of coal gasification reactivities determined in bench scale reactors: Effect of pyrolysis conditions on char reactivities under high-pressure CO2. Fuel 77:1411. Mitsubishi 250MW demo plant on target for mid-2007 testing. (2007). Gas Turbine World, January–February, pp. 42–45. Nelson, J. M., Leonard, R., Liu, G., Peng, W., Romans, D., Vimalchand, P., Smith, P. V., and Longanbach, J. (2003). Low-rank coal gasification studies at the PSDF. Paper presented at the 18th Low-Rank Fuels Symposium, Session 2B—Gasification, Billings, Mon­tana, June 24–26, 2003. Available at http://psdf.southernco.com/pdf/2003/paper-145.pdf. NETL. (2008). http://www.netl.doe.gov/technologies/coalpower/gasification/basics/3.html.

Gasification Technology to Produce Synthesis Gas

27

Paisley, M. A., and Overend, R. P. (2002). Verification of the performance of Future Energy Resources: Silvagas biomass gasifier. Operating Experience in the Vermont Gasifier, Paper presented at 19th International Pittsburgh Coal Conference, Pittsburgh, PA. Palonen, J., Anttikoski, T., and Eriksson, T. (2005). The Foster Wheeler gasification technology for biofuels: Refuse-derived fuel (RDF) power generation. Paper presented at Power Gen Europe. Available at http://www.fwc.com/publications/tech_papers/. Poulston, S., Granite, E. J., Pennline, H. W., Myers, C. J., Stanko, D. P., Hamilton, H., Rowsell, L., Smith, A., Ilkenhans, T., and Chu, W. (2007). Metal sorbents for high temperature mercury capture from flue gas. Fuel 86:2201. Probstein, R. F., and Hicks, R. E. (1990). Synthetic fuels. Cambridge, MA: pH Press. Ratafia-Brown, J., Manfredo, L., Hoffman, J., and Ramezan, M. (2002). Major environmental aspects of gasification-based power generation technologies. National Energy Technology Laboratory. Available at http://www.netl.doe.gov/technologies/coalpower/ gasification/pubs/. Reed, T. B., and Gauer, S. (2001). A survey of biomass gasification 2001. 2nd ed. National Renewable Energy Technology Laboratory, written under DOE Contract DE‑AC3683CH10093, Subcontract ECG-6-16604-01(BEF). Riemert, R. (1989). Gas production from coal, wood, and other solid feedstocks. In Ullman’s encyclopedia of industrial chemistry, 357–379, 5th ed. Weinheim: Wiley-VCH. Rudolph, P. F. H. (1984). Lurgi coal gasification (moving bed gasifier). In Handbook of synfuel technology, ed. R. A. Myer, pp. 3-127–3-148. New York: McGraw-Hill Publishing. Salvador, S., Commandre, J.-M., and Stanmore, B. R. (2003). Reaction rates for the oxidation of highly sulphurized cokes: The influence of thermogravimetric conditions and some coke properties. Fuel 82:715. Schlather, J., and Turk, B. (2007). Comparison of warm-gas desulfurization process versus traditional scrubbers for a commercial IGCC power plant. Paper presented at the 2007 Gasifi­ca­ tion Technology Conference, San Francisco. Available at http://www.gasification.org/. Shadle, L. J., Berry, D. A., and Syamlal, M. (2002). Coal conversion processes, gasification. In Kirk-Othmer encyclopedia, ed. S. Seitz. New York: John Wiley & Sons. Shah, V. A., and McFarland, J. (1988). Low cost ammonia and CO2 recovery. Hydrocarbon Proc. 67:43. Simbeck, D. R., Korens, N., Biasca, F. E., Vejtasa, S., and Dickenson, R. L. (1993). Coal gasification guidebook: Status applications, and technologies. EPRI-TR-102034, Research Project 2221-39. Siriwardane, R., Robinson, C., Shen, M., and Simonyi, T. (2007a). Novel regenerable sodiumbased sorbents for CO2 capture at warm gas temperatures. Energy and Fuels 21:2088. Siriwardane, R., Tian, H., Simonyi, T., and Webster, T. (2007b). Regenerable sorbent development for sulfur, chloride and ammonia removal from coal derived synthesis gas. Paper presented at 2007 International Conference on Coal Science and Technology, August 2007, Nottingham, England. Stringer, J. (1989). Material issues in pressurized fluidized bed combustors. Paper presented at Proceedings of the 1988 Fluidized Bed Combustion Technologies for Utility Appli­ cations, May 3–5, 1988, Palo Alto, CA. Supp, E. (1990). How to produce methanol from coal. Berlin: Spinger-Verlag. Takematsu, T., and Maude, C. (1991). Coal gasification for IGCC power generation. IEA CR 37. Tampa Electric Integrated Gasification Combined-Cycle Project. (2000). U.S. Department of Energy, Office of Fossil Energy, Clean Coal Technology Topical Report 19. Available at http://www.fossil.energy.gov/programs/powersystems/publications/ Clean_Coal_Topical_Reports/. Thambimuthu, K. V. (1993). Gas cleaning for advanced coal-based power generation. IEACR/53, IEA Coal Research, London.

28

Synthesis Gas Combustion: Fundamentals and Applications

Tillman, D. A. (1994). Fuels from waste. In Kirk-Othmer encyclopedia of energy technology. 4th ed., Vol. 12. New York: John Wiley & Sons. Trembly, J., Gemmen, R. S., and Bayless, D. J. (2007a). The effect of IGFC warm gas cleanup system conditions on the gas-solid partitioning and form of trace species in coal syngas and their interactions with SOFC anodes. J. Power Sources 163:986. Trembly, J., Gemmen, R. S., and Bayless, D. J. (2007b). The effect of coal syngas containing HC1 on the performance of solid oxide fuel cells: Investigations into the effect of operational temperature and HC1 concentration. J. Power Sources 169:347. Trembly, J., Gemmen, R. S., and Bayless, D. J. (2007c). The effect of coal syngas containing AsH3 on the performance of SOFCs: Investigations into the effect of operational temperature, current density and AsH3 concentration. J. Power Sources 171:818. Twigg, M. V. (1996). Catalyst handbook. London: Manson Publishing. Vogt, E. V., Weller, P. J., and Vanderburgt, M. J. (1984). The Shell coal gasification process. In Handbook of synfuel technology, ed. R. A. Myer, pp. 3-27–3-44. New York: McGrawHill Publishing. Wallace, F., Guan, X., Leonard, R., Nelson, M., Vimalchand, P., Peng, W., Smith, P. V., and Breault, R. W. (2006). Operation of the Transport Gasifier™ at the PSDF. Paper presented at the 31st International Technical Conference on Coal Utilization and Fuel Systems, Clearwater, FL, May 21–25, 2006. Available at http://psdf.southernco.com/ pdf/2006/paper-155.pdf. Wenglarz, R., Lippert, T., and Alvin, M. A. (October 1986). Alternative combustion turbine designs and clean-up systems for pressurized fluidized bed combustion power plants. EPRI-CS-4860, Electric Power Research Institute. WePower SCPC and IGCC. (2003). Information from April 2003 draft environmental impact statement, Elm Road generating station, Table  7-11, p. 157. Vol. 1. Public Service Commission of Wisconsin and Department of Natural Resources. Werthera, J., Saengera, M., Hartgea, E.-U., Ogadab, T., and Siagib, Z. (2000). Combustion of agricultural residues. Prog. Energy Comb. Sci. 26:1. Zulfiqar, M., Moghtaderi, B., and Wall, T. F. (2006). Flow properties of biomass and coal blends. Fuel Proc. Tech. 87:281.

Chemical Kinetics 2 Syngas and Reaction Mechanisms Marcos Chaos, Michael P. Burke, Yiguang Ju, and Frederick L. Dryer Contents 2.1 Introduction..................................................................................................... 29 2.2 Explosion Characteristics of H2-Containing Systems..................................... 30 2.2.1 Effect of Impurities: NOX.................................................................... 35 2.3 Recent and Proposed Updates to the H2/CO Kinetic Model........................... 37 2.4 High-Pressure/Low-Temperature Syngas Ignition and Kinetic Implications..................................................................................................... 43 2.4.1 Shock Tube Pressure Histories............................................................46 2.4.2 Modeling Approaches.......................................................................... 49 2.4.3 Other Systems...................................................................................... 51 2.5 Premixed Flame Propagation in High-Pressure Media................................... 52 2.6 Conclusion....................................................................................................... 61 Acknowledgments..................................................................................................... 63 References................................................................................................................. 63

2.1 Introduction Chapters 1 and 7 detail syngas generation technologies and their use in established and developing applications for power production as well as chemical synthesis. Syngas is receiving renewed interest, as it can be flexibly generated from a variety of nonpetroleum, renewable feedstocks. In addition, syngas production, along with the implementation of carbon capture and storage (CCS) technologies (Chiesa et al., 2005), has the potential for substantially “greening” fossil fuel and biomass energy resource use, particularly when polygeneration of fuels, chemical products, and electrical power generation are integrated with CCS (Larson et al., 2009). Although gasification technologies are well established and syngas combustion is widely used in, for example, refineries for process heat (see Chapter 1), its application to power generation is not sufficiently developed. Conditions relevant to the applications described in Chapter 2 (e.g., gas turbine combustors, internal combustion engines, etc.) are typically characterized by lower temperatures and higher pressures (T < 1000 K, 10 < P < 30 atm). At such conditions, the chemistry and combustion dynamics of syngas as well as hydrogen are not fully understood. This is despite the great amount of detailed studies that have been carried out through the years on the CO/H2/O2 29

30

Synthesis Gas Combustion: Fundamentals and Applications

kinetic system (e.g., Davis et al., 2005; Dixon-Lewis and Williams, 1977; Gardiner and Olson, 1980; Kim et al., 1994; Li et al., 2007; Rasmussen et al., 2008; Sun et al., 2007; Sung and Law, 2008; Westbrook and Dryer, 1984; Yetter et al., 1991a, 1991b), as this system is the fundamental basis of all hydrocarbon combustion chemistry. As a result of interest in syngas combustion noted above, robust fluid dynamics as well as chemical kinetic modeling tools are sought that are thoroughly validated against experiments spanning a wide range of operating conditions. The ultimate goal of these modeling efforts is to achieve accurate predictive behavior, for example, of dynamic combustor features such as blowout, flashback, stability, and autoignition, including how fuel composition might affect these properties (Lieuwen et al., 2008). Driven by the above discussion, data have recently become available (Beerer and McDonell, 2007; Bradley et al., 2007; Burke et al., 2007, 2009a; Kalitan et al., 2006; Mittal et al., 2006; Natarajan et al., 2007; Petersen et al., 2007a; Sivaramakrishnan et al., 2007; Sun et al., 2007; Walton et al., 2007a) for conditions that are sufficiently removed from the range of previous kinetic model validations, as reviewed by Chaos and Dryer (2008). Since this early work and that of Sung and Law (2008), additional experimental data have been published (Beerer and McDonell, 2008; Herzler and Naumann, 2008; Le Cong and Dagaut; 2008; Mertens et al., 2009; Natarajan et al., 2009; Rasmussen et al., 2008) that lend support to further considering the title topic. A common feature of the studies listed above is the use of undiluted systems (i.e., fuel–air), high pressures, and relatively low temperatures in autoignition studies (e.g., Beerer and McDonell, 2008), as well as preheated mixtures in laminar flame speed measurements (e.g., Natarajan et al., 2009). Under these conditions, there have been several reports noting considerable discrepancies between experimental data and kinetic model predictions. These observations motivate further consideration, and it is the purpose of this chapter to discuss the need for careful analysis of data collected at high pressures and the implications that experimental anomalies, known to be present at these conditions, may have on interpreting these data. This discussion is further motivated by recent evaluations of key reactions in the hydrogen system, which will be reviewed below. This chapter begins with an introduction to the chemistry driving the explosive characteristics of hydrogen oxidation and how they can be altered by the presence of impurities. A review then follows of available chemical kinetic models for simulation of H2/CO chemistry along with important updates to elementary reactions that can considerably affect the performance of these models at high pressures. Finally, recent experimental ignition and burning velocity data collected at conditions relevant to practical applications noted above are presented and analyzed; emphasis is placed on what implications these data have on further development of new and existing chemical kinetic models. For further overview and background of H2/CO kinetics, the reader is referred to the works of Chaos and Dryer (2008) and Rasmussen et al. (2008).

2.2 Explosion Characteristics of H2-Containing Systems Syngas derived from gasification can consist of high levels of hydrogen, depending on the feedstock used. Syngas combustion is very similar in character to that of hydrogen, particularly since the principal reaction producing CO2 is CO + OH = CO2 + H. The

Syngas Chemical Kinetics and Reaction Mechanisms

31

reactivity of any fuel-oxidizer mixture is driven by the generation and chain propagation of free active radicals. In the case of carbon monoxide oxidation, in the absence of any source of hydrogen atoms, reactions that can initiate the chain (mainly through propagation of O atoms) are slow at practical conditions, as are reactions of O atoms with CO. The addition of even small amounts of hydrogen or hydrogen-containing species (e.g., water) dramatically increases the rate of CO oxidation (Brokaw, 1967; Yetter et al., 1991a), as radicals are propagated through much faster hydrogen-related reactions. Hence, in design of syngas combustion systems it is important to have an understanding of the oxidation characteristics of hydrogen, as it will determine properties of interest, such as ignition and extinction limits, flame propagation, etc. This is best explained through the concept of explosion limits, that is, the pressure-temperature boundaries that demarcate regions of slow and explosively fast reaction of a given fuel-oxidizer mixture. Details for hydrogen oxidation explosion limits are given below, with emphasis on the conditions of interest in the present study. Figure 2.1 shows a typical depiction of the classical explosion limits for hydrogen– oxygen mixtures (Lewis and von Elbe, 1987). These limits arise because of competition between chain-branching and chain-terminating reactions either on surfaces or in the gas phase. Reactions of free radicals with other species yielding more than one free radical as a product would by themselves lead to an exponential growth of the radical pool (i.e., chain-branching reactions). On the other hand, chain-propagating reactions only produce one radical for every radical consumed, and reactions that remove radicals entirely can reduce the radical pool (i.e., chain-terminating reactions). The balance of overall chain propagation with termination processes such that the radical pool neither grows nor decays represents a critical branching factor for the chemical reacting system. The critical branching factor for a homogenous gas phase reacting system is a function of system properties, including its temperature, pressure, and reactant mixture fractions. The presence of diffusion and heterogeneous surfaces can also lead to radical production and destruction, modifying the critical branching factor, depending on the characteristic time scales of these processes relative to those in the gas phase. Finally, if a reaction is endothermic or exothermic, changes in local temperature as a result of reaction can also affect the critical branching factor. The loci of all critical branching factor conditions as a function of system parameters define regions of explosive chain branching and slow chemical reaction, which are termed explosion limits. For the following discussion it is assumed that the pressure of a reacting system maintained at a constant temperature of about 750 K is slowly increased from low (<0.001 atm) to high (>10 atm) pressure. The reaction properties associated with these constraints (Figure  2.1) cross the explosion limit curve for the hydrogen–oxygen system in three classical explosion limit regimes. The first limit, observed at very low system pressures, is determined by a balance between the removal of radicals on surfaces (wall effect), when diffusion time scales in the gas phase are short in comparison to characteristic reaction time scales and their production in the gas phase. At pressures below the limit, gas phase reaction branching is exceeded by termination (which is dominated by radical diffusion to and termination at surfaces), and the rate of branching is less than the critical branching factor. Above the limit, the rate of radical branching is sufficient to exceed the rate of termination, since there

32

Synthesis Gas Combustion: Fundamentals and Applications 102 Thermal-chain explosion (Mild ignition)

rd

3

101

Lim

100

No explosion

k1 = k2 [M]

1+

10–1

1 st

10–3

600

k6

Branched-chain explosion (Strong ignition)

Lim it

10–2

1 k7

2 nd

Pressure (atm)

it

2k1 = k2 [M]

Lim

it

800

1000 1200 Temperature (K)

1400

Figure  2.1  Explosion limit characteristics for stoichiometric H2 /O2 mixtures. Classical limits (Lewis and von Elbe, 1987) are shown by the solid line. Calculations (using reaction rates from the model of Li et al., 2004) for the pressure-temperature boundaries determining the second explosion limit (dashed line) and the extended second explosion limit (dash-dot line) are also shown (see text). The filled circles denote the conditions, in temperature-­pressure space, for the plots shown in Figure 2.2.

is a greater number of molecular collisions and diffusive time scales increase due to pressure. As the system pressure is further increased, the classical second explosion limit condition is reached, a limit that results entirely from gas phase kinetic processes (i.e., homogeneous gas phase kinetic branching and termination processes). In pure hydrogen oxidation, the most effective chain-branching reaction is H + O2 = O + OH (R1), while the competing reaction, H + O2(+M) = HO2(+M) (R2), leads to the production of a radical (HO2) that is much less reactive than OH or O. Table 2.1 shows a list of reactions considered in this chapter. The time scales for HO2 to regenerate active radical species at these parameter conditions are very large in comparison to the overall reaction time scale. With increasing pressure, the rate of ternary collisions in reaction (R2) increases relative to binary collisions in (R1). Performing a steady-state analysis of the radical pool at the second limit using these two reactions yields the concentration of M (i.e., a collision partner stabilizing reaction (R2)) for the classical second limit, given as [M] = 2k1/k2 (Lewis and von Elbe, 1987), where ki (i = 1, 2) represent the specific reaction rate constants of (R1) and (R2) in the

Syngas Chemical Kinetics and Reaction Mechanisms

33

Table 2.1 List of Reactions Discussed in the Text Label

Reaction

Label

Reaction

(R1) (R2) (R3) (R4) (R5) (R6) (R7) (R8) (R9)

H + O2 = O + OH H + O2(+M) = HO2(+M) HO2 + HO2 = H2O2 + O2 H2O2(+M) = OH + OH (+M) HO2 + H2 = H2O2 + H HO2 + H = H2 + O2 HO2 + H = OH + OH NO + HO2 = NO2 + OH NO2 + H = NO + OH

(R10) (R11) (R12) (R13) (R14) (R15) (R16) (R17)

CO + OH = CO2 + H HCO + M = H + CO + M HCO + O2 = HO2 + CO CO + HO2 = CO2 + OH HO2 + OH = H2O + O2 H2 + OH = H2O + H O + H2 = H + OH H + OH + M = H2O + M

forward direction (reverse rates are neglected as they involve processes that are very slow at these conditions relative to the forward processes). At higher pressures well above the classical second limit, a third limit condition is reached, as a result of two principal issues: (1) the high concentration of M leads to a sufficient concentration of HO2 such that reaction HO2 + HO2 = H2O2 + O2 (R3) and subsequent reactions of H2O2, principally H2O2(+M) = OH + OH(+M) (R4), transition the terminating character of (R2) into an overall process (represented by the combination of (R2) and the additional subsequent reactions) that is essentially chain propagating; and (2) the reaction process becomes very exothermic, and along with increased thermal diffusion time scales with increasing pressure, this leads to local mixture self-heating (the process is no longer isothermal). Reactions (R2) and (R3) are very exothermic, being moderated by the endothermicity of reaction (R4) (Lee and Hochgreb, 1998). At high pressures, the character of the reaction also changes with initial reaction temperature. Calculations shown in Figure 2.2a exemplify this behavior. These computations were performed for a constant-volume, adiabatic, and stoichiometric H2/O2 system at an initial pressure of 20 atm and for temperatures of 1000 K and 1300 K; these conditions are also marked in Figure 2.1. The displayed results are normalized in each case by the initial temperature and by the time at which the maximum rate of temperature rise is observed. At 1000 K, a significant increase in the system temperature is seen during a chemical induction period due to heat release associated with the above reactions (R2, R3, R4), while at 1300 K, no similar heat release effect is observed. Also shown in Figure 2.2b is a plot of characteristic reaction times (determined by the ratio of the initial H2 concentration to the maximum H2 consumption rate) as a function of temperature for the same system. Figure 2.2b shows that as temperature increases there is a distinct transition in characteristic reaction times indicating the presence of an explosion limit. Increasing temperatures produces an increasing flux of HO2 radicals via HO2 + H2 = H2O2 + H (R5), generating H radicals. H radicals, in turn, react with the abundant HO2 through HO2 + H = H2 + O2 (R6) and HO2 + H = OH + OH (R7). Reaction (R6) is chain terminating, whereas (R7) is chain branching. Therefore, a system that is initially straight chain in overall reaction character (i.e., R2→R3→R4) can become chain branching as a result of

34

Synthesis Gas Combustion: Fundamentals and Applications

Normalized Temperature

1.2 1000 K 1300 K

1.15 1.1 1.05 1 0

0.2

0.4

0.6

0.8

1

Normalized Time (a)

[H2]o/(d[H2]/dt)max (seconds)

1.00 Extended second limit

10–1 10–2 10–3

Ea = 46,500 cal/mol

10–4 10–5

Ea = 26,350 cal/mol 1000

1200 Temperature (K)

1400

(b) Figure 2.2  Constant-volume adiabatic calculations (using the model of Li et al., 2004) for a stoichiometric H2/O2 system at 20 atm. (a) Normalized temperature-time traces for two temperature conditions (solid line, 1000 K; dashed line, 1300 K) below and above the extended second limit (see Figure 2.1). (b) Characteristic reaction times as a function of temperature; overall activation energies (Ea) at temperatures above and below the extended second limit are listed and have been estimated by fitting the computed data to an equation of the form tc = A exp(Ea /RT), where tc is the characteristic time, R the universal gas constant, T the temperature, and A a preexponential factor.

Syngas Chemical Kinetics and Reaction Mechanisms

35

temperature rise (R2→R5→R7, R4). The limit observed under these conditions is an extension of the classical second limit, modified by the importance of added termination and branching routes (R6 and R7) at high pressures and temperatures. A steady-state analysis of the radical pool considering these additional reactions leads to the determination of a critical third-body concentration (i.e., pressure) marking the explosion limit as

[M] =

k1  1  k2    1 + k7  k6  

This limit is typically referred to as the extended second limit (Mueller et al., 1996, 1999a). The relevance of the extended second limit to overall system behavior depends upon a comparison of chemical kinetic time scales with convective, back-mixing, or diffusive time scales (Mueller et al., 1996, 1999a; Zheng and Law, 2004). Figure 2.2b shows that at temperatures below the extended second limit the overall reaction is characterized by a high activation energy. Thus, any self-heating that occurs during induction (see Figure 2.2a) is important to overall system behavior. Conditions located in the region between the third limit and the extended second limit can be described as leading to thermal-chain explosions (Figure  2.1), where heat release from slow chain-propagating reactions brings a reactive mixture to a chain-branched explosive condition, as explained above. This region, for example, is associated with mild ignition phenomena observed in shock tubes (Strehlow and Cohen, 1962; Meyer and Oppenheim, 1970; Oppenheim, 1985), as will be further detailed below. On the other hand, at pressures below and temperatures above the extended second limit, explosions are nonthermal and strictly due to chain branching, characterized by relatively low overall activation energies (see Figure 2.2b). For conditions relevant to practical syngas applications (e.g., lean premixed gas turbine engines), typical pressures are on the order of 10 to 30 atm and temperatures not in excess of 1000 K. These conditions place these systems in the thermalchain explosion regime. Therefore, chemical kinetics defining the extended second limit remain an important aspect in the design of such combustion systems. In fact, the recent experimental studies listed above have placed new emphasis on the importance of the chemistry of the hydroperoxyl radical (HO2) as well as hydrogen peroxide (H2O2) due to the conditions studied. These investigations are the subject of further analyses given in the next sections.

2.2.1 Effect of Impurities: NOX It is relevant to present in this section a brief discussion on the effects that impurities may have on the kinetic character of hydrogen and syngas oxidation, particularly as applied to the nature of the explosion limits discussed above. Although syngas is mostly composed of hydrogen and carbon monoxide, the presence of other trace

36

Synthesis Gas Combustion: Fundamentals and Applications

species can have a pronounced effect on combustion processes (Glarborg, 2007). These species alter kinetic behavior by catalyzing radical generation or recombination processes. Contaminants generated during feedstock gasification (to produce syngas) include nitrogen and sulfur oxides, among other species (see Chapter 1). For representative purposes, and to complement the discussions in Chapters 6 and 10, attention is given here to nitrogen oxides; details on the combustion chemistry of H2/CO in the presence of sulfur oxides can be found in the work of Mueller et al. (2000). The presence of small quantities of NOX can have significant influence on the underlying H 2/O2 kinetic behavior of syngas mixtures by drastically modifying the relative influence of HO2 chemistry at conditions near those of the extended second explosion limit (Ashmore and Tyler, 1962; Chaos and Dryer 2008; Mueller et al., 1998, 1999b). NOX provides an alternate consumption route for HO2 radicals via NO + HO2 = NO2 + OH (R8) and NO2 + H = NO + OH (R9). This route effectively competes with (R2) for H atoms. Reactions (R8) and (R9) therefore establish a catalytic cycle that releases OH so that fast fuel consumption can occur at temperatures well below the explosion limit of the unperturbed system. In the temperature range where this catalytic cycle is active, small amounts of NOX significantly increase the overall chemical reaction rate, principally through substantial reduction in the chemical induction time, leading to establishing critical branching. Figure 2.3 shows the effects of the addition of small amounts of NO on the constant pressure oxidation of H2/CO mixtures at pressures from 0.5 to 40 atm. This figure is similar to Figure 2.2 in that it plots the numerically predicted characteristic reaction times (using the model of Li et al. [2007] with NOX kinetic reactions adopted from Mueller et al. [1999b]). One can clearly observe that not only is the extended second explosion limit behavior modified as NO concentration increases, but the overall rate of reaction (defined by the fuel concentration reacted over the characteristic reaction time) is significantly altered. For low pressures and at temperatures below the explosion limit, HO2 radicals are the primary chain carriers, and the addition of even small quantities of NO can reduce the characteristic reaction times by orders of magnitude, as NO establishes chain braching as opposed to the thermal-chain processes described above. This result can have significant consequences on the auto­ ignition characteristics of hydrogen–oxygen and carbon monoxide–hydrogen–oxygen reactions if small amounts of combustion exhaust gases become back-mixed into entering reactants, an expected circumstance as a result of residual gas and exhaust gas recirculation (EGR) in the combustion of hydrogen and syngas in reciprocating engines (Chapter 10). As noted in Chapter 10, introduction of H2 in engine applications tends to increase NOX emissions due to an increase in reaction temperature. Even small levels of NOX in the EGR stream or in residual cylinder gases can affect ignition timing as well as engine performance (Chaos and Dryer, 2008). Note from Figure 2.3 that there is a nonlinear effect in terms of the amount of added NO, and further that the effect is also a function of both reaction temperature and pressure. Addition of an appropriate amount of NO causes the overall activation energy of the reaction to be nearly the same above and below the explosion limit condition. The difference in overall oxidation rate above and below the explosion limit

Syngas Chemical Kinetics and Reaction Mechanisms

37

Figure 2.3  Characteristic reaction times of H2/CO mixtures (defined as in Figure 2.2) as a function of temperature at various pressures and initial NO concentrations. For illustration purposes, mixture compositions similar to those of Mittal et al. (2006) are chosen (H2/CO/ O2 /N2 /Ar, 6.25/6.25/6.25/18.125/63.125 mol%). Reaction times were determined assuming isobaric systems.

eventually is decreased (due to the increased rate of reactions involving HO2), and the explosion limit becomes described by essentially the same overall temperature dependence. It is this function that is most important in defining the engineering parameters over which mixing can be accomplished without flashback or preignition in syngas combustion applications.

2.3 Recent and Proposed Updates to the H2/CO Kinetic Model Kinetic models that can be applied to hydrogen and syngas systems continue to be refined (Davis et al., 2005; Konnov, 2008; Li et al., 2004, 2007; O’Connaire

38

Synthesis Gas Combustion: Fundamentals and Applications

et al., 2004; Rasmussen et al., 2008; Sun et al., 2007; Wang et al. 2007) as new rate evaluations, thermochemical parameters with reduced uncertainties, and experimental validation data become available. For example, Li et al. (2004, 2007) recently revised prior work by updating the detailed H2/O2 model of Mueller et al. (1999a). The updates by Li et al. included a revision in the rate correlations used for the branching reaction (R1), taken from Hessler (1998), and for the low pressure of the competing reaction (R2), adopted from Michael et al. (2002). A major constraint in selecting these two correlations was that their ratio in the temperature range between 800 K and 900 K replicated the ratio experimentally determined by Mueller et al. (1998). This ratio is a representation of the critical third-body concentration (i.e., pressure) that marks the classical second explosion limit ([M] = 2k1/k2), as described above. In fact, as models develop, one has to ensure that he or she can properly represent the classical explosion limits of the hydrogen system; such tests, however, are often not performed. Figure 2.4 shows comparisons of explosion limit data against predictions from the kinetic models listed above (Davis et al., 2005; Konnov, 2008; Li et al., 2004, 2007; Sun et al., 2007). Agreement is good overall, but some differences are appreciable. For example, at 0.1 atm the models predict a range of temperatures for transition into the explosive regime of about 745 to 765 K. These differences manifest when using the models to predict observables such as ignition delay at conditions near the extended second limit (such computations using various

Davis et al. (2005) Konnov (2008) Li et al. (2004, 2007) Sun et al. (2007)

Pressure (atm)

100

10–1

10–2

Baldwin et al. (1967) Baulch et al. (1988) Egerton and Warren (2005) Muller et al. (1999a) Von Elbe and Lewis (1942) 650

700

750 800 850 Temperature (K)

900

950

Figure  2.4  Comparison of H2/O2 second explosion limit experimental data (symbols) with the criterion [M] = 2k1/k2 computed using rate values obtained from the kinetic models listed (lines) for M = N2. Data are for stoichiometric mixtures, obtained in a flow reactor (Mueller et al., 1999a), static reactors (Baldwin et al., 1967; Egerton and Warren, 1951; von Elbe and Lewis, 1942), and a well-stirred reactor (Baulch et al., 1988). Experiments used 2:1 (mol) H2:O2 mixtures except the studies of Baldwin et al. (28% H2, 14% O2, 58% N2, mol) and Mueller et al. (1% H2, 0.5% O2, 98.5% N2, mol). The data have been modified to take into account the third-body efficiencies of H2 and O2 relative to N2; efficiencies were taken from von Elbe and Lewis (1942). The bold dash-dot line denotes the explosion boundary for a system containing water (10 mol%), calculated using the model of Li et al. (2004, 2007).

39

Syngas Chemical Kinetics and Reaction Mechanisms

Baulch et al. (2005) Fernandes et al. (2008) Michael et al. (2002) Sellevag et al. (2008) Troe (2000)

k2,0(cm6/mol2/s)

1×1016 8×1015 6×1015 4×1015 2×1015 0.4

0.8

1.6 1.2 1000/T(K–1)

2

Figure  2.5  Temperature dependence of the low-pressure-limit reaction rate of H + O2(+M) = HO2(+M) (R2) for M = N2. The recent evaluation of Baulch et al. (2005) is also plotted with associated uncertainties.

models, including Li et al. [2004] and O’Connaire et al. [2004], can be found in the recent works of Konnov [2008] and Ströhle and Myhrvold [2007]). Experimental measurements and theoretical studies continue to appear in the literature for both (R1) and (R2) (Bates et al., 2001; Fernandes et al., 2008; Hahn et al., 2004; Hwang et al., 2005; Sellevåg et al., 2008; Troe, 2000; Troe and Ushakov, 2001) to further reduce uncertainties in predicted rates, particularly at temperatures below 1000 K. For example, Figure 2.5 compares recent correlations for the low pressure rate of reaction (R2) with N2 as the collisional partner. Although some uncertainties remain (e.g., see Hwang et al., 2005), all of them lie within 20% of each other and within the uncertainty estimates derived from the literature review of Baulch et al. (2005). Figure 2.6 shows calculations for the effective rate of (R2) using values obtained from the recent work of Fernandes et al. (2008) and Sellevåg et al. (2008). Overall, values differ by about 25% over the temperature and pressure range of interest to practical applications. Fernandes et al. (2008) claim these differences are due to a less complete database and a less detailed analysis considered by Sellevåg et al. (2008). Nevertheless, studies of (R2) remain important, especially in determining the collisional efficiencies of species such as water and carbon dioxide. Efficiencies are hard to determine experimentally, and associated uncertainties can be considerable; furthermore, the interaction of polar species such as water in (R2) is not completely understood (Michael et al., 2002). The presence of these species increases the recombination rate of (R2), effectively shifting the explosion limit behavior of H2/O2. The explosion limit data shown in Figure 2.4 were adjusted to make all values relative to N2. However, for comparison, model results are also plotted for a mixture consisting of 10% (mol) H2O. The presence of this highly effective collision partner has a very noticeable impact; pressure and temperature conditions for which a N2 system is explosive (e.g., 1 atm and 950 K) can become nonexplosive with the addition of water. This has direct implications in practical systems where back-mixing

40

Synthesis Gas Combustion: Fundamentals and Applications

1015 k2,

k2(cm3 mol/s)

1014 100 atm

1013

10 atm

1012

1 atm

1011

0.1 atm

1010 109 108

0

0.5

1

1.5

2

1000/T(K

–1)

2.5

3

3.5

Figure 2.6  Temperature and pressure dependence of (R2) from recent evaluations for M = N2. Solid lines represent the values of Fernandes et al. (2008); dashed lines are the recommendation of Sellevåg et al. (2008).

of reaction products in combustor recirculation zones (for example, to control NOX emissions) can change combustion behavior and stability. Furthermore, as discussed in Chapter 7, steam may also be used in gas turbines to achieve the desired power output by increasing mass flow. Other reactions of critical importance in the oxidation of H2/CO/O2 as well as moist CO mixtures have received further attention recently. Li et al. (2007) employed a weighted empirical fit of the entire body of experimentally measured rate constants for reactions CO + OH = CO2 + H (R10), HCO + M = H + CO + M (R11), and HCO + O2 = HO2 + CO (R12), all of which have a strong impact on the burning rate of hydrocarbon flames (Zhao et al., 2005). New high-pressure data (described above) have allowed analyses to be performed that have led to proposed updates in some of the reactions involving HO2, two of the most significant being the rate correlations for CO + HO2 = CO2 + OH (R13) and HO2 + OH = H2O + O2 (R14). Mittal et al. (2006, 2007) used data from high-pressure rapid compression machine (RCM) ignition of H2/CO/O2/N2/Ar mixtures and noted that the rate of (R13) could be up to a factor of 10 lower than some accepted values (e.g., Baulch et al., 1973). This is consistent with the recommendation of Mueller et al. (1999b), who argued, based on flow reactor data, that lower values for (R13) might be justified upon further evaluation. Theoretically, (R13) has received less attention than other reactions in the CO-hydrogen system (Allen et al., 1996; Sun et al., 2007; You et al., 2007). The detailed treatment of You et al. (2007) (which considered critical geometries, hindered internal rotations, as well as the complex potential energy surface of (R13)) supports a much lower rate for (R13) than those commonly used in kinetic models. Chaos and Dryer (2008) showed that implementing the rate of You et al. (2007) in the model of Li et al. (2007) considerably improved predictions against the data of Mittal et al. (2006) (Figure 2.7). Chaos and Dryer (2008) and Li

41

Syngas Chemical Kinetics and Reaction Mechanisms

Ignition Delay (ms)

6 Experiment (Mittal et al. 2006) Base model (Li et al. 2007) Base model (Li et al. 2007) w/update from You et al. (2007)

4

2

0

0

0.2

0.4 Rco

0.6

0.8

Figure 2.7  Comparison between measured (Mittal et al., 2006) and predicted RCM ignition delay times for a (H2 + CO)/O2/N2/Ar—12.5/6.25/18.125/63.125 (molar) mixture. RCO is the fraction of CO in (H2 + CO). Compressed conditions are 50 bar and 1044 K. Lines show the difference in the quality of model predictions when the rate of You et al. (2007) for (R13) is used in the model of Li et al. (2007). Modeling performed in accordance with the methods described in Mittal et al. (2006).

et al. (2007) also showed through computational singular perturbation (CSP) analyses (Kazakov et al., 2006) how reaction (R13) contributed to the ignition process through its effect on the induction period. Chaos and Dryer (2008) further concluded that the updates put forth by Mittal et al. (2006) and You et al. (2007) for (R13) can be readily absorbed into current kinetic models without affecting the quality of their predictions against validation targets that are insensitive to induction chemistry.* Reaction (R14) is of particular significance in oxygen-rich systems; for example, (R14) can alter the nature of the extended second limit as the relative importance of O and OH radicals as chain carriers varies with the presence of excess oxygen. Reaction (R14) has been mostly studied at atmospheric conditions (Braun et al., 1982; Burrows et al., 1981; Cox et al., 1981; DeMore, 1979, 1982; Goodings and Hayhurst, 1988; Keyser, 1988; Kurylo et al., 1981; Lii et al., 1980; Peters and Mahnen, 1973; Rozenshtein et al., 1984; Sridharan et al., 1984) due to the significance of this reaction in the HOx cycle of atmospheric chemistry. More recently, high temperature measurements (Hippler et al., 1995; Kappel et al., 2002) have shown that (R14) exhibits an uncommon and highly non-Arrhenius behavior, indicative of the formation of an activated complex (see Figure 2.8). Sivaramakrishnan et al. (2007) identified (R14) as a sensitive reaction in studying oxidation of H2/CO/O2/Ar mixtures in a high-pressure shock tube (up to 450 atm). They parameterized reaction (R14) by fitting experimental data and capturing the unusually narrow and deep minimum in the rate at about 1000 to 1200 K. However, the fit gives larger weight to the data *

Li et al. now distribute their 2007 model with the substitution of the parameters of You et al. (2007) for those of Mueller et al. (1999a).

42

Synthesis Gas Combustion: Fundamentals and Applications 1014

kR8(cm3/mol/s)

1014

1013

1013

0.4 0.5

1

1.5

0.6

2 2.5 1000/T(K–1)

0.8 3

1 3.5

4

Figure  2.8  Rate constants for HO2 + OH = H2O + O2 (R14); 3, Peeters and Mahnen (1973); 5, DeMore (1979); 9, Lii et al. (1980); G, Cox et al. (1981); -, Kurylo et al. (1981); ,, Braun et al. (1982); #, DeMore (1982); , Goodings and Hayhurst (1988); , Keyser (1988); /, Hippler and Troe (1992) (reevaluation of data from Hippler et al., 1990); A, Hippler et al. (1995); +, Kappel et al. (2002); =, Srinivasan et al. (2006); dashed line, rate expression of Sivaramakrishnan et al. (2007); solid line, fit by Chaos and Dryer (2008) considering the rate minimum measured by Hippler et al. (1995); dash-dot line, fit by Chaos and Dryer (2008) considering the rate minimum measured by Kappel et al. (2002). The figure insert provides details of data and rate fits at high temperatures.

of Hippler et al. (1995) rather than to the more recent, and arguably more reliable, measurements of Kappel et al. (2002) (Figure 2.8). The expression thus derived was able to improve predictions (using the model of Davis et al. (2005)) against the data of Sivaramakrishnan et al. (2007). Chaos and Dryer (2008) noted that the expression proposed by Sivaramakrishnan et al. (2007) could reach values over the collision limit at temperatures greater than about 1400 K. They performed least squares analyses by considering the data of Srinivisan et al. (2006) and Goodings and Hayhurst (1988), who showed very little temperature dependence of the rate for (R14) for temperatures greater than about 1300 K. Two fits were performed, taking into account the two different minima observed by Hippler et al. (1995) and Kappel et al. (2002). The resulting expressions are plotted in Figure 2.8. Li et al. (2007) performed CSP analyses of model results under the conditions of Sivaramakrishnan et al. (2007). Chaos and Dryer (2008) further extended these analyses to show that over the temperature range of interest (1200 to 1400 K) for the study of Sivaramakrishnan et al. (2007), the data only support the likelihood of a lower rate for reaction (R14). As pressure increases, reaction (R14) strongly competes with (R7) for HO2 radicals. Reaction (R7) is one of the major direct sources of OH at high pressures (especially near the extended second limit), thus determining the amount of OH available to react with CO through (R10). Therefore, reactions involving HO2 that compete for and generate OH radicals (such as reactions (R7) and

Syngas Chemical Kinetics and Reaction Mechanisms

43

(R14)) become more important at high pressures in the H2/CO system since the availability of OH determines the rate of CO oxidation. Sivaramakrishnan et al. (2007) noted that their data could be reconciled by increasing the rate of (R13), which is consistent with the above argument, as (R13) both consumes CO and releases OH; however, increasing this rate degrades model predictions against flow reactor data (Kim et al., 1994), and as discussed above, recent studies support a much lower rate for (R13). The update proposed by Sivaramakrishnan et al. (2007) as well as by Chaos and Dryer (2008) essentially lowers the rate of (R14) over the temperature range of interest (see Figure 2.8), thus slowing a termination path for both HO2 and OH. At higher pressures (>250 atm), (R14) is not as sensitive (Chaos and Dryer, 2008) since the flux of HO2 to form H2O2 through (R3) is considerably larger than through reaction (R14) and, at high temperatures, H2O2 decomposes rapidly to yield OH (R4), removing the significance of reaction (R14). Clearly, additional studies of (R14) are required; the lack of experimental data at intermediate temperatures (400 to 900 K), however, makes finding an expression that accurately captures the observed temperature dependence challenging. Moreover, reactions (R6) and (R7) continue to have large uncertainties associated with their rates and temperature dependences, which, as shown above, can have a direct effect in the determination of the oxidation characteristics of systems near the extended second limit.

2.4 High-Pressure/Low-Temperature Syngas Ignition and Kinetic Implications Ignition of H2/CO mixtures has traditionally been studied at relatively low pressures and high temperatures (e.g., Dean et al., 1978). These conditions correspond to the chain-branched or strong ignition regime shown in Figure 2.1. The need to establish an experimental database at conditions found in practical applications has recently led to considerable interest in syngas ignition at high pressures and relatively low temperatures. Perhaps the most representative of such studies is that of Petersen et al. (2007a), who reported new ignition delay data for syngas–air mixtures in a shock tube and a flow reactor. These new data are summarized in Figure 2.9, along with other recently published data from the rapid compression study of Walton et al. (2007a) and from an earlier high-pressure flow reactor study described by Peschke and Spadaccini (1985). All of the ignition delay data have been normalized to conditions of 20 atm pressure. Petersen et al. (2007a) noted that at reaction temperatures lower than about 1050 K, experimental observations of ignition delay, although consistent among the different apparatuses considered, all begin to differ considerably from kinetic model predictions generated using homogeneous, zero-dimensional, isochoric assumptions (i.e., constant internal energy, U, and volume, V) typically employed by kineticists (Figure 2.9). The above discussion raises considerable concern as these results bring into question the predominant fundamental kinetic understanding of gas phase oxidation phenomena for the H2/CO system. Hydrogen and carbon monoxide oxidation are the fundamental basis for all hydrocarbon combustion chemistry, and severe

44

Synthesis Gas Combustion: Fundamentals and Applications 107 P = 20 atm

Ignition Delay (µs)

106

2%/ms

Constant U, V

5%/ms

105

10%/ms

104 103 Petersen et al. (2007), shock tube Petersen et al. (2007), flow reactor Walton et al. (2007a), rapid compression Peschke and Spadaccini (1985), flow reactor Blumenthal et al. (1995), shock tube

102 101 100

0.7

0.9

1.1 1.3 1000/T(K–1)

1.5

Figure  2.9  Ignition delay times of syngas-air (Peschke and Spadaccini, 1985; Petersen et al., 2007a; Walton et al., 2007a) and hydrogen–air mixtures (Blumenthal et al., 1995). Conditions: 38.6% H2 + 51.1% CO + 10.3% CO2 + Air, ϕ = 0.5, 16.5 < P < 28.9 atm (Petersen et al., 2007a; shock tube), 11.9 < P < 23 atm (Peschke and Spadaccini, 1985); 50% H 2 + 50% CO + Air, 0.33 < ϕ < 0.6, 5.0 < P < 5.3 atm (Petersen et al., 2007a; flow reactor); (6.7 < H2 < 13.6%) + (4.5 < CO < 9.1%) + (16.2 < O2 < 18.6%) + (44.1 < N2 < 63.2%), 0.3 < ϕ < 0.7, 12 < P < 23.5 atm (Walton et al., 2007a); 15% H2 in air, 35 < P < 47 bar (Blumenthal et al., 1995). Filled and open circles correspond to strong and weak ignition events, respectively (Blumenthal et al., 1995). All experimental data have been normalized to 20 atm assuming proportionality to P–1. Lines show chemical kinetic computations using the model of Li et al. (2007) under constant U, V assumption as well as for different preignition linear pressure rise rates (see text). To improve clarity, modeled results are not shown for times greater than 10 s, although it is noted that constant U, V predictions can reach values of approximately 1000 s for the lowest temperatures (T ~ 630 K).

deficiencies in reaction rates or thermochemistry sufficient to be the cause of these disparities would be far reaching. Yet, the noted discrepancies between syngas ignition delay experimental results and gas phase kinetic predictions are repeatable and real, and thus, they impose limitations to design of combustion systems using syngas or hydrogen, or both. As noted in Section 2.2, carbon monoxide oxidation is strongly influenced by the presence of small amounts of hydrogen-containing species, including moisture, hydrocarbons, and most importantly, hydrogen itself (e.g., Yetter et al., 1991a). Subsequently, ignition phenomena for syngas-air mixtures will be similar in character to those for hydrogen–air mixtures. In fact, hydrogen shock tube ignition delay results exhibit similar disparities between predictions and observations and have been frequently discussed in the published literature since the early 1960s. This has been detailed by Chaos and Dryer (2008) and is shown in Figure 2.10, which shows experimental hydrogen shock tube ignition data (Blumenthal et al., 1995; Craig, 1966; Fujimoto, 1963; Martynenko et al., 2004; Slack, 1977; Snyder et al.,

45

Syngas Chemical Kinetics and Reaction Mechanisms

105

106

Fujimoto (1963)

104

103 H2/O2/Ar 2/1/7 P~2 atm

102

Ignition Delay (µs)

101 0.9 10

6

10

4

102

1

1.1

1.2

10 10

1

101 0.8

0.9

1

1.1

1.2

Martynenko et al. (2004) Stoichiometric H2/Air

4

10

103 102

P = 2 atm

1

10

2.1 < P < 3.86 atm 10.1 < P < 14.9 atm 0.7

0.8

0.9

1

1.1

5

10

104 P = 3 atm

Blumenthal et al. (1995) Wang et al. (2003)

103

101 10–1 0.5

H2/Air Φ=1 P~2 atm

102

10

Stoichiometric H2/O2 Voevodsky and Soloukhin (1965) P = 1 atm

10–1 103

103

5

100 3

Snyder et al. (1965) Craig (1966) Slack (1977)

105

104

102 0.7

0.9

1.1

101 1.3 1000/T(K–1)

15% H2 + 85% Air P~4 atm 0.9

1

1.1

Figure  2.10  Hydrogen shock tube ignition delays (symbols) from a number of experimental studies compared against chemical kinetic predictions (lines) using the model of Li et al. (2007). Modeling performed using the reported gas temperature and a constant U, V thermodynamic assumption.

1965; Voevodsky and Soloukhin, 1965; Wang et al., 2003) compared against model predictions. Note the remarkable similarity between Figure 2.10 and results reported by Petersen et al. (2007a) for syngas (Figure 2.9). In considering the observations above, Dryer and Chaos (2008) and Chaos and Dryer (2008) emphasized that the high sensitivity of induction chemistry to any type of experimental perturbations or nonidealities is what principally led to similarities in observations among the various experimental venues. Furthermore, they argued that disparities in observations and kinetic predictions were a result of the ideal

46

Synthesis Gas Combustion: Fundamentals and Applications

modeling assumptions applied and their inability to represent experimental conditions appropriately. In the particular case of shock tube observations, the disparities occur at temperatures below the extended second explosion limit, where mild ignition occurs (Strehlow and Cohen, 1962) and characteristic kinetic times are strongly influenced by induction chemistry and heat release involving HO2 and H2O2 reactions (see Section 2.2). Since the publication of the study of Chaos and Dryer (2008), several papers have drawn attention to the fact that significant pressure variations can occur during ignition in shock tubes at long test times (several milliseconds or longer). These revelations have significant impact on the arguments made with regard to syngas combustion as well as model validation based upon shock tube data presently in the literature. Below, further attention is brought to these issues, along with their implications in interpreting and modeling ignition delay data.

2.4.1 Shock Tube Pressure Histories Shock-induced self-ignition of undiluted fuel–air mixtures at the conditions of interest noted above are characterized by relatively long ignition delay times on the order of a few milliseconds. Ignition under these conditions is strongly coupled to the thermodynamic state and fluid dynamics of the gas behind the reflected shock wave. Ideally, the gas behind a reflected shock should be stationary and have uniform thermodynamic properties over the entire test volume. In the case of syngas-air mixtures, ideal analyses further predict that no significant preignition heat release can occur without substantial depletion of reactants, voiding the philosophical definition of ignition. Practically, however, even for dilute mixture studies, unavoidable nonidealities exist in the shocked gases, due to the ubiquitous presence of boundary layers. Incident shock attenuation, boundary layer growth, and shock wave–boundary layer interactions lead to nonuniform pressures and temperatures behind the reflected shock that gradually increase with time (Davidson and Hanson, 2004; Kahandawala et al., 2006; Petersen and Hanson, 2001). Furthermore, residual gas velocities may exist behind the reflected shock wave that further lead to the establishment of pressure gradients along the shock tube axis (Frenklach et al., 1984; Michael and Sutherland, 1986). These nonidealities can be minimized by using large-diameter shock tubes, dilute fuel-oxidizer mixtures in monoatomic gases, and short test times (less than about 500 µs). The latter two options, however, are not applicable to shock tube experiments that study higher concentration fuel–air mixtures of direct relevance to energy conversion applications. For such highly reactive, exothermic mixtures, the above phenomena are further compounded by the fact that at high pressures and lower temperatures, ignition in reflected shock tube regions is inhomogeneous (Terao, 1977), initiating at localized hot spots in a deflagrative manner, progressing eventually to detonation (Blumenthal et al., 1996; Fieweger et al., 1997; Wang et al., 2003). An example of this behavior is shown in Figure 2.11, which contrasts ignition processes during mild and strong combustion. Prior to the main ignition event, the finite exothermicity of the initial, localized, heterogeneous reaction sites induces further increases in pressure, which

47

Syngas Chemical Kinetics and Reaction Mechanisms

(a)

(b)

Figure 2.11  Shadowgraphs of the ignition process of 15% H2 + 85% air mixtures in a shock tube. (a) Mild ignition: 7.7 atm, 977 K, 100 µs separation between frames; (b) strong ignition: 3.4 atm, 1096 K, 40 µs separation between frames. Note that these conditions lie on either side of the extended second limit (see Figure 2.1). (Images have been adapted from Blumenthal et al., 1996. With permission.)

are quickly equalized throughout the gases behind the reflected shock due to the high speed of sound. These features are typical of the well-known mild ignition process. For nonreactive mixtures, Petersen and Hanson (2001) reported temperature increases (due to observed pressure gradients behind the reflected shock) of up to 40 K over 500 µs for representative pressure and temperature ranges of 24 to 530 atm and 1275 to 1900 K, respectively. For conditions relevant to practical fuels (i.e., 25 to 50 atm and 700 to 1000 K) and due to reactivity and the likelihood of much longer ignition times, more severe pressure and temperature effects are to be expected. It should be noted, however, that the magnitude of these changes also depends on the shock tube diameter (Petersen and Hanson, 2001), being less severe (but still considerable) as the tube diameter is increased. From the discussion above, it is clear that changes in the thermodynamic state of the gas behind the reflected shock wave are likely to be significant and, thus, can have pronounced effects on kinetic observations as well as the ignition process. Figure 2.12 shows the pressure-time history record during ignition of a sample syngas mixture studied by Reehal et al. (2007). A preignition pressure increase of nearly 50% over the reported nominal reflected shock pressure can be observed (Figure  2.12b). In such instances, kinetic modeling of the ignition delay process under the common assumption that the shock tube behaves as a constant-volume system with constant internal energy is no longer reasonable. Approximate homogenous modeling can be performed if individual experimental pressure histories are known for each test condition; however, this information rarely accompanies reported shock tube ignition delay measurements. With some frequency, an exemplar experimental

48

Synthesis Gas Combustion: Fundamentals and Applications

100

A

80

Syngas, P = 13.2 atm, T = 1022 K 18.01% H2 0.22% Ar 1.03% CO 16.13% O2 0.89% CH4 61.37% N2 2.36% CO2

60

Pressure (atm)

40 20 0 30

B

20

6 atm/ms

10

0 0

200

400 600 Time (µs)

800

1000

Figure 2.12  Pressure profile during syngas ignition obtained from Reehal et al. (2007). Panel (b) zooms into the profile shown in (a) to show preignition pressure rise.

pressure profile is presented, but generally the ordinate scaling chosen (large enough to capture pre- and postignition pressure values) is such that pressure history prior to ignition is not known with sufficient accuracy (Figure 2.12a). Even though the aforementioned nonidealities present in shock tubes at the conditions of interest have been known for some time and noted by shock tube experimentalists (e.g., Blumenthal et al., 1996; Fieweger et al., 1997; Petersen and Hanson, 2001), recent shock tube studies of fuel–air mixtures have failed to consider these phenomena in interpreting the reported data (e.g., Petersen et al., 2007a). Furthermore, because of the lack of pressure-time history information noted above and the established, common practice to assume constant U, V in modeling shock tube data, the experimental and general chemical kinetic modeling communities (including ourselves) have ignored the implications to model validation. The issues outlined above have recently begun to receive more attention, as shock tube experimentalists continue to understand and attempt to determine early ignition events at lower temperatures and for high-energy density mixtures (Li et al., 2008; Pang et al., 2009; Petersen et al., 2007b; Shen et al., 2009). In these studies ignition

Syngas Chemical Kinetics and Reaction Mechanisms

49

delay data are further examined and pressure variations prior to ignition are identified and taken into account when interpreting and modeling the data. Their implications into how ignition delay chemical kinetic modeling should be approached are discussed below.

2.4.2 Modeling Approaches In several previous studies, it has been suggested that preignition pressure variations in reflected shock tube observations can be approximated by assuming they are polytropic processes (Fieweger et al., 1997; Petersen and Hanson, 2001). Recently, some investigators (e.g., Pang et al., 2009) have used pressure histories and associated temperatures deduced from isentropic relations to adjust ignition data affected by preignition pressure rise, while others (e.g., Petersen et al., 2007b) have opted to compute an effective or average pressure and temperature based on the observed pressure history and plot the data with respect to these adjusted values. The presence of dynamic pressure features prior to ignition, however, can have considerable effects on radical initiation processes, as discussed in Mittal et al. (2008). Therefore, it is important to include measured pressure histories in modeling approaches rather than using averaged or effective values. Li et al. (2008) proposed a specific modeling tool for shock tube applications, CHEMSHOCK, which allows for the treatment of time-varying pressures coupled with chemical kinetics. CHEMSHOCK solves the coupled energy and chemical species system of differential equations using a two-step process: at every time step the system is first solved assuming constant U, V conditions, and then pressure and temperature are adjusted isentropically to match the measured pressure profile while keeping the chemical composition fixed. Pang et al. (2009) showed promising results using CHEMSHOCK in modeling ignition delay data of H2/O2 mixtures. Alternatively, SENKIN (Lutz et al., 1987), part of the CHEMKIN II (Kee et al., 1989) suite of programs, is capable of embodying the same assumptions with similar generality. A time-dependent polytropic compression (or expansion) results in both volume and density changing with time. In contrast to the CONP (constant enthalpy and pressure) and CONV (constant internal energy and volume, commonly used for shock tube modeling), the VTIM (i.e., volume as a function of time) option in SENKIN can be employed to emulate a time-dependent polytropic state change. The user must only provide time functions for variations in the specific volume of the system. Knowing the functional variation of pressure from measured data and assuming isentropic compression/expansion, the functions to be provided are





1  P (t )  ν (t ) =   ρ0  P0 

−1

dν 1 ν (t ) dP =− dt γ P (t ) dt

γ

50

Synthesis Gas Combustion: Fundamentals and Applications

Ignition Delay (µs)

where ν is the specific volume, P the measured pressure, γ the specific heat ratio, and P0 and ρ0 the initial pressure and density, respectively, behind the reflected shock wave. This approach has been successfully used in modeling rapid compression machine ignition processes (Mittal et al., 2008), while Petersen and Hanson (2001) used a similar methodology to account for observed pressure variations in nonreacting mixtures behind reflected shock waves. Using this approach, the syngas data of Reehal et al. (2007) were further considered. It was assumed that preignition pressure rise, on average, can be approximated by a linear positive gradient, as evidenced by Pang et al. (2009). It is noted, however, that for high reactant concentrations (i.e., fuel–air mixtures) and due to the mild ignition processes discussed above, observed pressure rises may deviate from this linearity (e.g., see Figure 2 in Petersen et al., 2007b). As shown in Figure 2.12b, an approximate rise of 6 atm/ms can be applied to the data. Figure 2.13 plots the dataset for the mixture shown in Figure 2.12, along with model predictions using a constant U, V assumption, as well as VTIM with the pressure rise shown in Figure 2.12b. For the shortest test time, the constant U, V model predictions agree reasonably with the data, but at longer test times, the gradual pressure (and temperature) increase results in considerable departure of predictions and measurements, as noted above. In this case, Figure  2.13 shows that the constant U, V approach quickly fails to provide a good estimate of ignition delay at lower temperatures. SENKIN modeling with VTIM assuming a linear pressure profile provides much improved predictions. The above results are encouraging and warrant consideration in interpreting the syngas data of Petersen et al. (2007a). It is noted that Petersen et al. (2007a) did

104

103

102 0.85

0.9

0.95

1

1000/T(K–1)

1.05

1.1

Figure 2.13  Syngas ignition delay data (symbols) and predictions using the model of Li et al. (2007). Data are those of Reehal et al. (2007) for the mixture shown in Figure 2.12 at an average pressure of 11.8 atm. The dashed line corresponds to modeling assuming constant U, V; the solid line is the result of calculations using the pressure gradient shown in Figure 2.12b and employing CHEMKIN-VTIM (see text). The gray symbol corresponds to the case shown in Figure 2.12.

Syngas Chemical Kinetics and Reaction Mechanisms

51

not discuss the characteristics of the pressure signals associated with their shock tube data; however, in recent work (Mertens et al., 2008; Reehal et al., 2007) these researchers have shown that syngas shock tube ignition can exhibit considerable pressure variations. Furthermore, emission from the hydroxyl radical collected at the shock tube’s endwall (Reehal et al., 2007) indicates that these data show severe mild ignition characteristics that, as explained above, in conjunction with boundary layer effects lead to compression of the test mixture over the ignition delay period. In fact, Figure  2.9 also shows shock tube data from the H2-air ignition study of Blumenthal et al. (1995), who were able to identify and perform temporal measurements on the appearance of flame kernels and subsequent transition to detonation in the mild ignition regime. Cases that exhibited mild ignition processes are identified in Figure 2.9 as open circles; note that these closely agree with the shock tube data from Petersen et al. (2007a), which considerably deviate from constant U, V model predictions. Since pressure profiles for the data reported by Petersen et al. (2007a) are not available, Figure 2.9 shows results using assumed linear pressure gradients of 2, 5, and 10%/ms (i.e., 0.4, 1, and 2 atm/ms) prior to ignition. These values may be conservative, as the presence of mild ignition events will certainly change pressure signals considerably; for example, Figure 2.12b shows a pressure gradient of nearly 50%/ms. It can be seen that these simplified assumptions yield good agreement with the data over the entire temperature range. However, the accuracy of the measured pressure histories in the preignition region as well as the validity of the isentropic compression assumption produce additional uncertainties that must be considered in developing and validating models utilizing such data.

2.4.3 Other Systems One of the more intriguing points raised by data in Figure 2.9 is that data collected in flow reactor and rapid compression venues apparently follow the trend established by the shock tube data in the mild ignition regime. Given that nonidealities present in shock tubes (discussed above) do not apply to these experiments, the question remains as to what causes accelerated ignition in these systems. Chaos and Dryer (2008) suggested that both flow reactors and rapid compression measurements can suffer similarly from other experimental perturbations. The strong sensitivity of chemical induction processes in the thermal-chain explosive regime to any source of perturbation leads to reductions in the overall ignition delay period in the mild ignition region (see Figure 2.9), similar to those observed for shock tube studies. The presence of reaction fronts (Walton et al., 2007b) as well as particles (Elsworth et al., 1969; Haskell, 1970) may influence rapid compression studies. On the other hand, catalytic processes may affect flow reactor measurements (Chaos and Dryer, 2008). Similar to the effects of NOX on combustion systems discussed above, all these perturbations act to remove the rate-limiting nature of reactions below the explosion limit, and the exothermicity of the system is increased, as detailed in Section 2.2. This rapidly drives the reaction across the explosion limit, considerably shortening induction and, therefore, ignition delay time. The multiplicity of and the strong sensitivities of induction chemistry to experimental perturbations are very difficult to characterize accurately in research experiments

52

Synthesis Gas Combustion: Fundamentals and Applications

and are unlikely to be controlled in real engineering applications. The implications for developing lean premixing schemes for advanced syngas combustion applications are that designs should accommodate the likely presence of perturbations and their effects on ignition delay through, for example, empirical relations (Beerer and McDonell, 2008), if stimulated flashback into the mixing region is to be precluded.

2.5 Premixed Flame Propagation in High-Pressure Media One parameter of great importance in the design of gas turbine combustors using syngas or hydrogen is the laminar burning velocity, and in turn, the turbulent mass burning rate. Measurements of syngas burning velocities have been mostly studied at atmospheric conditions (e.g., Hassan et al., 1996; McLean et al., 1994). A number of recent advancements in flame speed measurement methodologies have resulted in a substantial extension in the range of initial parameters and mixture compositions that can be studied. Tse et al. (2000) devised a constant-pressure dual-chambered cylindrical bomb apparatus, extending the pressure range for the stretched spherical flame method by nearly an order of magnitude. The device enabled H2 and H2/CO flame speed measurements up to 40 atm (Tse et al., 2000; Sun et al., 2007). Bradley et al. (2007) conceived a method to extract planar burning velocities from wrinkled outwardly propagating flames and employed the technique to measure burning velocities of lean H2-air flames up to 10 atm. Natarajan et al. (2007) demonstrated that the Bunsen flame approach for flame speed measurement yields values that are consistent with other, more widely accepted methodologies. Flame speeds were determined using conical H2/CO/CO2/He flames at pressures to 15 atm and preheat temperatures to 700 K (Natarajan et al., 2007, 2009). In a study of transient effects arising from ignition in a homogeneous mixture, Chen et al. (2009a) found that unsteady and nonlinear effects are substantial for small, spherical flames of nonunity Lewis number. Burke et al. (2009b) and Chen et al. (2009b) showed that flame speeds observed close to the wall in cylindrical and spherical chambers are systematically lower due to confinement of the flow field. Correction techniques were developed that extend the flame radius range that can be used for accurate determinations of flame speed, allowing for burning velocity measurements of dilute mixtures with prolonged ignition transients. The techniques of Burke et al. (2009b) and Chen et al. (2009a, 2009b) permitted experimental study on the burning rates of H2/O2/CO/CO2/diluent flames of temperatures from 1500 to 1800 K in a cylindrical bomb up to 25 atm (Burke et al., 2007, 2009a). Similar to the recent ignition delay observations discussed in the previous section, the above flame speed studies extend results significantly beyond previous works with which modeling results presently in the literature were compared. Furthermore, it appears that current kinetic models result in predictions that fail to replicate these recent data (Figure 2.14). Figure  2.14 shows that there exists considerable scatter in available atmospheric pressure measurements (Figure  2.14a) (Burke et al., 2007; Hassan et al., 1997; McLean et al., 1994; Sun et al., 2007), especially at rich conditions. A similar trend is also observed for high-pressure H2/CO flame conditions (Figure 2.14b) (Burke et al., 2009a; Sun et al., 2007). Inconsistencies that may be present in data

53

Syngas Chemical Kinetics and Reaction Mechanisms

Laminar Burning Velocity (cm/s)

200 160 120 80 40 0

Laminar Burning Velocity (cm/s)

Burke et al. (2007) Hassan et al. (1997) McLean et al. (1994) Sun et al. (1994) 1

0

2 3 Equivalence Ratio (a)

4

5

Burke et al. (2009a) Sun et al. (2007)

100

10 atm

80

20 atm

60

40 1

2 3 Equivalence Ratio (b)

4

Figure  2.14  Laminar burning velocities of syngas mixtures. (a) H2:CO (1:1) in air at 1 atm; (b) H2:CO (1:3) + O2:He (1:7) mixtures at high pressure. Symbols are experimental measurements. The solid line shows predictions from the model of Li et al. (2007); the dashed line are results assuming a Fe(CO)5 content of 200 ppm in the CO source (using the kinetic subset of Rumminger et al., 1999).

processing methodologies to obtain laminar burning velocities (e.g., Burke et al., 2007, 2009b; Chen et al., 2009b) cannot fully explain the observed disparities. Sun et al. (2007) noted these discrepancies and developed a syngas kinetic model that exhibited improved predictions against high-pressure H2/CO flames. Sun et al. also demonstrated acceptable model performance against other H2/CO targets (Dean et al., 1978; Fotache et al., 2000; Mueller et al., 1999b; Yetter et al., 1991b). However, as shown in Figure 2.15, model updates implemented by Sun et al. (2007) seriously

54

Synthesis Gas Combustion: Fundamentals and Applications

Laminar Burning Rate (g/cm2/s)

0.4

0.3

10 atm 15 atm 20 atm

0.2

0.1 0.8

1.2 1.6 Equivalence Ratio

2

2.4

Figure 2.15  Measured and calculated unstretched laminar mass burning rates for H2 + (O2:He, 1:11.5) flames at 10, 15, and 20 atm. Symbols are experimental data (Tse et al., 2000); solid lines are predictions using the model of Sun et al. (2007); dashed lines are predictions using the model of Li et al. (2007).

degrade the quality of predictions against high-pressure pure hydrogen burning velocities. On the basis of the above discussion, Chaos and Dryer (2008) considered the possibility of experimental contamination effects due to unintentional addition of iron pentacarbonyl, Fe(CO)5. For laboratory combustion experiments, many researchers have employed high-pressure carbon-steel cylinders as the source of the fuel. CO can readily react at high pressure with metals present in steel to form carbonyls, especially Fe(CO)5. Depending on handling, steel CO cylinders obtained from commercial suppliers with initially little or no contaminants are prone to contamination by Fe(CO)5 over time (Tepe et al., 1999). An example of experimental contamination by Fe(CO)5 is the study of Williams and Shaddix (2007), who recently reported observing wall deposits formed when operating a swirl-stabilized combustor running on simulated syngas-air mixtures; these deposits were found to be mostly iron oxides originating from Fe(CO)5 in the CO source used. Metallic compounds have been shown to have strong flame inhibition effects (Lask and Wagner, 1960; Linteris et al., 2008; Reinelt and Linteris, 1996; Rumminger and Linteris, 2000, 2002; Rumminger et al., 1999; Vanpee and Shirodkar, 1978). When present in premixed flames, iron pentacarbonyl can reduce the burning velocity considerably (Lask and Wagner, 1960; Linteris et al., 2008; Reinelt and Linteris, 1996; Rumminger and Linteris, 2000). The presence of chromium, nickel, and iron carbonyls has also been shown to accelerate carbon monoxide and hydrogen oxidation in shock tubes (Izod et al., 1972; Linteris and Babushok, 2009; Matsuda, 1972a, 1972b). For the conditions of the study of Petersen et al. (2007a), however, the presence of Fe(CO)5 is not sufficient (Chaos and Dryer, 2008; Linteris and Babushok, 2009) to explain the results discussed in the section above. Finally, in gas turbines used in

Syngas Chemical Kinetics and Reaction Mechanisms

55

commercial IGCC installations, the formation of large iron deposits on nozzle hardware has been noted. These deposits lead to restriction/alteration of the airflow into the combustor, overheating parts of the combustion chamber. These iron deposits have been reported to originate from Fe(CO)5 introduced into the chamber after CO in the syngas used reacted with steel pipes (García, 2006). Furthermore, iron pentacarbonyl is a well-established challenge in the development of gasification technologies (Rezaiyan and Chereminisoff, 2005). To further investigate the effects of Fe(CO)5 contamination on H2/CO burning velocities, Chaos and Dryer (2008) added the Fe(CO)5 kinetic rate correlations and transport properties developed by Rumminger et al. (1999) to the model of Li et al. (2007) and performed calculations for the flame conditions of Figure  2.14. In the computations, it was assumed that the CO source contained 200 ppm of Fe(CO)5 so that the overall maximum iron pentacarbonyl mixture concentrations for a wide range of equivalence ratios were always less than 75 ppm, approximately. Figure 2.14 compares the model-predicted burning velocities for pure fuel (solid lines) and for mixtures using CO contaminated by 200 ppm of Fe(CO)5 (dashed lines) with experimental data. The figure confirms that Fe(CO)5 has a very noticeable effect on rich flames, in qualitative agreement with the onset of the experimental disparities found in the literature. Experimental data, however, are needed to conclusively determine contamination by iron pentacarbonyl. As part of the present study, the CO cylinder used in the experiments of Burke et al. (2007) was analyzed by Fourier transform infrared spectroscopy means. A Nicolet Magna IR 560 spectrometer was used for the tests. The instrument consisted of a 2.0 L gas cell with an optical path length of 10.0 m maintained at 100°C and 0.07 atm (~1 psia). Rather than trying to handle Fe(CO)5 (due to its high toxicity) to perform a calibration, pure CO was obtained from Airgas, stored in an aluminum cylinder to prevent formation of iron pentacarbonyl. The gas from this cylinder was used to establish a background signal. Absorption spectra from the CO gas stored in the steel cylinder used by Burke et al. (2007) were then recorded by averaging 32 scans at a resolution of 0.25 cm–1 over a 600 to 4000 cm–1 wave number range. The results of the tests are shown in Figure  2.16; the typical rovibrational CO band heads (R and P branches) do not appear in the spectra as pure, Fe(CO)5-free carbon monoxide was used as background. However, strong absorption bands are evident, centered around 2013 and 2033 cm–1. These are positioned at relatively higher frequencies than the CO rovibrational frequencies and can be ascribed to the presence of gas phase Fe(CO)5 (McDowell, 1971; Myrstad and Fredriksen, 1998; Rao et al., 1989), as it only has two CO stretching modes that are IR active (Rao et al., 1989). Based on the absorbance values measured and using the extinction coefficients reported by McDowell (1971) for the 2013 and 2033 cm–1 bands, consistent concentration values of 25.7 and 24.3 ppm, respectively, were calculated. This contamination level is approximately an order of magnitude lower than that used to generate the plots in Figure 2.14. Therefore, the data of Burke et al. (2007) can be considered free from iron pentacarbonyl effects. The discussion above motivated additional consideration of the important kinetic pathways affecting the propagation of flames at high pressure, particularly for diluted

56

Synthesis Gas Combustion: Fundamentals and Applications 0.5

Absorbance

0.4 2014 cm–1

0.3 0.2 0.1 0 2100

2033 cm–1

2075

2000 2050 2025 Wave Number (cm–1)

1975

1950

Figure  2.16  Infrared spectra of the carbon monoxide gas used in the experiments of Burke et al. (2007). The main absorption bands of Fe(CO)5 are identified.

and low flame temperature conditions. Such conditions are of considerable significance, as combustion dilution is essential to controlling NOX emissions in syngas combustion applications. Burke et al. (2009a) performed experimental and modeling studies of H2/CO/CO2 flames of temperatures from 1500 to 1800 K at high pressures. Significant disagreement among the model predictions as well as with experimental data were noted to correlate strongly with high pressure and lower flame temperature conditions where the reactive portion of the flame was restricted to a smaller, higher-temperature window. The discrepancies were found to exist even for pure hydrogen flames. Preliminary experimental data from Burke et al. (2009a) and model predictions (Li et al., 2007; Davis et al., 2005; Sun et al., 2007; Konnov, 2008) for the mass burning rate of both lean and rich H2/O2/diluent mixtures are shown in Figure 2.17. The experimental data show a negative pressure dependence of the burning rate for pressures above ~15 atm for these low flame temperature conditions. Negative pressure dependence of the burning rate for syngas flames could have interesting consequences for thermoacoustic stability behavior in engines. While the predictions of Li et al. (2007) and Davis et al. (2005) reproduce experimental results well to about 10 atm, no model reproduces the new measurements for pressures above 10 atm across all equivalence ratios. Large disparities in predicted burning rates at higher pressures (up to a factor of 2) are observed among the models and against the experimental data. The disparity in predictions yielded from the various models is particularly noteworthy, as nearly the same target data were used in developing and validating each of the models. Analyses conducted using the model of Li et al. (2007) indicate that sensitivity of mass burning rate predictions to elementary rates increases considerably with pressure, as demonstrated in Figure 2.18. The reactions, to which the pressure dependence for rich conditions is most sensitive, are primarily reactions that involve H

57

Syngas Chemical Kinetics and Reaction Mechanisms

Mass Burning Rate (g/cm2/s)

0.16 0.12 0.08 0.04 0

0

20

10

30

Pressure (atm) (a)

Mass Burning Rate (g/cm2/s)

1.2

0.8

0.4

0

0

10 20 Pressure (atm)

30

(b)

Figure  2.17  Laminar burning velocities of hydrogen mixtures at constant flame temperature (~1600 K) for various pressures: (a) H2/O2/He (1.7/1/11.5) mixture, ϕ = 0.85; and (b) H2/O2/Ar (5/1/9.5) mixture, ϕ = 2.5. Symbols are experimental measurements from Burke et al. (2009a) (open circles) and Tse et al. (2000) (open triangles), lines are model results: Li et al. (2004, 2007), bold solid line; Davis et al. (2005), bold dashed line; Sun et al. (2007), solid line; Konnov (2008), dashed line.

atoms, namely, (R1), (R2), (R6), and (R7) (Figure 2.18b). Rates of these four reactions define the extended second explosion limit, as discussed above. The reactions that appear to govern the pressure dependence for lean conditions are the reactions above in addition to reactions involving OH radicals, namely, (R14) and H2 + OH = H2O + H (R15) (Figure 2.18a). Other reactions include O + H2 = H + OH (R16) and H + OH + M = H2O + M (R17). Reaction (R16) is a branching reaction that becomes increasingly important at lean conditions, as more O radicals are produced due to

58

Synthesis Gas Combustion: Fundamentals and Applications

1 atm

HO2 + H = H2 + O2 (R6)

10 atm 20 atm

H + OH + M = H2O + M (R15) O + H2 = H + OH (R14) HO2 + OH = H2O + O2 (R12) HO2 + H = OH + OH (R7) H2 + OH = H2O + H (R13) H + O2 (+M) = HO2 (+M) (R2) H + O2 = O + OH (R1) –0.8

O + H2 = H + OH (R14)

–0.4 0 0.4 Sensitivity Coefficient (a)

0.8

1 atm 10 atm

HO2 + OH = H2O + O2 (R12)

20 atm

H + OH + M = H2O + M (R15) H2 + OH = H2O + H (R13) HO2 + H = H2+O2 (R6) H + O2(+M) = HO2 (+M) (R2) HO2 + H = OH + OH (R7) H + O2 = O + OH (R1)

–0.8

–0.4 0 0.4 Sensitivity Coefficient (b)

0.8

Figure 2.18  Sensitivity of mass burning rates to rate parameters as a function of pressure for (a) H2/O2/He mixtures (Figure 2.17a) and (b) H2/O2/Ar mixtures (Figure 2.17b).

59

Syngas Chemical Kinetics and Reaction Mechanisms

the presence of excess oxygen. Even at lean conditions, however, reaction (R1) is the dominant branching reaction (Figure 2.18). Reaction (R17) is highly exothermic and has been previously shown to affect predictions of laminar burning rates (Li et al., 2004); in fact, Li et al. (2004) modified the rate of (R17) to improve model predictions at flame conditions. Uncertainties in collision parameters of species stabilizing the recombination of H and OH in (R17) remain and have been the subject of recent studies (Sellevåg et al., 2008; Srinivasan and Michael, 2006). At increasing pressures, the OH radical is present in higher concentrations and plays a larger role in the kinetic pathways for lean conditions. Reaction (R14) competes with (R6) and (R7) for HO2 radicals and inhibits the overall kinetics by converting two radicals to stable species. As discussed above, the reaction rate of (R14) has an uncommon and highly non-Arrhenius temperature dependence (see Figure  2.8). Temperature window sensitivity analyses (Zhao et al., 2005) of burning rate to (R14) for the lean H2/O2/He mixture of Figure 2.17a were conducted (Burke et al., 2009a). The results (shown in Figure  2.19) reveal that the temperature window of peak sensitivity to (R14) at pressures above 10 atm is almost exclusively in the deep well of the rate correlation (around 1000 K; see Figure 2.8). Therefore, accurate characterization of the rate of (R14) in this region is critical to predicting the burning rate of lean syngas mixtures at pressures of practical significance. Indeed, simulations performed using the model of Li et al. (2007) substituting the two rate expressions proposed by Chaos and Dryer (2008) for (R14) yield substantially different results at pressures above 10 atm. Substitution of the fit based on the data of Hippler et al. (1995), which exhibit a local minimum at ~1200 K, yields substantially higher values than those shown in Figure 2.17, whereas substitution of the fit based on Kappel et al. (2002)

Temperature Sensitivity Coefficient

0 1 atm

–0.005

10 atm

–0.01

–0.015

20 atm

–0.02 0.5

1

1.5 1000/T(K–1)

2

2.5

Figure  2.19  Pressure dependence of temperature window sensitivities of burning rates to HO2 + OH = H2O + O2 (R14) for the lean H2/O2/He mixture of Figure 2.17a. The laminar burning rate exhibits strong sensitivity to the reaction in the location of the local rate minimum of this reaction (see Figure 2.8).

60

Synthesis Gas Combustion: Fundamentals and Applications

data, which exhibit a local minimum at ~1000 K, has little effect on the predictions. Additionally, direct substitution of recent rate expressions for (R2) and (R17) proposed by Srinivasan and Michael (2006), Sellevåg et al. (2008), and Pang et al. (2009) into the mechanism of Li et al. (2007) does not substantially improve, and in some cases degrades, predictions. Direct substitution of the proposed rate expression from Michael et al. (2000) for (R6) yields improvements to the predictions, but does not completely reconcile differences between the models and experimental data. At present it appears that under fuel-rich conditions, the relative rates of HO2 + H = H2 + O2 (R6) and HO2 + H = OH + OH (R7) must be significantly revised to bring predictions into reasonable consistency with experimental results. As these reactions remain important even under fuel-lean conditions, further refinements of predictions involving HO2 + OH = H2O + O2 (R14) are dependent on them. Additional experimental and theoretical analyses on reactions (R6), (R7), and (R14) are needed to support resolution of models at conditions important to syngas combustion at high pressures. The effect of pressure on the flame structure based on flux of H radicals through the competing pathways, (R1) and (R2), and H radical mole fraction using the model of Li et al. (2007) is illustrated in Figure  2.20 (Burke et al., 2009a). Figure  2.20 also shows the temperature of the extended second limit for the given pressures (see also Figure 2.1). The results indicate that with increasing pressure, the temperature of extended second limit increases accordingly, moving significantly toward the postflame region for lean and rich mixtures. Therefore, the portion of the flame

0.002

0.004

0 10 atm

0.12

0.0008

0.08

0.0004

0.04

0 0.0006

0

20 atm

0.2

0.0004

0.1

0.0002 0

400

800 1200 Temperature (K)

(a)

0 1600

Mole Fraction

Mole Fraction

0

0.004

0.01

0.001

0.002

0.006

1 atm

0.02

0 0.006

10 atm

0.4 0.3

0.004

0.2

0.002 0 0.003

0.1 0 1.2

20 atm

0.002

0.8

0.001

0.4

0

  

0.002 0

400

1200 800 Temperature (K)

Flux (mole/cm3/s)

0.006

0.0012

0.03

0.003

1 atm

Flux (mole/cm3/s)

0.01 0.008

0 1600

(b)

Figure  2.20  Comparison of relevant aspects of the flame structure at various pressures for (a) H 2/O2/He mixtures and (b) H 2/O2/Ar mixtures at the conditions shown in Figure 2.17. Black lines represent H radical mole fraction, and gray lines show the flux of H radicals through H + O2 = O + OH (R1) (solid line) and H + O2(+M) = HO2(+M) (R2) (dashed line). The temperature of the extended second limit temperature is marked by the vertical lines.

Syngas Chemical Kinetics and Reaction Mechanisms

61

that undergoes strong branching kinetics (i.e., at temperatures above the extended second limit) is reduced to a smaller temperature window at higher pressures. The peak H radical mole fraction is decreased substantially and also moves toward the postflame region with increasing pressure. Furthermore, the reaction zone region in which (R1) and (R2) compete is restricted to higher temperatures and a smaller overall temperature range as pressure is increased. Similar shifts toward smaller, higher-temperature windows occur in the flux profiles of other controlling reactions, including (R6) and (R7). Such a restriction of the strongly reactive portion of the flame to higher temperatures can be expressed as an increase in the overall activation energy or Zeldovich number of the mixture with pressure. Comparing Figure 2.20a and b, the responses of the flame structure to pressure are different for different equivalence ratios. The peak fluxes of the H radical through (R1) and (R2) occur closer to the unburned region in rich flames, where the concentration of oxygen is higher. Since the peak flux of the H radical occurs at lower temperatures, where (R2) is favored relative to (R1), the peak flux through (R2) relative to (R1) is higher than that at lean conditions. At both lean and rich conditions, the mole fraction of HO2 compared to H is substantially higher at higher pressures, as a much larger portion of the flame is below the extended second limit. Consequently, pathways involving HO2 play a much larger role in the overall kinetics. Reactions involving HO2 with H and OH (and possibly even O for very lean conditions) exhibit higher sensitivities to the burning rate at higher pressures. Additionally, since the reactive portion of the flame is restricted to higher temperatures at high pressures, the peak sensitivity to these reactions occurs at higher temperatures (e.g., Figure 2.19). Since (R6), (R7), and (R14) continue to have large uncertainties in their rates and temperature dependences, as discussed above, these reactions deserve further attention, particularly at higher temperatures (above 1000 K). In general, high-pressure syngas as well as hydrogen flames emphasize HO2 kinetics and radical-radical recombination reactions such as (R17). Additionally, most of the flame reactivity, i.e., branching) occurs in a relatively small portion of the flame as, for increasing pressures, the extended second limit shifts to higher temperatures. Since many new techniques have been developed in the past decade that extend the pressure and temperature ranges of flame studies and recent interest in syngas combustion is focused on conditions relevant to gas turbines, further investigations of high-pressure and temperature syngas flame kinetics are surely forthcoming. As many of the recent studies (Tse et al., 2000; Sun et al., 2007; Natarajan et al., 2009; Burke et al., 2009a) have motivated changes to the models or highlight model discrepancies with experimental data, substantial improvements in modeling flames at gas turbine conditions will likely occur over the next decade.

2.6 Conclusion Syngas mixtures are an increasingly important commodity in developing low­polluting solutions to energy security and resources in a carbon-constrained world. Interest in syngas combustion has inspired an extension of the existing experimental validation resources to considerably higher pressures and lower temperatures that are sufficient for testing the comprehensive nature of existing detailed chemical kinetic

62

Synthesis Gas Combustion: Fundamentals and Applications

models. The present chapter has considered some of the unique properties of syngas combustion kinetics, summarized a number of recent theoretical and experimental advancements on the fundamental understanding of important elementary reactions, reviewed new experimental efforts to provide validation data for high-pressure, high-energy-density kinetic behavior, and discussed the kinetic issues that relate to improving predictions of these new data. The higher-pressure, lower-temperature conditions encountered in practical syngas applications point to the importance and the need of further theoretical as well as experimental studies of elementary reactions involving HO2 and H2O2. We note that recent reevaluations of the reaction rate for CO + HO2 = CO2 + OH (R13) are primarily important to ignition delay measurements at high pressures due to modifications of induction chemistry, and have little influence on postinduction observations. It appears that modifications in the reaction of HO2 + OH = H2O + O2 (R14) improve comparisons of predictions with highpressure oxidation, but recommended rate correlations for this reaction should be modified to avoid exceeding collisional rates at high temperatures. We show that recent investigations of high-pressure ignition and flame propagation of H2/CO mixtures should be cautiously evaluated prior to implementing any changes to improve the agreement of predictions from chemical kinetic models. H2/CO ignition measurements in shock tubes, rapid compression machines, and flow reactors can exhibit aberrations at low temperature and high pressures that cause observations to considerably differ from homogenous gas phase predictions. Although phenomena characterized as mild ignition in hydrogen–oxygen shock tube experiments are historically well known and can even be kinetically differentiated from strong ignition observations, the sources of chemical induction perturbations that lead to their manifestation are not understood in quantitative detail. Behavior of syngas mixtures under similar conditions is derived almost entirely as a result of perturbations of hydrogen–oxygen chemical induction kinetics, with only minor differences from the presence of carbon monoxide. Numerous processes can affect induction chemistry (i.e., compressible flow, physical mixing, impurities, and catalytic surface coupling), and it is likely that even in shock tubes, no one cause is universally responsible for the mild ignition observations. Moreover, that multiple sources of chemical induction perturbations can all lead to similar magnitudes of reduction in ignition delay in the mild ignition regime is the reason that observations in shock tubes can be correlated with experimental observations in other venues (flow reactors, rapid compression machine experiments). However, it is unlikely that in real systems, the multiple sources of perturbations can be controlled to an extent such that homogenous kinetic predictions provide limiting, realistic design criteria. While the exact nature and relative importance of each perturbing source in the various experimental venues remain to be determined, simple engineering approximations can be valuable for the safe design of lean premixing systems for gas turbines, based upon fundamental experimental data. Carbon monoxide stored in high-pressure carbon-steel cylinders is commonly used in laboratory research. These sources are prone to contamination by metal carbonyls, and it has been shown through computations that these contaminants, especially iron pentacarbonyl, can considerably affect fuel-rich H2/CO laminar burning velocities. The potential effect of this contaminant in applied conditions is relatively

Syngas Chemical Kinetics and Reaction Mechanisms

63

unknown, but it is important that fuel-rich burning rate measurements used for validation of kinetic models should be carefully scrutinized for contaminant effects. Finally, we note that all of the hydrogen–oxygen kinetic models in the literature are deficient in yielding predictions of laminar burning rates at all equivalence ratios, pressures, and flame temperatures of likely importance to gas turbine applications. While the key reactions that most affect burning rates under these conditions are identified, a suitable solution, supported by fundamental kinetic results, remains to be achieved. Additional fundamental work on the reaction of HO2 with OH and H radicals, as well as characterizations of H2O2 and these radicals under reaction conditions, would be extremely beneficial in further constraining model validation at high pressures and temperatures.

Acknowledgments This work was supported by U.S Department of Energy grant DE-FG0286ER13503 (FLD) from the Chemical Sciences, Geosciences and Biosciences Division, Office of Basic Energy Sciences, Office of Science, and by grants DE-FG26-05NT42544 (FLD) and DE-FG26-06NT42716 (FLD, YJ) from the Pittsburgh National Energy Technology Laboratories. Support from Siemens Power Generation, Inc. (technical monitor, Dr. Scott Martin) (FLD) and by the Petroleum Research Fund from the American Chemistry Society under grant PRF 43460-AC5 (YJ) is also acknowledged. The authors thank Dr. Gregory Linteris for providing an electronic version of the iron pentacarbonyl kinetic model developed at NIST.

References Allen, T. L., Fink, W. H., and Volman, D. H. (1996). Theoretical studies of the mechanism of the gas phase reaction of the hydroperoxo radical with carbon monoxide to form hydroxyl radical and carbon dioxide. J. Phys. Chem. 100:5299. Ashmore, P. G., and Tyler, B. J. (1962). Reaction of hydrogen atoms with nitrogen dioxide. Trans. Farad. Soc. 58:1108. Baldwin, R. R., Jackson, D., Walker, R. W., and Webster, S. J. (1967). Interpretation of the slow reaction and second limit of hydrogen oxygen mixtures by computer methods. T. Faraday Soc. 63:1676. Bates, R. W., Golden, D. M., Hanson, R. K., and Bowman, C. T. (2001). Experimental study and modeling of the reaction H + O2 + M → HO2 (M = Ar, N2, H2O) at elevated pressures and temperatures between 1050 and 1250 K. Phys. Chem. Chem. Phys. 3:2337. Baulch, D. L., Bowman, C. T., Cobos, C. J., Cox, R. A., Just, T., Kerr, J. A., Pilling, M. J., Stocker, D., Troe, J., Tsang, W., and Walker, R. W. (2005). Evaluated kinetic data for combustion modeling: Supplement II. J. Phys. Chem. Ref. Data 34:757. Baulch, D. L., Drysdale, D. D., Duxbury, J., and Grant, S. (1973). Evaluated kinetic data for high temperature reactions, 3. London: Butterworths. Baulch, D. L., Griffiths, J. F., Pappin, A. J., and Sykes, A. F. (1988). Stationary-state and oscillatory combustion of hydrogen in a well-stirred flow reactor. Combust. Flame 73:163. Beerer, D. J., and McDonell, V. G. (2007). Experimental study of ignition delay for application to hydrogen and syngas fired lean premixed gas turbine engines. Paper E01 presented at Proceedings of the 5th U.S. Combustion Meeting, San Diego.

64

Synthesis Gas Combustion: Fundamentals and Applications

Beerer, D. J., and McDonell, V. G. (2008). Autoignition of hydrogen and air inside a continuous flow reactor with application to lean premixed combustion. J. Eng. Gas Turbines Power 130:051507. Blumenthal, R., Fieweger, K., Komp, K. H., and Adomeit, G. (1996). Gas dynamic features of self ignition of non diluted fuel/air mixtures at high pressure. Combust. Sci. Tech. 113/114:137. Blumenthal, R., Fieweger, K., Komp, K. H., Adomeit, G., and Gelfand B. E. (1995). Selfignition of H2-air mixtures at high pressure and low temperature. In Proceedings of the 20th International Symposium on Shock Waves, 935. Pasadena, CA, July 23–28. Bradley, D., Lawes, M., Liu, K., Verhelst, S., and Woolley, R. (2007). Laminar burning velocities of lean hydrogen–air mixtures at pressures up to 1.0 MPa. Combust. Flame 149:162. Braun M., Hofzumahaus, A., and Stuhl F. (1982). VUV flash-photolysis study of the reaction of HO with HO2 at 1 atm and 298 K. Ber. Bunsenges. Phys. Chem. 86:597. Brokaw, R. S. (1967). Ignition kinetics of the carbon monoxide-oxygen reaction. Proc. Combust. Inst. 11:1063. Burke, M. P., Chaos, M., Dryer, F. L., and Ju, Y. (2009a). The dependence of mass burning rates of H2/CO/CO2 flames on pressure and flame temperature. To appear. Burke, M. P., Chen, Z., Ju, Y., and Dryer, F. L. (2009b). Effect of cylindrical confinement on the determination of laminar flame speeds using outwardly propagating flames. Combust. Flame 156:771. Burke, M. P., Qin, X., Ju, Y., and Dryer, F. L. (2007). Measurements of hydrogen syngas flame speeds at elevated pressures. Paper A16 presented at Proceedings of the 5th U.S. Combustion Meeting, San Diego. Burrows, J. P., Cox, R. A., and Derwent, R. G. (1981). modulated photolysis of the ozonewater vapour system: Kinetics of the reaction of OH with HO2. J. Photochem. 16:147. Chaos, M., and Dryer, F. L. (2008). Syngas combustion kinetics and applications. Combust. Sci. Tech. 180:1051. Chen, Z., Burke, M. P., and Ju, Y. (2009a). Effects of Lewis number and ignition energy on the determination of laminar flame speed using propagating spherical flames. Proc. Combust. Inst. 32:1253. Chen, Z., Burke, M. P., and Ju, Y. (2009b). Effects of compression and stretch on the determination of laminar flame speed using propagating spherical flames. Combust. Theor. Model. 13:343. Chiesa, P., Consonni, S., Kreutz, T., and Williams, R. H. (2005). Co-production of hydrogen, electricity and CO2 from coal with commercially ready technology. Part A. Performance and emissions. Int. J. Hydrogen Energy 30:747. Cox, R. A., Burrows, J. P., and Wallington, T. J. (1981). Rate coefficient for the reaction OH + HO2 = H2O + O2 at 1 atmosphere pressure and 308 K. Chem. Phys. Lett. 84:217. Craig, R. R. (1966). A shock tube study of the ignition delay of hydrogen-air mixtures near the second explosion limit. Report AFAPL-TR-66-74, Air Force Aeropropulsion Lab, Wright-Patterson. Davidson, D. F., and Hanson, R. K. (2004). Interpreting shock tube ignition data. Int. J. Chem. Kinet. 36:510. Davis, S. G., Joshi, A. V., Wang, H., and Egolfopoulos, F. (2005). An optimized kinetic model of H2/CO combustion. Proc. Combust. Inst. 30:1283. Dean, A. M., Steiner, D. C., and Wang, E. E. (1978). A shock tube study of the H2/O2/CO/Ar and H2/N2O/CO/Ar systems: Measurement of the rate constant for H + N2O = N2 +OH. Combust. Flame 32:73. DeMore, W. B. (1979). Reaction of HO2 with O3 and the effect of water vapor on HO2 kinetics. J. Phys. Chem. 83:1113.

Syngas Chemical Kinetics and Reaction Mechanisms

65

DeMore, W. B. (1982). Rate constant and possible pressure dependence of the reaction hydroxyl + hydroperoxo. J. Phys. Chem. 86:121. Dixon-Lewis, G., and Williams, D. J. (1977). The oxidation of hydrogen and carbon monoxide. In C. H. Branford and C. F. H. Tipper (eds.), Comprehensive chemical kinetics, 1–248. Vol. 17. Amsterdam: Elsevier. Dryer, F. L., and Chaos, M. (2008). Ignition of syngas/air and hydrogen/air mixtures at low temperatures and high pressures: Experimental data interpretation and kinetic modeling implications. Combust. Flame 152:293. Egerton, A., and Warren, D. R. (1951). Kinetics of the hydrogen/oxygen reaction. I. The explosion region in boric acid-coated vessels. Proc. R. Soc. A Math. Phys. 204:465. Elsworth, J. E., Haskell, W. W., and Read, I. A. (1969). Non-uniform ignition processes in rapid-compression machines. Combust. Flame 13:437. Fernandes, R. X., Luther, K., Troe, J., and Ushakov, V. G. (2008). Experimental and modelling study of the recombination reaction H + O2 (+M) → HO2 (+M) between 300 and 900 K, 1.5 and 950 bar, and in the bath gases M = He, Ar, and N2. Phys. Chem. Chem. Phys. 10:4313. Fieweger, K., Blumenthal, R., and Adomeit, G. (1997). Self-ignition of S.I. Engine model fuels: A shock tube investigation at high pressure. Combust. Flame 109:599. Fotache, C. G., Tan, Y., Sung, C.-J., and Law, C. K. (2000). Ignition of CO/H2/N2 versus heated air in counterflow: Experimental and modeling results. Combust. Flame 120:417. Frenklach, M., Li Kwok Cheong, C. K., and Oran, E. S. (1984). LDV measurements of gas flow behind reflected shocks. Prog. Astronaut. Aeronaut. 95:722. Fujimoto, S. (1963). Chemical reaction in a shock wave I: The ignition delay of a hydrogen– oxygen mixture in a shock tube. Bull. Chem. Soc. Jpn. 36:1233. García, F. (2006). Syngas combustion experience at the Puertollano IGCC plant. Panel discussion GT2006-91352 at ASME Turbo Expo 2006, Barcelona, Spain, May 8–11. Gardiner, Jr., W. C., and Olson, D. B. (1980). Chemical kinetics of high temperature combustion. Ann. Rev. Phys. Chem. 31:377. Glarborg, P. (2007). Hidden interactions—Trace species governing combustion and emissions. Proc. Combust. Inst. 31:77. Goodings, J. M., and Hayhurst, A. N. (1988). Heat release and radical recombination in premixed fuel-lean flames of H2 + O2 + N2—Rate constants for H + OH + M → H2O + M and HO2 + OH → H2O + O2. J. Chem. Soc. Faraday Trans. 2:745. Hahn, J., Krasnoperov, L., Luther, K., and Troe, J. (2004). Pressure dependence of the reaction H + O2 (+Ar) → HO2 (+Ar) in the range 1–900 bar and 300–700 K. Phys. Chem. Chem. Phys. 6:1997. Haskell, W. W. (1970). Fuel ignition in a rapid compression machine: Sensitivity to flame ignition by particles. SAE Paper 700059. Hassan, M. I., Aung, K. T., and Faeth, G. M. (1996). Markstein numbers and unstretched laminar burning velocities of wet carbon monoxide flames. AIAA Paper 96-0912. Herzler, J., and Naumann, C. (2008). Shock tube study of the ignition of lean CO/H2 fuel blends at intermediate temperatures and high pressure. Combust. Sci. Tech. 180:2015. Hessler, J. P. (1998). Calculation of reactive cross sections and microcanonical rates from kinetic and thermochemical data. J. Phys. Chem. A 102:4517. Hippler, H., Neubauer, H., and Troe, J. (1995). Shock wave studies of the reactions HO + H2O2 → H2O + HO2 and HO + HO2 → H2O + O2 between 930 and 1680 K. J. Chem. Phys. 103:3510. Hippler, H., and Troe, J. (1992). Rate constants of the reaction HO + H2O2 → HO2 + H2O at T ≥ 1000 K. Chem. Phys. Lett. 192:333. Hippler, H., Troe, J., and Willner, J. (1990). Shock wave study of the reaction HO2 + HO2 → H2O2 + O2: Confirmation of a rate constant minimum near 700 K. J. Chem. Phys. 93:1755.

66

Synthesis Gas Combustion: Fundamentals and Applications

Hwang, S. M., Ryu, S.-O., De Witt, K. J., and Rabinowitz, M. J. (2005). High temperature rate coefficient measurements of H + O2 chain-branching and chain-terminating reaction. Chem. Phys. Lett. 408:107. Izod, T. P. J., Kistiakowsky, G. B., and Matsuda, S. (1972). Mechanism of oxidation of carbon monoxide in the presence of chromium carbonyl. J. Chem. Phys. 56:1083. Kahandawala, M. S. P., Corera, A. P., Williams, S., Carter, C. D., and Sidhu, S. S. (2006). Investigation of kinetics of Iso-octane ignition under scramjet conditions. Int. J. Chem. Kinet. 38:194. Kalitan, D. M., Petersen, E. L., Mertens, J. D., and Crofton, M. W. (2006). Ignition of lean CO/H2/air mixtures at elevated pressures. ASME Paper GT2006-90488. Kappel, Ch., Luther, K., and Troe, J. (2002). Shock wave study of the unimolecular dissociation of H2O2 in its falloff range and of its secondary reactions. Phys. Chem. Chem. Phys. 4:4392. Kazakov, A., Chaos, M., Zhao, Z., and Dryer, F. L. (2006). Computational singular perturbation analysis of two-stage ignition of large hydrocarbons. J. Phys. Chem. A 110:7003. Kee, R. J., Rupley, F. M., and Miller, J. A. 1989. CHEMKIN II: A Fortran chemical kinetics package for the analysis of gas phase chemical kinetics. Technical Report SAND89-8009, Sandia National Laboratories. Keyser, L. F. (1988). Kinetics of the reaction OH + HO2 → H2O + O2 from 254 K to 382 K. J. Phys. Chem. 92:1193. Kim, T. J., Yetter, R. A., and Dryer, F. L. (1994). New results on moist CO oxidation: High pressure, high temperature experiments and comprehensive kinetic modeling. Proc. Combust. Inst. 25:759. Konnov, A. A. (2008). Remaining uncertainties in the kinetic system of hydrogen combustion. Combust. Flame 152:507. Kurylo, M. J., Klais, O., and Laufer, A. H. (1981). Mechanistic investigation of the HO + HO2 reaction. J. Phys. Chem. 85:3674. Larson, E. D., Fiorese, G., Liu, G., Williams, R. H., Kreutz, T. J., and Consonni, S. (2009). Co-production of synfuels and electricity from coal + biomass with zero net carbon emissions: An Illinois case study. Energy Procedia 1:4371. Lask, G., and Wagner, H. Gg. (1960). Influence of additives on the velocity of laminar flames. Proc. Combust. Inst. 8:432. Le Cong, T., and Dagaut, P. (2008). Experimental and detailed kinetic modeling of the oxidation of methane and methane/syngas mixtures and effect of carbon dioxide addition. Combust. Sci. Tech. 180:2046. Lee, D., and Hochgreb, S. (1998). Hydrogen autoignition at pressures above the second explosion limit (0.6–4.0 MPa). Int. J. Chem. Kinet. 30:385. Lewis, B. and von Elbe, G. (1987). Combustion, flames and explosion of gases. 3rd ed. Orlando, FL: Academic Press. Li, H., Owens, Z. C., Davidson, D. F., and Hanson, R. K. (2008). A simple reactive gas­ dynamic model for the computation of gas temperature and species concentrations behind reflected shock waves. Int. J. Chem. Kinet. 40:89. Li, J., Zhao, Z., Kazakov, A., Chaos, M., Dryer, F. L., and Scire, Jr., J. J. (2007). A comprehensive kinetic mechanism for CO, CH2O, and CH3OH combustion. Int. J. Chem. Kinet. 39:109. Li, J., Zhao, Z., Kazakov, A., and Dryer, F. L. (2004). An updated comprehensive kinetic model of hydrogen combustion. Int. J. Chem. Kinet. 36:566. Lieuwen, T., McDonell, V. G., Petersen, E. L., and Santavicca, D. (2008). Fuel flexibility influences on premixed combustor blowout, flashback, autoignition, and stability. J. Eng. Gas Turbines Power 130:011506. Lii, R.-R., Corse, R. A., Sauer, M. C., and Gordon, S. (1980). Temperature dependence of the gas-phase self-reaction of hydroperoxo radicals in the presence of ammonia. J. Phys. Chem. 84:813.

Syngas Chemical Kinetics and Reaction Mechanisms

67

Linteris, G. T., and Babushok, V. I. (2009). Promotion or inhibition of hydrogen-air ignition by iron-containing compounds. Proc. Combust. Inst. 32:2535. Linteris, G. T., Rumminger, M. D., and Babushok, V. I. (2008). Catalytic inhibition of laminar flames by transition metal compounds. Prog. Energ. Combust. Sci. 34:288. Lutz, A. E., Kee, R. J., and Miller, J. A. (1987). SENKIN: A Fortran program for predicting. Homogeneous gas phase chemical kinetics with sensitivity analysis. Technical Report SAND87-8248, Sandia National Laboratories. Martynenko, V. V., Penyaz’kov, O. G., Ragotner, K. A., and Shabunya, S. I. (2004). Hightemperature ignition of hydrogen and air at high pressures downstream of the reflected shock wave. J. Eng. Phys. Thermophys. 77:785. Matsuda, S. (1972a). Gas phase homogeneous catalysis in shock waves. II. The oxidation of carbon monoxide by oxygen in the presence of iron pentacarbonyl. J. Chem. Phys. 57:807. Matsuda, S. (1972b). Gas phase homogeneous catalysis in shock waves. III. The oxidation of carbon monoxide by oxygen in the presence of nickel carbonyl. J. Phys. Chem. 76:2833. McDowell, R. S. (1971). Metal carbonyl vapors: Rapid quantitative analysis by infrared spectrophotometry. Am. Ind. Hyg. Assoc. J. 32:621. McLean, I. C., Smith, D. B., and Taylor, S. C. (1994). The use of carbon monoxide/hydrogen burning velocities to examine the rate of the CO + OH reaction. Proc. Combust. Inst. 25:749. Mertens, J. D., Kalitan, D. M., Barrett, A. B., and Petersen, E. L. (2009). Determination of the rate of H + O2 + M → HO2 + M (M = N2, Ar, H2O) from ignition of syngas at practical conditions. Proc. Combust. Inst. 32:295. Mertens, J. D., Mussmann, S. S., Kalitan, D. M., and Petersen, E. L. (2008). A chemical kinetics model for the fast ignition of syngas at lower temperatures and higher pressures. Paper 08S-3 presented at spring meeting of the Western States Section of the Combustion Institute, Los Angeles, March. Meyer, J. W., and Oppenheim, A. K. (1970). On the shock induced ignition of explosive gases. Proc. Combust. Symp. 13:1153. Michael, J. V., and Sutherland, J. W. (1986). The thermodynamic state of the hot gas behind reflected shock waves: Implications to chemical kinetics. Int. J. Chem. Kinet. 18:409. Michael, J. V., Su, M. C., Sutherland, J. W., Carroll, J. J., and Wagner, A. F. (2002). Rate constants for H + O2 + M → HO2 +M in seven bath gases. J. Phys. Chem. A 106:5297. Michael, J. V., Sutherland, J. W., Harding, L. B., and Wagner, A. F. (2000). Initiation in H2/O2: Rate constants for H2 + O2 → H + HO2 at high temperature. Proc. Combust. Inst. 28:1471. Mittal, G., Chaos, M., Sung, C.-J., and Dryer, F. L. (2008). Dimethyl ether autoignition in a rapid compression machine: Experiments and chemical kinetic modeling. Fuel. Process. Technol. 89:1244. Mittal, G., Sung, C.-J., Fairweather, M., Tomlin, A. S., Griffiths, J. F., and Hughes, K. J. (2007). Significance of the HO2+CO reaction during the combustion of CO+H2 mixtures at high pressures. Proc. Combust. Inst. 31:419. Mittal, G., Sung, C.-J., and Yetter, R. A. (2006). Autoignition of H2/CO at elevated pressures in a rapid compression machine. Int. J. Chem. Kinet. 38:516. Mueller, M. A., Kim, T. J., Yetter, R. A. and Dryer, F. L. (1999a). Flow reactor studies and kinetic modeling of the H2/O2 reaction. Int. J. Chem. Kinet. 31:113. Mueller, M. A., Yetter, R. A., and Dryer, F. L. (1996a). Explosion limit behavior of dilute hydrogen–oxygen mixtures from 0.3 to 3.0 atm and 850 to 950 K. Paper presented at fall meeting of the Eastern States Section of the Combustion Institute, Hilton Head Island, SC, December. Mueller, M. A., Yetter, R. A., and Dryer, F. L. (1999b). Flow reactor studies and kinetic modeling of the H2/O2/NOX and CO/H2O/O2/NOX reactions. Int. J. Chem. Kinet. 31:705. Mueller, M. A., Yetter, R. A., and Dryer, F. L. (1998). Measurement of the rate constant for H + O2 + M → HO2 + M (M = N2, Ar) using kinetic modeling of the high-pressure H2/O2/NOX reaction. Proc. Combust. Inst. 27:177.

68

Synthesis Gas Combustion: Fundamentals and Applications

Mueller, M. A., Yetter, R. A., and Dryer, F. L. (2000). Kinetic modeling of the CO/H2O/O2/ NO/SO2 system: Implications for high-pressure fall-off in the SO2 + O (+M) = SO3 (+M) reaction. Int. J. Chem. Kinet. 32:317. Myrstad, T., and Fredriksen, G. R. (1998). Removal of Fe(CO)5 from CO gas as detected by FTIR spectroscopy. Chem. Eng. Technol. 21:297. Natarajan, J., Kochar, Y., Lieuwen, T., and Seitzman, J. (2009). Pressure and preheat dependence of laminar flame speeds of H2/CO/CO2/O2/He mixtures. Proc. Combust. Inst. 32:1261. Natarajan, J., Lieuwen, T., and Seitzman, J. (2007). Laminar flame speeds of H2/CO mixtures: Effect of CO2 dilution, preheat temperature, and pressure. Combust. Flame 151:104. O’Connaire, M., Curran, H. J., Simmie, J. M., Pitz, W. J., and Westbrook, C. K. (2004). A comprehensive modeling study of hydrogen oxidation. Int. J. Chem. Kinet. 36:603. Oppenheim, A. K. (1985). Dynamic features of combustion. Phil. Trans. R. Soc. Lond. A 315:471. Pang, G. A., Davidson, D. F., and Hanson, R. K. (2009). Experimental study and modeling of shock tube ignition delay times for hydrogen–oxygen–argon mixtures at low temperatures. Proc. Combust. Inst. 32:181. Peeters, J., and Mahnen, G. (1973). Reaction mechanisms and rate constants of elementary steps in methane-oxygen flames. Proc. Combust. Inst. 14:133. Peschke, W. T., and Spadaccini, L. J. (1985). Determination of autoignition and flame speed characteristics of coal gases having medium heating values. Report EPRI AP-4291, Electric Power Research Institute. Petersen, E. L., and Hanson, R. K. (2001). Nonideal effects behind reflected shock waves in a high-pressure shcok tube. Shock Waves 10:405. Petersen, E. L., Kalitan, D. M., Barrett, A. B., Reehal, S. C., Mertens, J. D., Beerer, D. J., Hack, R. L., and McDonell, V. G. (2007a). New syngas/air ignition data at lower temperature and elevated pressure and comparison to current kinetics models. Combust. Flame 149:244. Petersen, E. L., Lamnaouer, M., de Vries, J., Curran, H. J., Simmie, J., Fikri, M., Schulz, C., and Bourque, G. (2007b). Discrepancies between shock-tube and rapid compression machine ignition at low temperatures and high pressures. Paper P911 presented at Proceedings of the 26th International Symposium on Shock Waves, Göttingen, Germany, July. Rao, K. M., Spoto, G., Guglielminotti, E., and Zecchina, A. (1989). IR investigation of Fe(CO)5 adducts at the surface of silicalite, ZSM-5, zeolite Y, and γ-Al2O3. Inorg. Chem. 28:243. Rasmussen C. L., Hansen, J., Marshall, P., and Glarborg, P. (2008). Experimental measurements and kinetic modeling of CO/H2/O2/NO, conversion at high pressure. Int. J. Chem. Kinet. 40:454. Reehal, S. C., Kalitan, D. M., Hair, T., Barrett, A. B., and Petersen, E. L. (2007). Ignition delay time measurements of synthesis gas mixtures at engine pressures. Paper C24 presented at Proceedings of the 5th U.S. Combustion Meeting, San Diego. Reinelt, D., and Linteris, G. T. (1996). Experimental study of the inhibition of premixed and diffusion flames by iron pentacarbonyl. Proc. Combust. Inst. 26:1421. Rezaiyan, J., and Chereminisoff, N. P. (2005). Gasification technologies: A primer for engineers and scientists. Boca Raton, FL: CRC Press. Rozenshtein, V. B., Gershenzon, Yu. M., Il’in, S. D., and Kishkovjtch, O. P. (1984). Reactions of HO2 with NO, OH and HO2 studied by EPR/LMR spectroscopy. Chem. Phys. Lett. 112:473. Rumminger, M. D., and Linteris, G. T. (2000). Inhibition of premixed carbon monoxidehydrogen–oxygen–nitrogen flames by iron pentacarbonyl. Combust. Flame 120:451. Rumminger, M. D., and Linteris, G. T. (2002). The role of particles in the inhibition of counterflow diffusion flames by iron pentacarbonyl. Combust. Flame 128:145. Rumminger, M. D., Reinelt, D., Babushok, V., and Linteris, G. T. (1999). Numerical study of the inhibition of premixed and diffusion flames by iron pentacarbonyl. Combust. Flame 116:207.

Syngas Chemical Kinetics and Reaction Mechanisms

69

Sellevåg, S. R., Georgievskii, Y., and Miller, J. A. (2008). The temperature and pressure dependence of the reactions H + O2 (+M) → HO2 (+M) and H + OH (+M) → H2O (+M). J. Phys. Chem. A 112:5085. Shen, H.-P. S., Vanderover, J., and Oehlschlaeger, M. A. (2009). A shock tube study of the auto-ignition of toluene/air mixtures at high pressures. Proc. Combust. Inst. 32:165. Sivaramakrishnan, R., Comandini, A., Tranter, R. S., Brezinsky, K., Davis, S. G., and Wang, H. (2007). Combustion of CO/H2 mixtures at elevated pressures. Proc. Combust. Inst. 31:429. Slack, M. W. (1977). Rate coefficient for H + O2 + M = HO2 + M evaluated from shock tube measurements of induction times. Combust. Flame 28:241. Snyder, A. D., Robertson, J., Zanders, D. L., and Skinner, G. B. (1965). Shock tube studies of fuel–air ignition characteristics. Report AFAPL-TR-65-93, Air Force Aeropropulsion Lab, Wright-Patterson. Sridharan, U. C., Qiu, L. X., and Kaufman, F. (1984). Rate constant of the OH + HO2 reaction from 252 to 420 K. J. Phys. Chem. 88:1281. Srinivasan, N. K., and Michael, J. V. (2006). The thermal decomposition of water. Int. J. Chem. Kinet. 38:211. Srinivasan, N. K., Su, M.-C, Sutherland, J. W., Michael, J. V., and Ruscic, B. (2006). Reflected shock tube studies of high-temperature rate constants for OH + NO2 → HO2 + NO and OH + HO2 → H2O + O2. J. Phys. Chem. A 110:6602. Strehlow, R. A., and Cohen, A. (1962). Initiation of detonation. Phys. Fluids 5:97. Ströhle, J., and Myhrvold, T. (2007). An evaluation of detailed reaction mechanisms for hydrogen combustion under gas turbine conditions. Int. J. Hydrogen Energy 32:125. Sun, H. Y., Yang, S. I., Jomaas, G., and Law, C. K. (2007). High-pressure laminar flame speeds and kinetic modeling of carbon monoxide/hydrogen combustion. Proc. Combust. Inst. 31:439. Sung, C.-J., and Law, C. K. (2008). Fundamental combustion properties of H2/CO mixtures: Ignition and flame propagation at elevated pressures. Combust. Sci. Tech. 180:1097. Tepe, R. K., Vassallo, D., Jacksier, T., and Barnes, R. M. (1999). Iron pentacarbonyl determination in carbon monoxide. Spectrochim. Acta B 54:1861. Terao, K. (1977). Explosion limits of hydrogen–oxygen mixtures as a stochastic phenomenon. Jpn. J. Appl. Phys. 16:29. Troe, J. (2000). Detailed modeling of the temperature and pressure dependence of the reaction H + O2 (+M) → HO2 (+M). Proc. Combust. Inst. 28:1463. Troe J. and Ushakov, V. G. (2001). Theoretical studies of the HO + O ↔ HO2 ↔ H + O2 reaction. II. Classical trajectory calculations on an ab initio potential for temperatures between 300 and 5000 K. J. Chem. Phys. 115:3621. Tse, S. D., Zhu, D. L., and Law, C. K. (2000). Morphology and burning fluxes of expanding spherical flames in H2/O2/inert mixtures up to 60 atmospheres. Proc. Combust. Inst. 28:1793. Vanpee, M., and Shirodkar, P. P. (1978). Study of flame inhibition by metal compounds. Proc. Combust. Inst. 17:787. Voevodsky, V. V., and Soloukhin, R. I. (1965). On the mechanism and explosion limits of hydrogen–oxygen chain self-ignition in shock waves. Proc. Combust. Inst. 10:279. von Elbe, G., and Lewis, B. (1942). Mechanism of the thermal reaction between hydrogen and oxygen. J. Chem. Phys. 10:366. Walton, S. M., He, X., Zigler, B. T., and Wooldridge, M. S. (2007a). An experimental investigation of the ignition properties of hydrogen and carbon monoxide mixtures for syngas turbine applications. Proc. Combust. Inst. 31:3147. Walton, S. M., He, X., Zigler, B. T., and Wooldridge, M. S. (2007b). An experimental investigation of iso-octane ignition phenomena. Combust. Flame. 150:246. Wang, B. L., Olivier, H., and Grönig, H. (2003). Ignition of shock-heated H2-air-steam mixtures. Combust. Flame 133:93.

70

Synthesis Gas Combustion: Fundamentals and Applications

Wang, H., You, X., Joshi, A. V., Davis, S. G., Laskin, A., Egolfopoulos, F., and Law, C. K. (2007). USC Mech Version II: High-temperature combustion reaction model of H2/CO/ C1-C4 compounds, http://ignis.usc.edu/USC_mech_II.htm (accessed May 2007). Westbrook, C. K., and Dryer, F. L. (1984). Chemical kinetics modeling of hydrocarbon combustion. Prog. Energ. Combust. Sci. 10:1. Williams, T. C., and Shaddix, C. R. (2007). Contamination of carbon monoxide with metal carbonyls: Implications for combustion research. Combust. Sci. Tech. 175:1225. Yetter, R. A., Dryer, F. L., and Rabitz, H. (1991a). A comprehensive reaction mechanism for carbon monoxide/hydrogen/oxygen kinetics. Combust. Sci. Tech. 79:97. Yetter, R. A., Dryer, F. L., and Rabitz, H. (1991b). Flow reactor studies of carbon monoxide/ hydrogen/oxygen kinetics. Combust. Sci. Tech. 79:129. You, X., Wang, H., Goos, E., Sung, C. J., and Klippenstein, S. J. (2007). Reaction kinetics of CO + HO2 → products: Ab initio transition state theory study with master equation modeling. J. Phys. Chem. A 111:4031. Zhao, Z., Li, J., Kazakov, A., and Dryer, F. L. (2005). Temperature-dependent feature sensitivity analysis for combustion modeling. Int. J. Chem. Kinet. 37:282. Zheng, X. L., and Law, C. K. (2004). Ignition of premixed hydrogen/air by heated counterflow under reduced and elevated pressures. Combust. Flame 136:168.

Flame Properties 3 Laminar of H /CO Mixtures 2

Jayaprakash Natarajan and Jerry M. Seitzman Contents 3.1 Introduction..................................................................................................... 71 3.2 Premixed Flame Properties............................................................................. 72 3.2.1 Adiabatic Flame Temperature............................................................. 72 3.2.2 Flame Structure and Flame Thickness................................................ 74 3.2.3 Flame Propagation...............................................................................80 3.2.3.1 H2/CO Ratio.......................................................................... 81 3.2.3.2 Pressure................................................................................. 81 3.2.3.3 Preheat Temperature............................................................. 86 3.2.3.4 Diluents................................................................................. 88 3.2.3.5 Flame Stretch........................................................................90 3.2.4 Flame Extinction.................................................................................92 3.3 Nonpremixed Flame Properties....................................................................... 95 3.4 Conclusions......................................................................................................96 References.................................................................................................................96

3.1 Introduction Laminar flames represent an excellent starting point for understanding syngas flame properties. Moreover, knowledge of laminar flame properties is essential to understanding the behavior of turbulent flames. Finally, various computational tools used to design an efficient fuel-flexible combustor require an understanding of the fundamental combustion properties of these mixtures, such as flame temperature, laminar flame speed, strain sensitivity, and extinction strain rates. Thus, this chapter provides an overview of laminar flame properties of syngas fuel mixtures, primarily with air as the oxidizer. We begin with premixed laminar flames, where fuel and oxidizer are well mixed before reaching the flame. In particular, the first section emphasizes the structure and propagation characteristics (that is, flame speed) for syngas flames. Because the operating environment for syngas fuels can range from atmospheric burners for process heating to gas turbine engines, this section also describes the dependence of the premixed flame properties on mixture composition, pressure, and preheat temperature. The effects of strain on flame propagation are also discussed briefly, including the influence of H2 on the extinction

71

72

Synthesis Gas Combustion: Fundamentals and Applications

characteristics of CO and CH4 flames. Finally, we describe some important attributes of nonpremixed laminar syngas flames.

3.2 Premixed Flame Properties 3.2.1 Adiabatic Flame Temperature The first parameter of interest is the adiabatic flame temperature (Tad ). It is defined as the equilibrium temperature of the products when the reactants are burned at constant pressure without any heat transfer to or from the surroundings. In flames, reactants are converted to products essentially at constant pressure; thus, the maximum temperature of the flame is typically close to the adiabatic flame temperature in the absence of nonunity Lewis number, differential diffusion, and strain effects (Law and Sung, 2000). Flame temperature is an important parameter for a number of reasons; for example, NOX production is highly sensitive to temperature through the thermal (Zeldovich) mechanism, which tends to dominate NOX production beyond about 1800 K. Moreover, flame temperature also has a significant influence on flame propagation and extinction. Figure 3.1 compares the adiabatic flame temperatures for different pure fuel gases (CH4, H2, and CO) and a nominal syngas fuel mixture (35% H2, 35% CO, and 30% CO2). Of the three pure fuels, CO has the highest flame temperature for a given equivalence ratio, while CH4 has the lowest. For a lean mixture with equivalence ratio Φ = 0.6, the CO flame temperature is around 300 K higher than that of CH4 and the differences are even greater for rich mixtures. Methane-air mixtures have lower flame temperatures than CO and H2, because CH4 requires four times more oxidizer (on a molar basis) to achieve a stoichiometric mixture. For an undiluted oxidizer (e.g., pure O2), the CH4 flame temperatures are much closer to those of the other fuels.

Adiabatic Flame Temperature Tad (K)

2500

CH4 H2 CO Syngas

1.25 1.06

2300

1.04

2100

1.04

1900 1700 1500

0.5

0.75

1

1.25

1.5

1.75

2

Equivalence Ratio Ф

Figure 3.1  Adiabatic flame temperatures for CH4, H2, CO, and typical syngas (35% H2, 35% CO, and 30% CO2) fuels and air at atmospheric pressure and 300 K initial temperature.

73

Laminar Flame Properties of H2 /CO Mixtures

Though both CO and H2 require the same amount of oxidizer for a given equivalence ratio, the CO flame temperature is slightly higher than for H2 owing to the higher molar heating value of CO. At Φ = 0.6, the CO flame temperature is around 130 K greater than Tad for hydrogen. This difference is greatly reduced as the mixture nears stoichiometric conditions, since the higher temperatures there lead to reduced CO2 levels, and therefore less heat release associated with the additional fuel. The chosen syngas mixture’s flame temperature is lower than that for either pure H2 or CO due to the significant amount of diluent in the syngas fuel. For the 30% CO2 dilution considered here, the flame temperature is similar to that for the CH4 flame (within ~100 K) over the range of practical equivalence ratios. Thus, we see that undiluted syngas mixtures would have higher temperatures than conventional methane (or natural gas) fuel, while syngas compositions with typical levels of dilution will have flame temperatures closer to those encountered in natural gas combustion. For all the fuels shown in Figure  3.1, the peak Tad occurs at slightly fuel-rich conditions. This can be explained by considering the balance between the energy required to raise the product temperature (characterized by the specific heat of the combustion products) and the amount of heat release, which is reduced by the extent of product dissociation (Law, 2006). For these fuels, the trade-off leads to the peak Tad occurring at slightly less rich conditions (Φ ≅ 1.04 for CH4, 1.06 for H2, and 1.25 for CO flames) than the peak heat release point. For the diluted syngas fuel, the peak adiabatic flame temperature occurs at a lower equivalence ratio than for the pure CO or H2 flames. This is mainly due to the lower stoichiometric Tad, which reduces the extent of product dissociation. Apart from fuel and oxidizer composition, other factors that influence adiabatic flame temperature include pressure and (initial) reactant temperature (Tu). Figure 3.2 highlights the effect of pressure on adiabatic flame temperature for the nominal

Adiabatic Flame Temperature Tad (K)

2500 1.1 2300

1.02

p = 1 atm; Tu = 300 K p = 30 atm; Tu = 300 K p = 1 atm; Tu = 700 K

1.04

2100

1900

1700

1500 0.5

0.75

1 1.25 1.5 Equivalence Ratio Ф

1.75

2

Figure  3.2  Influence of pressure and preheat temperature on adiabatic flame temperatures for typical syngas (35% H2, 35% CO, and 30% CO2) fuel burning with air.

74

Synthesis Gas Combustion: Fundamentals and Applications

syngas fuel mixture. As the pressure increases from 1 to 30 atm (for a fixed Tu), there is no significant change in adiabatic flame temperature, except for a small increase (~50 K) near stoichiometric conditions. The temperature rise is associated with the high-pressure suppression of product dissociation, which is an issue only in the region of high Tad (e.g., near Φ ≅ 1). The suppression of product dissociation also tends to shift the peak flame temperature toward Φ = 1. The influence of reactant temperature on adiabatic flame temperature is also evident in Figure 3.2. The adiabatic flame temperature increases with reactant temperature, and the rise in Tad decreases as the reactant mixture approaches stoichiometric. For the case considered here, the increase in reactant temperature by 400 K (from room temperature to 700 K) causes Tad to increase by ~300 K at Φ ≅ 0.5 and 2.0, but only by ~200 K for Φ near 1. Higher levels of product dissociation at the high temperatures associated with near stoichiometric conditions lower the energy available to raise the product temperature; thus, there is a lower temperature increase there. More product dissociation with higher initial reactant temperatures also shifts the peak Tad to richer mixtures. As shown in Figure 3.2 for the nominal syngas fuel composition, the peak in Tad moves from Φ ≅ 1.04 for room temperature reactants to Φ ≅ 1.1 for the preheated reactants.

3.2.2 Flame Structure and Flame Thickness In order to understand laminar flame propagation in syngas mixtures, it is important to have a picture of the premixed syngas flame structure. The detailed structure of a one-dimensional, premixed laminar flame can be faithfully simulated using the steady-state mass, species, and energy conservation equations, state equations, and a comprehensive reaction mechanism for the fuel and oxidizer of interest. For the results presented here, we employ the freely propagating flame code, Chemkin PREMIX (Kee et al., 2006), which includes a detailed package to evaluate the transport (diffusive) properties for complex gas mixtures. There are a number of detailed chemical reaction mechanisms that include the fundamental H2/CO kinetics (see Chapter 2); for the numerical results presented here, a detailed C1 mechanism (Li et al., 2007) is used. The chemical structure of a syngas flame significantly changes with the amount of H2 in the fuel mixture, owing to the widely different kinetic and transport properties of H2 and CO. For example, we consider two lean syngas mixtures: one with very low H2 content (5% by mole) and the other with a medium level of H2 (50%). The simulations are performed for undiluted mixtures at room temperature and pressure. Figure 3.3 shows the profiles of temperature and heat release rate associated with key exothermic reactions along the flame coordinate for the low H2 case. The main heat release reaction here is CO + OH ↔ CO2 + H. In the absence of any hydrogen in the reactants (which can lead to OH production), CO oxidation is rather slow. The heat release rate profile of this CO + OH reaction is very broad; it starts at a temperature of around 800 K, peaks around 1300 K, and continues until the final flame temperature is reached. Notably, this reaction is also the main source for H atoms produced in the flame zone (McLean et al., 1994).

75

Laminar Flame Properties of H2 /CO Mixtures 2.5

1800

H + O2(+M) = HO2(+M) HO2 + O = O2 + OH

2.0

1500

H2 + OH = H2O + H

Temperature

CO + O(+M) = CO2(+M)

1.5

Total heat release rate/2

CO + OH = CO2 + H

1200

1.0

900

0.5

600

0.0

0

0.05

0.1

0.15

0.2

Temperature (K)

Heat Release Rate (10–9erg/cm3s)

HO2 + H = OH + OH

300

Flame Coordinate (cm)

Figure 3.3  Chemical structure of a lean (Φ = 0.6) H2/CO fuel mixture with 5% mole fraction of H2 at p = 1 atm and Tu = 300 K.

The H atoms produced in the flame zone diffuse upstream into the incoming reactant mixture. There, the H atoms react with O2 to form the hydroperoxyl radical, HO2, through the three-body radical termination reaction H + O2(+M) ↔ HO2(+M). As shown in Figure 3.3, this reaction is also one of the main heat release reactions in the leading edge of the flame. The HO2, which is relatively stable at low temperatures, reacts with H and O atoms to produce a pool of OH radicals as the reactants move into the flame zone. These OH radicals attack CO and convert it to CO2, thus completing the cycle. The other significant route for CO oxidation is through the reaction CO + O(+M) ↔ CO2(+M). The importance of this reaction is drastically reduced as the amount of H2 in the fuel mixture increases beyond 15 to 20%. The main H2 oxidation reaction (H2 + OH ↔ H2O + H) has little direct contribution to the overall heat release; yet it is still an important reaction as it competes for OH radicals with the main CO oxidation reaction in low H2 content syngas flames. Figure 3.4 presents the corresponding profiles of temperature and heat release rate for a lean syngas fuel mixture with a medium (50%) level of H2. Since the equivalence ratio is the same as that used for the low H2 simulation, the flame temperature is only slightly lower. The first notable difference between the low and medium H2 cases is that the overall heat release rate increases by a factor of 4, owing to the higher reactivity of H2 compared to CO. This higher reactivity, combined with the higher diffusivity of H2 compared to CO, results in an increase in flame propagation speed as H2 levels rise in a syngas fuel.

76

Synthesis Gas Combustion: Fundamentals and Applications 9.0

Total heat release rate/2

1800

Temperature

1500 H + O2(+M) = HO2(+M)

6.0

HO2 + H = OH + OH H2 + OH = H2O + H

4.5

1200

CO + OH = CO2 + H H + OH (+M) = H2O(+M)

900

Temperature (K)

Heat Release Rate (10–9erg/cm3s)

7.5

3.0 600

1.5

0.0 0.05

0.075

0.1 0.125 0.15 Flame Coordinate (cm)

0.175

300 0.2

Figure 3.4  Chemical structure of a lean (Φ = 0.6) H2/CO fuel mixture with 50% mole fraction of H2 at p = 1 atm and Tu = 300 K.

Another important difference can be observed by comparing the temperature profiles with the overall heat release rate profiles (see Figures 3.3 and 3.4). For the medium H2 case, the overall heat release starts earlier, where the temperature is closer to the initial reactant temperature. In other words, the thickness of the nominal preheat zone decreases as H2 levels rise in syngas fuel mixtures. Looking at the contribution from individual exothermic reactions, this early rise in heat release is mainly due to the contribution of the H atom termination reaction H + O2(+M) ↔ HO2(+M). This reaction is also the dominant heat release reaction throughout most of the reaction zone, with a peak in its heat release rate when the temperature has risen to ~900 K (it is surpassed by the CO2 and H2O formation reactions (CO + OH and H + OH + M) only after most of the heat release has occurred). To better understand the balance between different reactions, consider the production and destruction of H atoms. There is a significant amount of H atom production in the later stage of the flame, through the main H2 and CO oxidation reactions H2 + OH and CO + OH, respectively. These two reactions dominate H atom production, as opposed to the low H2 case where CO + OH ↔ CO2 + H was the only significant source of H atoms. Similar to the low H2 case, the produced H atoms diffuse back to the incoming (cold) reactants. There they react with O2 to form HO2, causing a sharp rise in HO2 concentration in the early part of the flame. Recall that this reaction is highly exothermic and consumes H atoms. The HO2 reacts further, with H atoms, and produces two OH radicals (i.e., HO2 + H ↔ OH + OH), again consuming more H atoms in the early part of the flame. Then the produced pool of OH radicals attacks H2 and CO to form more H atoms in the later portion of the flame, and thus complete

77

Laminar Flame Properties of H2 /CO Mixtures Φ = 0.6 Φ = 0.8 Φ = 1.0 Φ = 2.0

Flame Thickness, δf (µm)

1000

750

500

250

0

20

40

% H2

60

80

100

Figure 3.5  Laminar flame thickness from simulations for a range of H2/CO fuel mixtures at p = 1 atm and Tu = 300 K.

the cycle. This medium H2 content syngas flame structure is very similar to that for pure H2 flames (Law, 2006), where the reactions occurs through the entire flame zone, which can be split into H atom consumption and H production layers, and there is no preheat zone as such. This is in contrast to conventional CH4-air flames, where the majority of the heat release reactions occur in the later regions of the flame, only after the temperature has risen significantly. Hence, it is clear that the relative thickness of the reaction zone to the diffusional (preheat) zone thickness increases as the amount of H2 increases in the fuel mixture. Also, the amount of H2 in the fuel mixture has significant influence on the overall flame thickness. We can define the overall flame thickness (δf ) based on the temperature gradient, that is, δf = ∆T/∇Tmax, the ratio of the temperature rise in the flame to the maximum temperature gradient (Law and Sung, 2000). Figure 3.5 shows that the flame becomes much thinner as the level of H2 in the syngas fuel increases. Generally, flame thickness can be considered a function of the diffusive and chemical rates of the mixture, that is,

δ f ∝ ρα RR ,

where ρ is the density of the mixture, α is the thermal diffusivity, and RR is the overall reaction rate (Glassman, 1996). The increase in the overall reaction rate with the amount of H2 is greater than the increase in diffusivity, resulting in an overall decrease in flame thickness. Similarly, an increase in equivalence ratio can also cause the flame thickness to decrease. Another parameter that has a key influence on flame thickness is pressure. Figure  3.6 shows the variation of flame thickness with pressure for three (lean,

78

Synthesis Gas Combustion: Fundamentals and Applications

Φ = 0.6 Φ = 1.0 Φ = 2.0

1.5

1.2 0.9

100 0.6

pδf (atm mm)

Flame Thickness, δf (µm)

1000

0.3 10

0

10

20

30

0 40

Pressure (atm)

Figure  3.6  Pressure dependence of laminar flame thickness (solid lines) and pressureflame thickness product (dashed lines) from simulations for a 50:50 H2/CO fuel mixture at Tu = 300 K.

stoichiometric, and rich) 50% H2 syngas fuel mixtures. In all cases, the flame thickness decreases monotonically with pressure. The influence of pressure on flame thickness is drastic at lower pressures but reduced at elevated pressures, with leaner flames exhibiting a greater decline in pressure dependence. As noted above, the flame thickness can be modeled as

δ f ∝ ρα RR .

Since ρα is nearly insensitive to pressure and the reaction rate normally increases with pressure, the flame thickness generally decreases with pressure. Figure 3.6 also shows the variation of the pressure-flame thickness product (pδf ) with pressure. The intent is to compare the relative decrease in flame thickness with pressure at different equivalence ratios. As seen in the figure, pδf is roughly constant for stoichiometric mixtures, with a weak minimum near 5 atm. The rich (Φ = 2) mixture produces a similar result. Thus, we find that the drop in flame thickness is nearly proportional to pressure (i.e., δf ∝ 1/p) for the stoichiometric and rich flames. On the other hand, pδf increases with pressure for the lean mixture, indicating that the flame thickness is a weaker function of pressure for lean mixtures. Above 10 atm, the increase in pδf for the Φ = 0.6 case is nearly linear, i.e., pδf ∝ p, indicating δf ~ constant. Figure  3.7 highlights the influence of H2 level on the variation of flame thickness with pressure. As shown, the decrease in flame thickness is slightly more significant for high H2 mixtures, especially at lower pressure ranges. The influence of preheat temperature on flame thickness of syngas fuels is seen in Figure 3.8, which shows the variation of δf for 300 < Tu < 700 K for a 50:50 H2/CO

79

Laminar Flame Properties of H2 /CO Mixtures 1.8

H2 = 20% H2 = 50% H2 = 80%

1.5 1.2

100 0.9

pδf (atm mm)

Flame Thickness, δf (µm)

1000

0.6 10

0

10

20 Pressure (atm)

30

0.3 40

Figure  3.7  Pressure dependence of laminar flame thickness (solid lines) and pressureflame thickness product (dashed lines) from simulations for three H2/CO fuel mixtures at Φ = 0.6 and Tu = 300 K.

700

Φ = 0.6 Φ = 0.8

Flame Thickness, δf (µm)

Φ = 1.0 600

Φ = 2.0

500

400

300 300

400

500 600 Preheat Temperature Tu (K)

700

Figure 3.8  Laminar flame thickness as a function of preheat temperature from simulations for four 50:50 H2/CO fuel–air mixtures at p = 1 atm.

fuel mixture. In general, the flame thickness increases with preheat temperature, in a roughly linear manner. For the leanest (Φ = 0.6) mixture, the reactant temperature has a small effect below ~450 K. Though the increase in preheat temperature increases the overall reaction rate, which tends to decrease the flame thickness,

80

Synthesis Gas Combustion: Fundamentals and Applications

the increase in thermal and mass diffusivities with preheat temperature dominates, hence the increase in flame thickness.

3.2.3 Flame Propagation The fundamental parameter that characterizes flame propagation in a premixed reactant mixture is the laminar flame speed (SL). Since a stationary flame is achieved when the local flow velocity normal to the flame equals the flame speed, SL is relevant to determining blow-off and flashback velocities of a burner (Zhang et al., 2007), as well as other burner parameters, such as quenching diameter. Apart from the utility of flame speed information in the design of fuel-flexible syngas combustion systems, SL is also a useful parameter for validating and optimizing chemical kinetic models. Extensive research has been performed to characterize the laminar flame speed for syngas-type fuel mixtures at a variety of operating conditions, as it is strongly influenced by reactant mixture composition, pressure, and preheat temperature. Laminar burning velocities for near-stoichiometric and fuel-rich CO/H 2 mixtures have been measured with conical flames stabilized with Mach Hebra nozzle burners (Scholte and Vaags, 1959), Bunsen burners (Günther and Janisch, 1971), spherically expanding flames (Strauss and Edse, 1958), and flat flames (Yumlu, 1967). In early flame speed studies, the effect of flame stretch (see Section 3.2.3.5) on the measured flame speed was not considered. In later studies, the stretch-corrected measurements of laminar flame speed and its strain sensitivity for H2/CO fuel mixtures were obtained in counterflow flames (Vagelopoulos and Egolfopoulos, 1994) and spherically expanding flames (McLean et al., 1994; Brown et al., 1996; Hassan et al., 1996, 1997; Sun et al., 2006). Vagelopoulos and Egolfopoulos (1994) and McLean et al. (1994) reported laminar flame speeds for H2/CO mixtures for a range of H2/CO and equivalence ratios. These stretch-corrected measurements are primarily for atmospheric pressure, though there are exceptions. Hassan et al. (1996, 1997) measured the laminar flame speed and Markstein length in spherically expanding flames at 4 atm for H2/CO fuel mixtures with only 5% H2. They reported that the onset of severe instabilities restricts the flame speed measurements for high H2 content mixtures even at a few atmospheres. In order to overcome these flame instabilities, Sun et al. (2006) used O2/He as an oxidizer rather than standard air and reported flame speed information for H2/CO fuel mixtures up to 10 atm for fuels containing up to 50% H2, and up to 40 atm for mixtures with 5% H2. Also, all the above data (both at atmospheric and elevated pressure) for H2/CO fuel mixtures are restricted to room temperature reactants. Laminar flame speeds of H2/CO mixtures have been reported for a range of reactant preheat temperatures (up to 700 K) and fuel compositions at both atmospheric (Natarajan et al., 2007a, 2007b, 2008b) and elevated (up to 15 atm) (Natarajan et al., 2008a, 2009) pressure conditions. These more recent studies not only include the influence of diluents such as CO2 and N2, but also focus on the lean mixtures relevant to many low emissions combustors. This section summarizes the findings of these studies, including the general influence of (1) H2/CO ratio, (2) pressure, (3) preheat temperature, (4) dilution, and (5) flame stretch on the laminar flame speed of syngas fuel mixtures. Many of these

Laminar Flame Properties of H2 /CO Mixtures

81

effects can be qualitatively modeled with a simplified analysis of a flame based on a balance between heat release and diffusion (Glassman, 1996). From this model, the flame speed can be modeled as

SL ∝

α ⋅ RR ρ

(3.1)

where α is the thermal diffusivity, RR is the overall reaction rate, and ρ is the unburned gas density. 3.2.3.1 H2/CO Ratio We begin by examining the influence of H2 level in the syngas fuel on laminar flame speed. The results again are based on simulations with Chemkin PREMIX. As shown in Figure 3.9, the flame speed increases as more H2 is added to the fuel. This is not due to changes in flame temperature, as Tad changes only slightly with H2 level, as noted above. The SL behavior is explained by the simple flame speed model presented in Equation 3.1: (1) the overall reactivity of the fuel mixture increases with the amount of H2, as illustrated in the flame structure description (Section 3.2.2), and (2) the low molecular weight of hydrogen acts to increase the diffusivity of the reactant mixture. Interestingly, the increase in flame speed is slightly more pronounced for low H2 amounts; the flame speed increases more rapidly as the amount of H2 increases to 15%. This is mainly due to the sensitivity of the CO oxidation rate to the presence of small amounts of hydrogen-containing species. As pointed out in the flame structure discussion, the main CO oxidation reaction shifts from the slower CO + O(+M) reaction to the relatively faster CO + OH reaction as the amount of H2 increases. The higher reaction rates result in the rapid increase in flame speed. For higher levels of H2, for example, >20%, there is still an increase in flame speed, partly due to higher diffusivities, though the sensitivity to H2 is reduced. Above ~60 to 70% H2 for the lean mixtures (Figure 3.9a), the sensitivity of the flame speed to H2 content slightly increases again. 3.2.3.2 Pressure The influence of pressure on SL is presented in Figure 3.10, based on simulations for a 50:50 H2/CO fuel mixture at lean to rich conditions, with the results normalized by the laminar flame speed at 1 atm. The laminar flame speed decreases nearly logarithmically with pressure (as indicated by the curve fits in the figure). The increase in pressure raises the overall reaction rate (RR), as molecular collision rates rise with pressure. Hence, higher pressures would tend to produce faster laminar flame speeds (see Equation 3.1). On the other hand, the increased density of the reactant mixture necessitates more thermal energy transfer from the reaction zone to raise the reactant temperature in the preheat zone. Since thermal diffusivity (α) is also inversely proportional to pressure, the net effect of increasing pressure is a reduced flame speed. Flame temperature changes with pressure are relatively small at these conditions (as noted previously), so flame temperature effects do not greatly influence the SL

82

Synthesis Gas Combustion: Fundamentals and Applications

160

Φ = 0.6 Φ = 0.8

Flame Speed (cm/s)

120

80

40

0

0

300

20

40

% H2 (a)

60

80

100

60

80

100

Φ = 1.0 Φ = 2.0

Flame Speed (cm/s)

240 180 120 60 0

0

20

40

% H2 (b)

Figure 3.9  Laminar flame speeds from simulations for a range of H2/CO fuel mixtures for (a) lean and (b) stoichiometric and rich cases, at p = 1 atm and Tu = 300 K.

pressure dependence. Unlike the flame thickness variations (see Figure 3.6), the drop in flame speed with pressure is slightly more pronounced for lean mixtures than for stoichiometric conditions. Also shown in Figure 3.10 is the variation of normalized mass burning rate (or equivalently for a one-dimensional flame, the mass flux = ρuSL ) with pressure. The

83

Laminar Flame Properties of H2 /CO Mixtures

SL/SL,atm

0.75

9

0.5

6

0.25

0

(ρuSL)/(ρuSL)atm

15 Φ = 0.6 Φ = 1.0 Φ = 2.0 12

1

3

1

10 Pressure (atm)

0 100

Figure 3.10  Laminar flame speed (solid lines) and mass burning rate (dashed lines) as a function of pressure for a 50:50 H2/CO fuel mixture at Tu = 300 K (symbols = simulations, lines = curve fits).

mass burning rate specifies the reactant consumption rate, and therefore the flame’s heat release rate (Watts or BTU/s). As opposed to SL , the mass burning rate rises with pressure, as the increase in reactant density is greater than the decrease in flame speed. The increase in mass burning rate with pressure is less pronounced for the lean (Φ = 0.6) mixture, which also exhibits a drop in the pressure dependence above ~10 to 15 atm. Varying the H2 level in the fuel has only a weak effect on the pressure dependence of the flame speed and mass burning rate. This is illustrated in Figure 3.11, which shows results for the lean case (Φ = 0.6). The drop in normalized flame speed with pressure is slightly more pronounced for high H2 content fuel mixtures, while the increase in normalized burning rate with pressure is less pronounced (above ~10 atm). One way to capture the pressure dependence of flame propagation speeds is through simplified models, such as those that employ single-step (overall) chemical reaction rates (Egolfopoulos and Law, 1997) and assume the flame has a two-zone structure (a broad diffusion dominated zone followed by a thin reaction zone). While the flame structures shown in Section 3.2.2 reveal that reactions occur throughout the flame zone in fuels with significant levels of hydrogen, it is still useful to analyze syngas flames with these assumptions. In this case, we find

 E R m ′′ ≈ ρSL ∝ p n 2 ρα exp − a   2Trz 

(3.2)

where m ′′ is the burning mass flux, n is the overall reaction order, Ea /R is the activation energy of the overall reaction normalized by the gas constant, and Trz is the apparent reaction zone temperature (typically modeled as Tad for thin reaction zones).

84

Synthesis Gas Combustion: Fundamentals and Applications 10 8

SL/SL,atm

0.75 6 0.5

4

0.25

0

(ρuSL)/(ρuSL)atm

H2 = 20% H2 = 50% H2 = 80%

1

2

1

10 Pressure (atm)

0 100

Figure  3.11  Laminar flame speed (solid lines) and mass burning rate (dashed lines) as a function of pressure for three H2/CO fuel mixtures at Φ = 0.6 and Tu = 300 K (symbols = simulations, lines = curve fits).

Neglecting small changes in ρα and Tad with pressure due to compositional variations, the reaction order can be calculated from Equation 3.2:

 ∂ ln m ′′  n = 2   ∂ ln p  Tad

(3.3)

The apparent reaction order has been calculated for a 50:50 H2/CO fuel mixture for lean, stoichiometric, and rich conditions at various pressures (Figure  3.12). At atmospheric pressure, n is ~1.9 for the rich (Φ = 2) case, 1.75 for the stoichiometric mixture, and 1.5 for the lean (Φ = 0.6) case. So for low pressures, n drops with equivalence ratio. In addition, the reaction order decreases monotonically with pressure for all the cases shown. However, the pressure dependence is most pronounced for the rich mixture. So for high pressures (above 20 to 30 atm), the rich and lean mixtures have similar apparent reaction orders, while the stoichiometric mixture has a higher n. The pressure dependence of the flame propagation speeds also brings to light the relative importance of chain branching and terminating reactions in syngas flames. For example, one of the main chain branching reactions in the H2/O2 system is H + O2 ↔ OH + O, which is a two-body reaction. On the other hand, the chain terminating reaction H + O2 + M ↔ HO2 + M is third order. Hence, a pressure rise will tend to increase the relative rate of the H chain termination reaction compared to the chain branching step. This leads to a reduction of overall reaction rate, which hinders flame propagation. Since lean mixtures have a lower n at low pressures, a reduction in H atoms affects flame propagation more for lean mixtures—hence the drastic reduction in mass burning rates for lean conditions. Moreover, the fact that n ~ 2 at

85

Laminar Flame Properties of H2 /CO Mixtures 2

Φ = 0.6 Φ = 1.0

Reaction Order, n

1.75

Φ = 2.0

1.5 1.25 1 0.75 0.5

1

10 Pressure (atm)

100

Figure 3.12  Variation of overall reaction order with pressure for a 50:50 H2/CO fuel mixture at three equivalence ratios and Tu = 300 K (symbols = simulations, lines = logarithmic fits).

low pressures indicates the dominance of the two-body branching reaction (over the three-body terminating reaction) there. The amount of H2 in the fuel mixture also influences the overall reaction order. For example, Figure 3.13 shows results for three H2 levels at a fixed fuel–air ratio. While variations in H2 level do not change the monotonic decrease in n with pressure, the H2 level does change the value of n. For pressure above ~2–3 atm, high H2 1.75

H2 = 20% H2 = 50%

Reaction Order, n

1.5

H2 = 80%

1.25 1 0.75 0.5 0.25

1

10 Pressure (atm)

100

Figure 3.13  Variation of reaction order with pressure for three H2/CO fuel mixtures at Φ = 0.6 and Tu = 300 K (symbols = simulations, lines = logarithmic fits).

86

Synthesis Gas Combustion: Fundamentals and Applications

content syngas mixtures have a lower reaction order. This suggests that the H termination reaction is more prominent for high H2 mixtures, which should have higher H atom concentrations. In addition, it is useful to note that the calculated n is positive for all the cases considered, which indicates that the mass burning rate will increase with pressure for these syngas mixtures. However, n drops rapidly with pressure for the high H2 mixture, and negative n values have been observed for lean, pure H2 mixtures at ~20 atm (Law and Sung, 2000). Hence, it is possible that mass burning rates may drop with pressure for high H2 syngas fuels under some conditions. 3.2.3.3 Preheat Temperature Reactant preheating occurs in various combustion systems, for example, due to compression (in high-pressure combustion) or waste heat recovery. Therefore, it is important to examine the dependence of flame speed on reactant temperature. According to the model represented by Equation 3.1, reactant preheating influences the laminar flame speed mainly through changes in reaction rate and diffusive properties. At a constant pressure, the functional dependence of laminar flame speed on preheat temperature can be approximated from the simplified model of Equation 3.2, that is,

SL ∝

 E R α exp − a  ρ  2Trz 

(3.4)

An increase in preheat temperature increases the reaction zone temperature (Trz ) and the diffusivity (α) of the mixture, and decreases the unburned density; all tend to increase SL . To explore the sensitivity to Tu, we begin by presenting a comparison of measured and simulated laminar flame speeds for atmospheric pressure, lean syngas fuels for various amounts of preheating (Figure 3.14) (Natarajan et al., 2007a, 2007b). For the low-temperature cases (300 to 500 K), the simulations produce a fairly close approximation of the measurements, which should help assure the reader that the simulation results presented thus far provide a reasonably accurate representation of syngas flame speeds. For the higher reactant temperatures (Tu > 500 K), the simulations provide a less accurate match to the data, but qualitatively the results are similar. Thus, the simulation results presented below for high reactant temperatures should be considered primarily as indicators of the general influence of preheat. Figure 3.15 shows a similar comparison of measured and simulated laminar flame speeds for different syngas compositions at elevated pressure (Natarajan et al., 2009). At both the low (300 K) and high (600 K) reactant temperatures, the flame speed increases as H2 is added to the fuel, as demonstrated previously (Section 3.2.3.1). The two preheat cases have different equivalence ratios, which were adjusted in order to achieve nearly the same adiabatic flame temperature for both preheat temperatures. This approach isolates the effect of preheat temperature. Thus, the increase in flame speed with Tu for fixed Tad presented in Figure 3.15 can be attributed to a combination of increase in diffusivity and the increase in the reaction rate in the leading part of the syngas flame (that is, the HO2 formation zone presented in Figure 3.4).

87

Laminar Flame Properties of H2 /CO Mixtures

Flame Speed (cm/s)

Tu = 700 K

C1 Mech (Li et al.)

500

Tu = 600 K

400 300

Tu = 500 K

200

Tu = 400 K

100

Tu = 300 K

0 0.55

0.65

0.75 0.85 Equivalence Ratio

0.95

1.05

Figure 3.14  Laminar flame speed for a 50:50 H2:CO syngas composition for various preheat temperatures at p = 1 atm (symbols = measurements, lines = simulations) (Natarajan et al., 2007a).

150

C1 Mech (Li et al.)

Flame Speed (cm/s)

120

Tu = 600 K; Φ = 0.6

90

60 Tu = 300 K; Φ = 0.8

30

0

0

20

40

% of H2

60

80

100

Figure 3.15  Laminar flame speeds for a range of H2/CO fuel mixtures (1:9 O2:He oxidizer) at elevated pressure (p = 15 atm) (symbols = measurements, lines = simulations) (Natarajan et al., 2009).

88

Synthesis Gas Combustion: Fundamentals and Applications 8

Φ = 0.6

SL/SL,atm and (ρuSL)/(ρuSL)atm

Φ = 0.8 6

Φ = 1.0 Φ = 2.0

4

2

0 300

400

500 600 Preheat Temperature Tu (K)

700

Figure 3.16  Normalized flame speed (solid lines) and mass burning rate (dashed lines) as a function of preheat temperature and equivalence ratio from simulations for a 50:50 H2/CO fuel at p = 1 atm.

Figure 3.16 provides a more detailed examination of the influence of Tu for different fuel–air ratios, with results shown for both laminar flame speed and mass burning rate. The simulations reveal SL increases exponentially with Tu; as Tu rises from room temperature to 700 K, the flame speed increases by a factor of 4 to 8 for the range of equivalence ratios shown. Overall, flame speeds and mass burning rates of lean mixtures show the most sensitivity to preheating. It is interesting to note that the rate of increase in mass burning rate is less than that for flame speed. As the reactants are heated from room temperature to 700 K, the mass burning rate increases by only two to three times. This is due to the decrease in the density of the unburned reactants with preheat temperature. 3.2.3.4 Diluents As previously noted, syngas fuels exhibit a large variability in composition, not only in the combustible fuel content, but also in diluents such as CO2, N2, and H2O. Fuel dilution impacts flame propagation in at least three ways, through changes in (1) mixture-specific heat and adiabatic flame temperature, (2) chemical kinetic rates, and (3) radiative heat transfer. First, addition of diluents to the fuel reduces the flame temperature, and thus reduces the laminar flame speed (and narrows the flammability range) through reduction in the reaction rate. From Equation 3.4, the effect of flame temperature on overall reaction rate can be estimated through the activation energy (Ea) term. This global parameter can be determined from Equation 3.4, assuming that λ/cp is nearly constant; the result is

89

Laminar Flame Properties of H2 /CO Mixtures

d ln m ′′ E =− a d (1 Tad ) 2R



(3.5)

Overall Activation Energy, Ea (kcal/mole)

To properly determine the activation energy, it is thus important to vary only the flame temperature. One approach for estimating Ea is to change Tad by replacing a small amount (2 to 6%) of the N2 in air with argon. This leads to a small increase in the flame temperature and hence the mass burning rate. Based on this approach, the calculated overall activation energy for a 50:50 syngas mixture at two lean conditions over a range of preheat temperatures is shown in Figure 3.17. Both lean mixtures show an increase in Ea with preheat temperature, reaching a nearly constant value above Tu ~ 600 K. As higher Ea values imply a greater sensitivity to flame temperature (see Equation 3.4), one can expect that diluent effects on flame speed may be more pronounced for preheated conditions. The simulations also indicate that Ea for the leaner (Φ = 0.6) condition is significantly less than for the more stoichiometric (Φ = 0.8) case for this 50:50 H2/CO fuel. It should be noted that this trend is opposite of that observed for pure H2 under lean conditions using the same calculation approach (Law and Sung, 2000). Second, some diluents are not inert. For example, the chemical kinetic effects of CO2 dilution can be manifested through the main CO oxidation reaction CO + OH ↔ CO2 + H. Higher CO2 dilution levels lead to enhanced backward reaction rates, and hence reduced rates of CO oxidation and H atom production. This will reduce flame propagation speeds. Chemical kinetic studies have emphasized this

70

Φ = 0.6 Φ = 0.8

60

50

40

30 200

300

400 500 600 Preheat Temperature Tu (K)

700

Figure  3.17  Variation of overall activation energy with preheat temperature for 50:50 H2/CO fuel mixtures at lean equivalence ratios and p = 1 atm (symbols = simulations, lines = curve fits).

90

Synthesis Gas Combustion: Fundamentals and Applications

point by comparing the flame speeds of mixtures with CO2 dilution, and a fictitious, chemically inert species with the same specific heat as CO2. The results showed that CO2 diluted flame speeds were lower (Zhu et al., 1988). This kinetic effect is expected to have profound influence on flame speeds for lean H2/CO flames due to the importance of H atom concentration, especially for low H2 syngas fuels. As seen in the flame structure examples, H atoms control the main branching (H + O2 ↔ O + OH) and termination (H + O2 + M ↔ HO2 + M) reactions. Third, addition of diluents can influence flame propagation through radiative heat transfer. For example, CO2 and H2O are effective absorbers and emitters of infrared radiation, unlike N2 and O2. The presence of significant amounts of CO2 or H2O in the reactants can result in reabsorption of energy radiated from the hot products. This enhancement of thermal energy transfer across the flame can increase laminar flame speeds and extend flammability limits (Ruan et al., 2001; Chen et al., 2006; Ju et al., 1998). Conversely, the presence of an effective radiative emitter within the initial part of the reaction zone can also lead to increased heat losses and reduced flame speeds. 3.2.3.5 Flame Stretch So far we have considered the influence of various physical and chemical parameters on laminar flame speed of a freely propagating, planar (one-dimensional) flame. However, such flames are rarely found in practical devices, and actually difficult to create even in controlled environments. Thus, we must also consider propagation speeds of flames that are not one-dimensional. Laminar flame speed and thus overall heat release rate are strongly influenced by flame stretch, especially for reactant mixtures consisting of species with drastically different diffusivities. The stretch (κ), defined as the fractional rate of change of flame surface area, is mainly due to tangential strain (velocity nonuniformity in the approach flow) and flame curvature (Law and Sung, 2000). For example, flow strain is relevant in a stagnation flame, while a propagating nonplanar flame can have stretch effects due solely to the motion of the curved flame. In this section, we characterize stretch effects on unburned flame speeds due to strain for typical syngas mixtures. The results are based on simulations of strained flames obtained with the Chemkin OPPDIF code (Kee et al., 2006), for identical premixed mixtures exiting opposing nozzles. Based on a standard approach (Wu and Law, 1984), the minimum velocity before the preheat zone is defined as the reference strained unburned flame speed (Su ), and the maximum gradient of the axial velocity ahead of the minimum velocity location is defined as the imposed strain rate (K). As an example, Figure 3.18 shows the effect of strain rate on the unburned flame speed for a lean, 10:90 H2:CO fuel mixture. Typically for lean H2/CO mixtures with low to moderate strain, the unburned flame speed increases linearly with imposed strain rate. For this low H2 content fuel, this is accompanied by a drop in the maximum flame temperature, due to a decrease in residence time in the reaction zone, thus leading to incomplete combustion. The linear dependence of flame speed on strain is generally expressed by the equation Su = Su0 – L M κ, where Su0 is the unstrained flame speed and L M is the unburned Markstein length (Law and Sung, 2000). The Markstein length is essentially the sensitivity of the unburned flame speed to strain.

91

32

1950

28

1925 Su = 0.038 K + 24.8

24

1900

20

1875

Maximum Temperature (K)

Flame Speed Su (cm/s)

Laminar Flame Properties of H2 /CO Mixtures

Flame speed 16

Max. temperature 0

30

60 90 Strain Rate K (1/s)

120

1850 150

Figure  3.18  Strained laminar flame speed and maximum flame temperature for a lean (Φ = 0.6) 10:90 H2:CO composition at p = 1 atm and Tu = 300 K (symbols = simulations, lines = linear fits).

As an example, the unstrained flame speed, based on extrapolating the simulations to zero strain rate in Figure 3.18, is 24.7 cm/s, while L M is 0.038 cm. Like laminar flame speed, the Markstein length is also affected by various parameters, such as the amount of H2 in the fuel mixture, equivalence ratio, pressure, and preheat temperature. Usually L M tends to scale with laminar flame thickness. To demonstrate this, Figure 3.19 shows the results of strained flame simulations for a 50:50 H2:CO fuel mixture, again for Φ = 0.6, and at two pressures. The influence of H2 level in the syngas is illustrated by a comparison of the results for the 10:90 (Figure 3.18) and 50:50 (Figure 3.19) H2:CO fuel mixtures at p = 1 atm and Φ = 0.6. As expected from the SL results presented previously, Su0 increases for the higher H2 content fuel. In addition, LM decreases from 0.038 cm to 0.026 cm. This correlates well with the reduced flame thickness for the higher H2 fuel (see Figure 3.5). The influence of pressure is evident from comparison of the 1 and 5 atm results shown in Figure 3.19. The higher-pressure condition produces a lower unstrained flame speed and lower Markstein length. Again, the reduced LM value parallels the demonstrated decrease in syngas flame thickness with pressure (see Figure 3.6). Similarly, the qualitative dependence of Markstein length on other parameters such as equivalence ratio and preheat temperature for syngas fuels can be determined from their influence on laminar flame thickness (Natarajan et al., 2007b, 2008b). Interestingly, increasing the amount of H2 in the H2/CO fuel mixture (from 10:90 to 50:50) also inverts the dependence of the maximum flame temperature on the strain rate (see Figures 3.18 and 3.19). For the 50:50 fuel mixture, the maximum flame temperature now increases with the strain rate. Though the increase in strain rate still decreases the residence time in the reaction zone, the addition of H2 also

92

Synthesis Gas Combustion: Fundamentals and Applications

Flame Speed Su (cm/s)

Su = 0.0258 K + 48.5

P = 1 atm P = 5 atm Max. temperature

55

2000

1975

Flame speed 1950

45 1925 35

25

Su = 0.0097 K + 28.9

0

250

500 Strain Rate K (1/s)

750

1900

Maximum Temperature (K)

65

1875 1000

Figure  3.19  Strained laminar flame speed and maximum flame temperature for a lean (Φ = 0.6) 50:50 H2:CO composition at two pressures and Tu = 300 K (symbols = simulations, lines = linear fits).

significantly decreases the flame’s Lewis number (Le, the ratio of thermal to mass diffusivity) (Law and Sung, 2000). Thus, there is a faster diffusion of reactants toward the reaction zone compared to its loss of thermal energy through conduction back to the fresh reactants for this positively strained flame; the net result is a local increase in the reaction zone temperature. This is further enhanced by preferential diffusion, owing to the higher diffusivity of H2. The result is a local increase in the equivalence ratio (toward Φ = 1) for this positively strained, lean mixture. As the strain rate is increased further, the drop in residence time eventually dominates, leading to a decrease in flame temperature.

3.2.4 Flame Extinction At sufficiently high imposed stretch rates (for example, due to higher deceleration rates experienced by a stagnating flow), the flame temperature and heat release rate begin to decrease drastically, and the flame eventually extinguishes. The corresponding strain rate is called the extinction strain rate (Kext ). Similar to laminar flame speed, extinction strain rate is a fundamental property of the reactant mixture, contributing to phenomena such as flame liftoff and combustor blowout. Extinction strain rates have also been used for partial validation of detailed chemical kinetic models (Dong et al., 2005). Various research efforts have focused on characterizing extinction strain rates for a variety of reactant mixtures that include CH4, H2, and CO (Vagelopoulos and Egolfopoulos, 1994; Law et al., 1986; Jackson et al., 2003; Choudhuri et al., 2008). Most of these experimental and numerical studies utilized the premixed opposing jet

93

Laminar Flame Properties of H2 /CO Mixtures

Extinction Strain Rate, Kext (1/s)

2000

1500

1000

500

0 0.4

0.7

1

1.3

1.6

Equivalence Ratio Φ

Figure 3.20  Extinction strain rates for premixed CH4-air flames at various equivalence ratios at p = 1 atm and Tu = 300 K (symbols = measurements, line = curve fit) (Law et al., 1986).

flow configuration to directly measure the extinction rates. As a basis for comparison, we begin by examining measured extinction strain rates (Figure 3.20) for CH4air flames (Law et al., 1986). Similar to SL and Tad, the extinction strain rate generally increases as the mixture equivalence ratio is adjusted toward stoichiometric. In other words, higher strain rates are required to extinguish fast-burning and high flame temperature reactant mixtures. Unlike SL and Tad, Kext peaks for slightly lean (rather than rich) mixtures. This behavior can be explained by considering both stretch and preferential diffusion effects. For rich CH4-air mixtures, Le is greater than 1, and hence the flame temperature drops with increasing stretch. But for lean CH4-air mixtures, Le is less than 1, and the flame temperature rises with stretch (at least before the onset of incomplete combustion), requiring more stretch to extinguish the flame. The presence of small amounts of H2 in relatively slow-burning reactant mixtures, such as CH4-air or CO-air, can strongly influence the extinction characteristics due to the intense burning nature of H2. Figure 3.21 shows measured extinction strain rates for various H2-doped, lean CH4-air flames at elevated preheat temperature (Tu = 578 K). As with the room temperature data (Figure 3.20), the extinction strain rate increases with equivalence ratio. Comparing the pure (0% H2 doping) case at Tu = 578 K (Figure 3.21) with room temperature results (Figure 3.20), the extinction strain rate increases significantly with preheat temperature for a given equivalence ratio. For example, Kext increases from ~750 to ~4000 s–1 due to the ~280 K increase in reactant temperature. The higher Kext values are associated with greater burning rates and flame temperatures for the preheated mixtures. For the same reasons, higher H2 levels are also seen to increase extinction strain rate (Figure 3.21). Figure 3.21 further illustrates that increasing the amount of H2 also extends the lean flammability limit of the reactant mixture. This can be observed by extending each

94

Synthesis Gas Combustion: Fundamentals and Applications

Extinction Strain Rate, Kext (1/s)

8000

H2 = 0% H2 = 5% H2 = 10%

6000

4000

2000

0 0.4

0.5

0.6 0.7 Equivalence Ratio Φ

0.8

0.9

Figure  3.21  Extinction strain rates for atmospheric pressure, premixed CH4-air flames with various amounts of H2 addition (0, 5, and 10%) at elevated preheat temperature (Tu = 578 K) (symbols = measurements, lines = curve fits) (Jackson et al., 2003).

of the constant H2 curves to Kext = 0, which in theory should correspond to the flammability limit of the reactant mixture. For syngas mixtures, the amount of H2 has a similarly significant influence on extinction characteristics. Figure  3.22 shows the variation of measured extinction strain rates for two lean mixtures as a function of H2 content in the fuel. As before, Kext increases with H2 and Φ due to the increase in burning rate with both, and due

Extinction Strain Rate, Kext (1/s)

800

Φ = 0.39 Φ = 0.32

600

400

200

0

0

10

20 30 % H2 in H2/CO Mixture

40

50

Figure  3.22  Extinction strain rates as a function of H2 level in H2/CO-air flames for two lean mixtures at p = 1 atm and Tu = 300 K (symbols = measurements, lines = curve fits) (Vagelopoulos and Egolfopoulos, 1994).

95

Laminar Flame Properties of H2 /CO Mixtures

to the increase in flame temperature with Φ. The addition of diluents to syngas fuels will tend to lower the extinction strain rate. This point has been emphasized in studies on the effects of CO2 dilution of CH4 and H2 fuels (Ren et al., 2001).

3.3 Nonpremixed Flame Properties The flame properties presented above were for perfectly premixed reactants, that is, where mixing has occurred well before the reactants encroach upon the flame region. In some combustions systems, premixing of fuel and oxidizer is not desirable, for example, due to flame stability or safety concerns. Thus, it is also helpful to describe the characteristics of laminar syngas flames that are nonpremixed. An example of such a flame structure is shown in Figure 3.23, for a 50:50 H2:CO fuel composition. As above, the simulation results were obtained with the Chemkin OPPDIF code. For this example, the flame occurs between two opposing jets, which is the same geometry used above to characterize stretch effects on laminar flames. In this case, the two nozzles contain different reactants: the fuel exits from one nozzle and air from the other. The figure shows the chemical composition (major species) and temperature as a function of distance along the axial centerline between the two nozzles. Various characteristics of undiluted, nonpremixed syngas flames are visible here. First, the peak temperature, which is sometimes considered a marker of the flame location, occurs on the oxidizer side of the stagnation plane, since a stoichiometric mixture of fuel and air requires more air. Thus, the fuel species must diffuse across the stagnation plane to reach the oxidizer. The high-temperature region is also typically characterized by high heat release rates, as well as the peak levels of flame radicals, such as O, OH, and H. The peaks in the H2O and CO2 mole fractions also occur here; for syngas fuels diluted with high levels of H2O or CO2, however, the major product peaks can occur elsewhere, depending on the level of dilution and the strain 1

2500 Stagnation Plane

T

Mole Fra ction

CO

1500

0.5 1000

H2

H2 O

0.25

O2

Temperature (K)

2000

0.75

500

CO2 0

0 0

2.5 5 7.5 Distance from Fuel Nozzle (mm)

10

Figure  3.23  Simulated structure of a nonpremixed strained (opposing jet flow, 170 s –1 strain rate) laminar flame for a 50:50 H2:CO fuel at p = 1 atm and Tu = 300 K.

96

Synthesis Gas Combustion: Fundamentals and Applications

rate (or residence time within the flame). It is also interesting to note how the mole fractions of CO and H2 begin to diverge some distance from the fuel nozzle due to the much higher diffusivity of the lighter H2. This is one of the key features of nonpremixed syngas flames. While the flame structure presented is characteristic for nonpremixed syngas flames, it should be noted that strain rate plays a key role, as it did for premixed strained flames. At sufficiently high levels of strain, peak temperatures and radical and product mole fractions drop, eventually reaching a point where the flame is extinguished. Analogous to the extinction strain trends observed for premixed flames, higher levels of strain are required to extinguish the flame as one increases the fraction of H2 in the fuel, raises the initial temperature of either reactant, or reduces the diluent level in the fuel (analogous to reducing the equivalence ratio for lean premixed flames).

3.4 Conclusions Understanding the fundamental laminar flame properties of syngas fuel mixtures is a critical aspect for developing syngas-fueled combustion devices. One of the major variables in syngas combustion is the composition of the fuel. While variations in the H2/CO ratio have little effect on flame temperature, they have a major impact on flame thickness, flame speed, flammability limits, stretch sensitivity, and extinction strain rate. Due to the high reactivity and diffusivity of hydrogen, the flame structure of syngas flames is also unlike that of natural gas flames. Additional components of many syngas fuels are diluents such as H2O and CO2. Understanding the influence of these diluents requires attention to their impact on radiative energy transfer and, as they are also combustion products, to their tendency to reduce overall product formation rates. This chapter also describes the dependence of laminar syngas flame properties on various physical parameters (pressure, reactant temperature, and stretch), which can in some cases be modeled and interpreted using global flame parameters such as overall reaction order, overall activation energy, and mixture thermal diffusivity. The sensitivity to flame stretch can also be correlated to flame thickness in premixed flames. These simplified modeling approaches can be beneficial in the development of scaling laws for combustor design tools.

References Brown, M. J., McLean, I. C., Smith, D. B., and Taylor, S. C. (1996). Markstein lengths of CO/H2/air flames, using expanding spherical flames. Proc. Combust. Instit. 26:875. Chen, Z., Qin, X., Xu, B., Ju, Y., and Liu, F. (2006). Studies of radiation absorption on flame speed and flammability limit of CO2 diluted methane flames at elevated pressures. Proc. Combust. Instit. 31:2693. Choudhuri, A. R., Subramanya, M., and Gollahalli, S. R. (2008). Flame extinction limits of H2-CO fuel blends. J. Eng. Gas Turbines Power 130:031501. Dong, Y., Holley, A. T., Andac, M. G., Egolfopoulos, F. N., Davis, S. G., Middha, P., and Wang, H. (2005). Extinction of premixed H2/air flames: Chemical kinetics and molecular diffusion effects. Combust. Flame 142:374.

Laminar Flame Properties of H2 /CO Mixtures

97

Egolfopoulos, F. N., and Law, C. K. (1997). Chain mechanisms in the overall reaction order in laminar flame propagation. Combust. Flame 80:7. Glassman, I. (1996). Combustion. 3rd ed. New York: Academic Press. Günther, R., and Janisch, G. (1971). Messwerte der Flammegeschwindigkeit von gasen und gasmischen. Chemie-Ing-Technol. 43:975. Hassan, M. I., Aung, K. T., and Faeth, G. M. (1996). Markstein numbers and unstretched laminar burning velocities of wet carbon monoxide flames. Paper AIAA 96-0912 presented at the 34th Aerospace Sciences Meeting and Exhibit, Reno, NV. Hassan, M. I., Aung, K. T., and Faeth, G. M. (1997). Properties of laminar premixed CO/H2/air flames at various pressures. J. Propulsion Power 13:239. Jackson, G. S., Sai, R., Plaia, J. M., Boggs, C. M., and Kiger, K. T. (2003). Influence of H2 on the response of lean premixed CH4 flames to high strained flows. Combust. Flame 132:503. Ju, Y., Masuya, G., and Ronney, P. D. (1998). Effects of radiative emission and absorption on the propagation and extinction of premixed gas flames. Proc. Combust. Instit. 27:2619. Kee, R. J., Rupley, F. M., Miller, J. A., Coltrin, M. E., Grcar, J. F., Meeks, E., Moffat, H. K., Lutz, A. E., Dixon-Lewis, G., Smooke, M. D., Warnatz, J., Evans, G. H., Larson, R. S., Mitchell, R. E., Petzold, L. R., Reynolds, W. C., Caracotsios, M., Stewart, W. E., Glarborg, P., Wang, C., and Adigun, O. (2006). CHEMKIN collection. Release 4.1. San Diego: Reaction Design. Law, C. K. (2006). Combustion physics. Cambridge, UK: Cambridge University Press. Law, C. K., and Sung, C. J. (2000). Structure, aerodynamics and geometry of premixed flamelets. Prog. Energy Comb. Sci. 26:459. Law, C. K., Zhu, D. L., and Yu, G. (1986). Propagation and extinction of stretched premixed flames. Proc. Combust. Instit. 21:1419. Li, J., Zhao, Z., Kazakov, A., Chaos, M., Dryer, F. L., and Scire Jr., J. J. (2007). A comprehensive kinetic mechanism for CO, CH2O, and CH3OH combustion. Int. J. Chem. Kin. 39:109. McLean, I. C., Smith, D. B., and Taylor, S. C. (1994). The use of carbon monoxide/hydrogen burning velocities to examine the rate of the CO + OH reaction. Proc. Combust. Instit. 25:749. Natarajan, J., Kochar, Y., Lieuwen, T., and Seitzman, J. (2008a). Laminar flame speed measurements of H2/CO/CO2 mixtures up to 15 atm and 600 K preheat temperature. ASME Paper 08-GT2008-51364 presented at Proceedings of the ASME/IGTI Turbo Expo 2008, Berlin. Natarajan, J., Kochar, Y., Lieuwen, T., and Seitzman, J. (2009). Pressure and preheat dependence of laminar flame speeds of H2/CO/CO2/O2/He mixtures. Proc. Combust. Instit., doi:10.1016/j.proci.2008.06.110. Natarajan, J., Lieuwen, T., and Seitzman, J. (2007a). Laminar flame speeds of H2/CO mixtures: Effect of CO2 dilution, preheat temperature and pressure. Combust. Flame 151:104. Natarajan, J., Lieuwen, T., and Seitzman, J. (2007b). Laminar flame speeds and strain sensitivities of mixtures of H2 with CO, CO2 and N2 at elevated temperatures. ASME Paper 07-GT2007-27967 presented at Proceedings of the ASME/IGTI Turbo Expo 2007, Montreal, Canada. Natarajan, J., Lieuwen, T., and Seitzman, J. (2008b). Laminar flame speeds and strain sensitivities of mixtures of H2/O2/N2 at elevated preheat temperatures. J. Eng. Gas Turbines Power 30:061502. Ren, J.-Y., Qin, W., Egolfopoulos, F. N., Mak, H., and Tsotsis, T. T. (2001). Methane reforming and its potential effect on the efficiency and pollutant emissions of lean methane-air combustion. Chem. Eng. Sci. 56:1541. Ruan, J., Kobayashi, H., Niioka, T., and Ju, Y. (2001). Combined effects of nongray radiation and pressure on premixed CH4/O2/CO2 flames. Combust. Flame 124:225. Scholte, T. G., and Vaags, P. B. (1959). Burning velocities of mixtures of hydrogen, carbon monoxide, and methane with air. Combust. Flame 3:511.

98

Synthesis Gas Combustion: Fundamentals and Applications

Strauss, W. A., and Edse, R. (1958). Burning velocity measurements by the constant-pressure bomb method. Proc. Combust. Instit. 7:377. Sun, H. Y., Yang, S. I., Jomass, G., and Law, C. K. (2006). High pressure laminar flame speeds and kinetic modeling of carbon monoxide/hydrogen combustion. Proc. Combust. Instit. 31:439. Vagelopoulos, C. M., and Egolfopoulos, F. N. (1994). Laminar flame speeds and extinction strain rates of mixtures of carbon monoxide with hydrogen, methane, and air. Proc. Combust. Instit. 25:1317. Wu, C. K., and Law, C. K. (1984). On the determination of laminar flame speeds from stretched flames. Proc. Combust. Instit. 20:1941. Yumlu, V. S. (1967). Prediction of burning velocities of carbon monoxide-hydrogen-air flames. Combust. Flame 11:190. Zhang, Q., Noble, D. R., and Lieuwen, T. (2007). Characterization of fuel composition effects in H2/CO/CH4 mixtures upon lean blowout. J. Eng. Gas Turbines Power 129:688. Zhu, D. L., Egolfopoulos, F. N., and Law, C. K. (1988). Experimental and numerical determination of laminar flame speeds of methane/(Ar, N2, CO2)–air mixtures as function of stoichiometry, pressure, and flame temperature. Proc. Combust. Instit. 22:1537.

Combustion 4 Fundamental Characteristics of Syngas Guillaume Ribert, Piyush Thakre, Zhe Wang, Richard A. Yetter, and Vigor Yang Contents 4.1 Introduction.....................................................................................................99 4.2 Premixed Systems.......................................................................................... 101 4.2.1 Ignition and Extinction in Closed Homogeneous and Perfectly Stirred Reactors................................................................................. 101 4.2.2 Premixed Flames............................................................................... 104 4.2.2.1 CO/H2 Mixtures.................................................................. 104 4.2.2.2 CO2, H2O, and N2 Dilution................................................. 107 4.2.2.3 Pressure and Temperature Effects...................................... 112 4.3 Counterflow Diffusion Flames...................................................................... 114 4.3.1 H2/CO Mixtures................................................................................. 117 4.3.2 CO2, H2O, and N2 Dilution................................................................ 117 4.3.3 Pressure Effects................................................................................. 120 4.4 Strained Counterflow Premixed Flames....................................................... 124 4.5 Conclusions.................................................................................................... 124 Acknowledgments................................................................................................... 126 References............................................................................................................... 126

4.1 Introduction Synthetic gas (syngas) is primarily a mixture of hydrogen (H2) and carbon monoxide (CO). Varying proportions of carbon dioxide (CO2), steam (H2O), nitrogen (N2), and small quantities of hydrocarbons, mainly methane (CH4), may also be present (Shilling and Lee, 2003). In order to employ syngas as a practical fuel for power and propulsion applications, detailed understanding and characterization of the key ­syngas combustion properties, such as self-ignition delay, flame speed, extinction limit, and heat-release rate, over a wide range of permissible fuel compositions, equivalence ratios, and operating conditions, will be necessary. Scholte and Vaags (1959) were the first to experimentally study the premixed flame speeds of mixtures of H2/CO, CO/CH4, and H2/CH4 with air, using various compositions. A decade later, kinetics studies of H2/CO oxidation based on shocktube experiments were carried out (Brabbs et al., 1970; Gardiner et al., 1972; Dean et al., 1978). McLean et al. (1994) obtained H2/CO burning velocities using a spherical 99

100

Synthesis Gas Combustion: Fundamentals and Applications

bomb apparatus. All the above measurements were limited to low pressures. Kim et al. (1994) studied the oxidation of moist CO up to 9.6 atm using a flow reactor. More recently, several experimental investigations of H2/CO autoignition and flame speeds were carried out at much higher pressures (Mittal et al., 2006; Sivaramakrishnan et al., 2007; Walton et al., 2007; Sun et al., 2007). One of the questions concerning the syngas oxidation mechanism was the relative importance of the dry oxidation of CO (R1, R2) and the oxidation of CO in the presence of hydrogen. It is now well known that while the reaction of the dry CO/O2 mixture is slow even at high temperatures, the overall reactivity is greatly accelerated if trace amounts of H2 and moisture are present. The oxidation route between CO and hydroxyl radicals (R3) is the dominant pathway, and it accounts for a significant portion of the heat release (Sung and Law, 2008). At higher pressures, however, reaction (R4) provides another route for the conversion of CO to CO2 (Sung and Law, 2008).

CO + O2 → CO2 + O

(R1)



CO + O + M → CO2 + M

(R2)



CO + OH → CO2 + H

(R3)



CO + HO2 → CO2 + OH

(R4)

A detailed reaction mechanism for H2/CO/O2 was developed by Yetter et al. (1991a, 1991b). Subsequently, a few analytical studies focused on modeling wet CO premixed flames in the presence of H2 and H2O, with simplified reduced chemistry (Wang and Rogg, 1993; Rightley and Williams, 1995, 1997). The Yetter et al. (1991a, 1991b) mechanism has since been updated (Kim et al., 1994) and further extended by Li et al. (2007) to predict a wide range of flame characteristics. Other mechanisms, with small variations in reactions and associated rate constants parameters, have also been developed (e.g., Davis et al., 2005). While the flame propagation and stability characteristics of lean premixed systems have been fairly well investigated for conventional hydrocarbon-air systems, not much is known about the characteristics of alternative gaseous fuels. Recently, Lieuwen and coworkers characterized the impact of syngas composition on flashback, blowout, dynamic stability, and autoignition (Zhang et al., 2005; Lieuwen et al., 2006, 2008; Noble et al., 2006). Other key characteristics, such as premixed flame speeds, have also been studied (Natarajan et al., 2005). Giles et al. (2006) conducted a numerical investigation of counterflow syngas nonpremixed flames with dilution by N2, CO2, and H2O. They noted that H2O and CO2 proved more effective than N2 in reducing NOX in H2/CO (50:50) flames. Further improvements to the model are still required, such as incorporation of the inhibitor effect of CO in H2/CO autoignition (Mittal et al., 2006) and more accurate flame speed calculations (Natarajan et al., 2005), especially at high pressures. Hydrogen has a lower autoignition temperature than natural gas and reacts rapidly with oxygen, which leads to a high burning velocity (Qin et al., 2000). Thus, switching from hydrocarbon fuels to syngas requires a detailed understanding of

Fundamental Combustion Characteristics of Syngas

101

flammability range, flame propagation speeds, and safety issues. In this chapter, an overview of syngas combustion characteristics is presented through a numerical study. First, the ignition delays and extinction times for the closed homogeneous reactor (HR) and perfectly stirred reactor (PSR) configurations are presented over a broad range of equivalence ratios and pressures. Simulations of laminar premixed and counterflow diffusion flames are then conducted for various syngas compositions, equivalence ratios, and operating conditions. Finally, the impact of dilution on syngas combustion with varying quantities of CO2, H2O, and N2 is considered. The GRI-Mech 3.0 chemical kinetics mechanism has been used unless mentioned other­ wise. The species proportions in all mixtures considered herein are described in terms of molar ratios.

4.2 Premixed Systems 4.2.1 Ignition and Extinction in Closed Homogeneous and Perfectly Stirred Reactors The combustion characteristics and operability limits of syngas flames differ considerably from those of conventional hydrocarbons. An HR and a PSR represent two different scenarios in terms of mixing characteristics (Turns, 2000). The former is a closed system, in which the mixture composition and temperature are spatially uniform but vary with time. The latter, on the other hand, is an open system, and the mixture composition and temperature in the reactor are different from their counterparts at the inlet. Thus, in the case of a PSR, the fresh reactant mixture at the inlet encounters a growing pool of intermediate reaction species and products in the reactor, and instantly mixes with them. This phenomenon bears significance in determining the ignition properties. The ignition delay for a PSR is defined as the residence time required for the onset of significant heat release without any external ignition source. In contrast, the blow-off/extinction time is specified as the residence time in which an already ignited mixture can be extinguished within a PSR. The ignition delay for an HR is defined as the time delay prior to a steep rise in temperature due to ongoing exothermic chemical reactions. In order to avoid undesirable autoignition, characteristic chemical reaction times should be greater than mixing times. Syngas fuel with equal CO/H2 (50:50 molar concentrations) was chosen as a reference mixture and CHEMKIN (2004) was used for analyses. Table  4.1 lists the ignition time delays for stoichiometric mixtures of CO/H2 (50:50), H2, and CH4 in air at T = 700 K, p = 20 atm as calculated employing the GRI-Mech 3.0 mechanism. For both the perfectly stirred and homogeneous reactors, the shortest ignition delay is exhibited by the CO/H2 mixture. Also included are the PSR ignition delays calculated using the kinetics scheme of Li et al. (2007), which is a comprehensive mechanism for the combustion of CO, CH2O, and CH3OH. The mechanism was developed in a hierarchical manner by substantial validation of submechanisms with experimental measurements involving the combustion of fuels such as H2, CO, and CH2O. The predictions using the Li et al. (2007) mechanism differ slightly from those obtained from the GRI-Mech 3.0 mechanism. The estimated ignition delays for all three cases using both the mechanisms are much higher than

102

Synthesis Gas Combustion: Fundamentals and Applications

Table 4.1 Ignition Delay Timesa at Stoichiometric Conditions (700 K, 20 atm; in air) Reactor Perfectly stirred reactor Homogeneous reactor a b c

CO/H2 (50:50)

H2

13.5 18c 50b

26 19c 104b

b

CH4 b

29b 187b

Unit: second. Based on GRI-Mech 3.0 mechanism. Based on Li et al. (2007) mechanism.

typical mixing times. The ignition delays for methane, H2, and the CO/H2 mixture decrease progressively, in that order. Consequently, the practical use of hydrogen and hydrogen-enriched fuels will require considerable care to ensure safe operations. Figure 4.1a shows the PSR and HR ignition time delays for CO/H2 mixtures as a function of temperature at 1 atm and different equivalence ratios ϕ. The comparison between the estimated and measured ignition delays is reasonable at high temperatures, but the model significantly overpredicts the delay at low temperatures. It should be noted, however, that most of the experiments were carried out at higher pressures. The disagreement at lower temperatures may be attributed in part to the pressure sensitivity of the CO/H2/O2 kinetic mechanism. There is significant scatter in the ignition delay data reported by different researchers, perhaps because the experiments were performed in several types of facilities, including shock-tube apparatus (Petersen et al., 2007), rapid compression facility (Walton et al., 2007), and flow reactors (Peschke and Spadaccini, 1985; Petersen et al., 2007). Moreover, the chosen ratio of H2/CO, the type and quantity of diluents, the equivalence ratio, and the operating pressure are not uniform throughout the various experimental studies. The same reasons may also cause the disagreement between the estimated and measured values at low temperature. The differences in the low-temperature ignition delay have been the subject of several recent studies and interpretations (Chaos and Dryer, 2008; Dryer and Chaos, 2008). The ignition delay decreases with the increase in temperature, albeit in a non­ uniform manner, as seen in Figure 4.1a. The apparent activation energy (slope of the curve) is different for the temperature ranges of 650–1000 K and 1000–2000 K. In the former range, the ignition delay for the HR is about ten times that for the PSR. In the latter range, it is only about twice that for the PSR. The lower value in the case of the PSR can be attributed to the perfect mixing of the incoming fresh reactant mixture with the developing pool of radicals in the reactor. Such mixing enhances the rate of chemical reactions. The steeper slope for the low-temperature range is the result of formation and consumption of relatively stable HO2 and H2O2 species, which precedes the formation of significant amounts of OH, O, and H. The HR ignition delay is highest for the fuel-lean case, followed by the stoichiometric and fuel-rich cases.

103

Fundamental Combustion Characteristics of Syngas

103

τ (s)

101 10–1 HR (φ = 1) HR (φ = 0.27) HR (φ = 3.5) PSR (φ = 1) Exp. (Petersen et al.) Exp. (Peschke et al.) Exp. (Walton et al.) Exp. (Markides et al.)

10–3 10–5 10–7

0.001

0.0005

0.0015

1/T (K–1)

0.002

0.0025

(a)

Extinction and Ignition Time (s)

103

PSR ext (φ = 1) PSR ext (φ = 3.5) PSR ext (φ = 0.27) PSR ext (φ = 1, Li et al.) PSR ignition (φ = 1)

101

10–1

10–3

10–5

10–7

0.001

0.002

1/T (K–1)

0.003

0.004

(b)

Figure 4.1  Ignition time delays for homogeneous and perfectly stirred reactors as a function of temperature for CO/H2 mixtures in air (1 atm).

Figure 4.1b shows that the extinction and ignition times for the PSR are comparable at high temperatures, but the former is much smaller (~106–7 times at 700 K) than the latter at low temperatures. This trend again shows the influence of blending an incoming fuel–air stream with an already ignited and radical species containing

104

Synthesis Gas Combustion: Fundamentals and Applications

mixture in the reactor. Intuition suggests that in order to extinguish an already ignited mixture, the residence time in the reactor must be decreased further. Thus, at low temperature, the critical residence time required to maintain combustion is much shorter than that for autoignition, and this implies that a small ignition source would enhance combustion tremendously. The extinction times overlap for fuel-rich and stoichiometric cases, but are much larger for the fuel-lean case at temperatures lower than ~1000 K. The extinction time predictions from the mechanisms of GRIMech 3.0 and Li et al. (2007) are quite close. The results show that even a small recirculation zone containing radical species can potentially serve as an ignition source. It has been suggested by Markides and Mastorakos (2005) that autoignition is not simply kinetically controlled, but rather, the location and onset of autoignition may be determined by mixing due to turbulent fluctuations. In addition, many other factors, such as compressible flow, physical mixing, operating conditions, and catalytic surface processes occurring in the mild ignition regimes, could be the sources of the departure of experimentally observed ignition delays from homogeneous gas-phase kinetic predictions. As enumerated in Table 4.1, the calculated ignition delays at high pressure (20 atm) are on the order of tens of seconds, closer to the measured values than the calculations at 1 atm.

4.2.2 Premixed Flames Laminar flame speed is one of the most important parameters characterizing premixed combustion. It dictates the flammability domain for a chosen fuel composition, equivalence ratio, and operating condition. Furthermore, it encompasses the information on the reaction mechanisms in the presence of diffusive transport. In the following, combustion characteristics—including flame speeds, flame structures, and flammability limits—are investigated for different syngas compositions. The effects of equivalence ratios and mixture dilution with CO2, N2, H2O, and CH4 are studied. In addition, the influences of temperature and pressure variations are considered. The calculations were performed using the one-dimensional PREMIX code from the CHEMKIN II package (Kee et al., 1986, 1992). 4.2.2.1 CO/H2 Mixtures Figure 4.2 shows the laminar flame speed (SL) as a function of equivalence ratio (ϕ) for five different CO/H2 (5:95, 25:75, 50:50, 75:25, and 95:5) compositions. The pressure is set to 1 bar and the initial temperature is 300 K. At a given equivalence ratio, the flame speed increases with the increase in the concentration of H2. The premixed flame speed is proportional to chemical reaction rates and molecular diffusivity, both of which are enhanced when the concentration of H2 is increased. The fuel-lean and fuel-rich flammability limits are approximately ϕ = 0.35 and ϕ = 5 for all mixture compositions. The flame speed increases with ϕ to reach a maximum value at a fuel-rich condition, and thereafter decreases. Table 4.2 lists the maximum values of SL and the corresponding equivalence ratios denoted by φ max SL . The flame speed increases with H2 content, but the associated φ max SL decreases.

105

Fundamental Combustion Characteristics of Syngas

Flame Speed (cm/s)

400

05% CO + 95% H2 25% CO + 75% H2 50% CO + 50% H2 75% CO + 25% H2 95% CO + 05% H2

300

200

100

1

0

2 3 Equivalence Ratio

4

5

Figure 4.2  Flame speeds as a function of equivalence ratio for various H2/CO mixtures, p = 1 atm and Tin = 300 K.

Table 4.2 Maximum Flame Speeds and Corresponding Equivalence Ratios for H2/CO Mixtures in Air H2:CO SL (cm/s) ϕmax SL

05:95

25:75

50:50

75:25

95:05

100:0

73 2.7

144 2.35

206 2.08

264 1.85

307 1.69

318 1.65

Figure  4.3 shows the flame structure of premixed combustion in terms of spatial variations of major and minor species and temperature for different equivalence ratios (ϕ = 0.6, 1.0, and 2.08) of a CO/H2 (50:50) mixture in air. The value of φ max SL = 2.08 corresponds to the maximum flame speed. At the fuel-lean condition, CO and H2 are entirely consumed to produce CO2 and H2O. In the fuel-rich case, however, a large quantity of unreacted CO remains after combustion. On the burned side, the CO concentration even increases slightly because of CO2 dissociation. Similarly, the nonnegligible mass fraction of H2 promotes the presence of intermediate species such as H. The increase in equivalence ratio up to 2.08 leads to a decrease in the characteristic flame thickness (δf). The trend is clear when H2O2 concentrations over different equivalence ratios are compared. The premixed flame thickness is related to the thermal diffusivity (aT) and the flame speed as δf ~ aT /SL . The observed trend is expected since the flame speed increases with the equivalence ratio up to φ max SL. The maximum flame temperature, however, is attained near the stoichiometric condition

106

Synthesis Gas Combustion: Fundamentals and Applications

CO

0.00 –0.2

Mass Fractions

H 2O

H2 0.0

0.2

O2

0.15

CO

Temperature

2.0

CO2

1.5

H 2O

0.10 0.05 0.00 –0.2

1.0

H2

0.5 0.0

0.2

2.0

CO

0.20 0.15

Temperature O2

CO2

0.10 0.05 0.00 –0.2

H2O H2

1.5 1.0 0.5

0.0 0.2 Position (cm)

0.4

4 2

8

0.1

0.2

0.4

0.3

φ = 1.0

6

H O OH HO2 × 10 H2O2 × 10

4 2 0 0.0

0.4

φ = 2.08

Mass Fractions

0.25

2.5

H O OH HO2 × 10 H2O2 × 10

6

0 0.0

0.4

φ = 1.0

0.20

0.30

0.5

Mass Fractions (10–3)

0.05

1.0

Mass Fractions (10–3)

0.10

0.25

1.5

Mass Fractions (10–3)

CO2

0.15

φ = 0.6 Temperature (1000 K)

O2

0.20

2.0

Temperature

Temperature (1000 K)

φ = 0.6

Temperature (1000 K)

Mass Fractions

0.25

8

0.1

0.2

0.3

0.4

φ = 2.08 H O OH HO2 × 10

6

H2O2 × 10

4 2 0 0.0

0.1

0.2 0.3 Position (cm)

0.4

Figure 4.3  Spatial distributions of major and minor species for a 50:50 H2/CO mixture at ϕ = 0.6, ϕ = 1.0, and ϕ = 2.08.

(ϕ = 1). Figure 4.4 shows the adiabatic flame temperatures and flame speeds for three different H2/CO compositions as a function of equivalence ratio. The maximum flame temperature always occurs at a fuel-rich condition close to ϕ = 1, but not near the equivalence ratio corresponding to the peak value of SL . The maximum adiabatic flame temperature occurs near the stoichiometric value, where all the available ­reactants are transformed entirely into products causing maximum heat release. For the combustion of conventional hydrocarbons fuels (methane, propane, etc.), however, the flame speed and adiabatic temperature profiles are correlated more closely when plotted as a function of equivalence ratio. The maximum flame speed for hydrocarbon fuels occurs around the stoichiometric condition, whereas for pure hydrogen it is around 1.7.

107

Fundamental Combustion Characteristics of Syngas

250

2200

200

2000 150 1800 100

1600

Flame Speed (cm/s)

2400 Adiabatic Temperature (K)

300

50:50 H2:CO 25:75 H2:CO 75:25 H2:CO

50

1400 1

2 3 Equivalence Ratio

4

5

0

Figure 4.4  Adiabatic flame temperature (lines) and flame speed (symbols and lines) as a function of equivalence ratio for 25:75, 50:50, and 75:25 H2/CO mixtures.

4.2.2.2 CO2, H2O, and N2 Dilution Since one or more of the species, CO2, H2O, N2, or CH4, are likely to be present in a practical syngas composition, the impact of these diluents on the flame structure and speed has to be carefully studied. The diluents can affect the combustion characteristics by influencing adiabatic flame temperature, specific heat of mixtures, chemical kinetics rates, and radiative heat transfer. Addition of a diluent to the fuel will reduce the flame temperature, and hence the laminar flame speed by reducing the reaction rate. Figure  4.5a,b,c shows the flame speeds of three CO/H2 compositions at different equivalence ratios as a function of percentage of dilution with CO2, H2O, or N2, where, for example, 20% of CO2 dilution implies a 40:40:20 H2/CO/CO2 mixture. The slopes of the curves show that the impact of diluents becomes most significant in the fuel-rich case (ϕ = 2.08). For a given dilution percentage and equivalence ratio, the flame speed is highest with N2, followed by H2O and then CO2. A similar trend is seen in Figure  4.6a, with respect to the maximum flame temperature attained. The diluents, H2O and CO2, may undergo decomposition and then be regenerated through the flame. Energy is spent in bond breaking and liberated in recombination, and as a consequence, the flame speeds and adiabatic temperatures are lower than their counterparts for dilution with N2. The addition of H2O and CO2, which also happen to be the final combustion products, may slightly enhance the backward reaction rates of a few exothermic elementary reactions. The increase in cp (triatomic H2O and CO2 vs. diatomic N2) and smaller diffusivities of H2O and CO2 (smaller than that of H2O), however, are likely to have a greater effect on the reduction of flame speed than their decomposition.

108

Synthesis Gas Combustion: Fundamentals and Applications 250 φ = 0.6 φ=1 φ = 2.08

Flame Speed (cm/s)

200

CO2 dilution

150

H2O dilution N2 dilution

100

50

0

20

40 Dilution (%) (a)

60

80

300 φ = 0.6 φ=1 φ = 2.08

Flame Speed (cm/s)

250 200

CO2 dilution

150

N2 dilution

H2O dilution

100 50

0

20

40

60

80

Dilution (%) (b)

Figure  4.5  Flame speed response to variable CO2, H2O, and N2 dilution for (a) 50:50, (b) 75:25, and (c) 25:75 H2/CO mixtures at three equivalence ratios.

Figure 4.6a shows that the CO2 mass fraction in the burned gases of the 50:50 H2/CO fuel mixture remains nearly constant for the considered range of dilution with H2O or N2. The same is true for the H2O mass fraction in burned gases as seen in Figure 4.6b, when the diluent is either CO2 or N2. The CO mass fraction in burned gases, however, reveals some chemical activity at 30% and 40% dilutions with CO2

109

Fundamental Combustion Characteristics of Syngas 150 φ = 0.6 φ=1 Flame Speed (cm/s)

φ = 2.08 100

CO2 dilution H2O dilution N2 dilution

50

0

20

40 Dilution (%) (c)

60

80

Figure 4.5  (Continued.)

and N2, respectively. No such activity for CO is seen for dilution with H2O. In addition to 50:50 H2/CO at ϕ = 1, several other simulations (not shown here) were conducted for 25:75 and 75:25 cases at different equivalence ratios. In these cases, the flame response to the variable dilution exhibits a similar behavior. In order to separate the thermal and chemical effects of dilution, the flame speeds of various H2/CO fuel mixtures are replotted as a function of the adiabatic flame temperature in Figure 4.7a,b,c. Three different equivalent ratios are considered over the dilution range of 0 to 70%. The highest equivalent ratio corresponds to the maximum flame speed in each case, i.e., 2.08 for 50:50, 1.85 for 75:25, and 2.35 for 25:75 H2/CO fuel mixtures, respectively. The strongest impact on the flame speed results from the thermal effect of dilution; however, the chemical effect is also evident. The chemical activity of H2O and CO2 is observed to change with the equivalence ratio. For the lean mixture, both H2O and CO2 dilution equally decreases the flame speed for an equivalent flame temperature. The situation for the fuel-rich case becomes considerably different. The chemical effect of H2O dilution is more prevalent, causing a steeper decrease in the flame speed at flame temperatures below 1800 K. For the stoichiometric mixture, CO2 dilution has the largest impact, particularly for temperatures greater than 1600 K. Methane was added to the 50:50 H2/CO composition, in combination with varying dilutions of H2O/CO2/N2. Table  4.3 summarizes the flame characteristics of 50:50 H2/CO stoichiometric mixtures in air with different levels of dilution in fuel. The characteristics tabulated include flame speed, adiabatic flame temperature, and CO2 mass fraction in burned gases. Introduction of 5% CH4 alone to a nondiluted H2/CO mixture does not change the flame speed or adiabatic temperature,

110

Synthesis Gas Combustion: Fundamentals and Applications

1.00

N2 dil. H2O dil. CO2 dil.

0.80

2500

Diluents Temperature COb2

2400

0.60

2200 2100

0.40

Temperature (K)

Mass Fractions

2300

2000 0.20

0.00

1900

0

10

20

30

40

1800

Dilution (%) (a) 0.3

H2Ob

N2 dil.

COu COb

H2O dil.

Mass Fractions

CO2 dil.

0.2

0.1

0.0

0

10

20 30 Dilution (%)

40

50

(b)

Figure 4.6  Effect of dilution (N2, H2O, or CO2) on adiabatic flame temperature and maximum mass fractions of major species for a 50:50 H2/CO mixture at ϕ = 1.0. Subscripts u and b indicate unburned and burned.

111

Fundamental Combustion Characteristics of Syngas 50:50 H2/CO Fuel Mixture CO2 H2O N2 φ = 1.0

sin

100

gd

ilu

tio

n

150

φ = 2.08

In cr ea

Flame Speed (cm/s)

200

50

I 0

1500

rea nc

sin

on

uti

il gd

2000

Flame Temperature (K)

Flame Speed (cm/s)

40

30

φ = 0.6 CO2 H2O

N2

20

10

0 1200

a cre In

sin

1400 1600 Flame Temperature (K) (a)

on

uti

il gd

1800

Figure 4.7  Effect of dilution (N2, H2O, or CO2) on adiabatic flame temperature and flame speed for (a) 50:50, (b) 75:25, and (c) 25:75 H2/CO mixtures at three equivalence ratios.

but slightly decreases the CO2 mass fraction. As compared to the results with dilution of 2.5:1.25:1.25 H2O/CO2/N2, the CH4/H2/CO flame has higher flame speed and adiabatic temperature, but a lower concentration of CO2. On increasing the overall dilution from 10 to 45%, but keeping the amount of CH4 constant at 5%, the adiabatic temperature is not significantly reduced. The flame speed, however, is noticeably lower than that for the corresponding CH4/H2/CO flame. The corresponding CO2 mass fraction in the burned gases, on the other hand, increases with dilution. The

112

Synthesis Gas Combustion: Fundamentals and Applications 75:25 H2/CO Fuel Mixture φ = 1.85 CO2 H2O N2

200

n

φ = 1.0

sin

gd

ilu

tio

150 cr ea

100

In

Flame Speed (cm/s)

250

50

a cre In

0 1000

sin

1500 2000 Flame Temperature (K)

60

φ = 0.6

CO2 H2O N2

50 Flame Speed (cm/s)

on

uti

il gd

40 30 20

ea cr In

10 0 1200

n

tio

1400

g

sin

lu di

1600 1800 Flame Temperature (K) (b)

2000

Figure 4.7  (Continued.)

flame speed remains nearly constant for the two cases of 45% dilution, with and without CH4. For any given level of dilution, the CO2 mass fraction in burned gases is smaller when CH4 is present. 4.2.2.3 Pressure and Temperature Effects Flame speed also varies with the inlet preheated temperature and pressure. Figure 4.8a shows the variation of flame speed with pressure, for a syngas mixture

113

Fundamental Combustion Characteristics of Syngas 25:75 H2/CO Fuel Mixture φ = 2.35 CO2 H2O N2

tio n

100

sin

gd

ilu

φ = 1.0

In cr ea

Flame Speed (cm/s)

150

50

Inc

0 1000

s rea

n

tio

ilu

d ing

1500 2000 Flame Temperature (K)

35

Flame Speed (cm/s)

φ = 0.6

CO2 H2O N2

30 25 20 15

g

10

I

r nc

sin ea

n

tio

lu di

5 0 1200

1400

1600 1800 Flame Temperature (K) (c)

2000

Figure 4.7  (Continued.)

of 40:40:20 H 2/CO/H 2O at different inlet temperatures and equivalence ratios. The flame speed reduces monotonically with the increase in pressure. A similar trend is seen for the flame thickness, as depicted by species concentrations in Figure  4.8b. Changes in pressure and temperature may alter the pathways and kinetic rates of the ongoing chemical reactions, leading to a change in the flame speed and the spatial distribution of mass fractions of species. In general, an increase in pressure raises the overall reaction rate, and consequently should tend

114

Synthesis Gas Combustion: Fundamentals and Applications

Table 4.3 Flame Characteristics of H2/CO (50:50) Stoichiometric Mixtures in Air with Various Diluents Degree of Dilution

Diluent Composition CH4:H2O:CO2:N2

SL (cm/s)

Tad (K)

YCO2

  0%   5%

0:0:0:0 5:0:0:0 0:2.5:1:25:1.25 5:2.5:1.25:1.25 0:5:2.5:2.5 5:7.5:3.75:3.75 0:10:5:5 5:20:10:10 0:22.5:11.25:11.25

120 121 114.3 101.7 108.1   89.2   94.4   53.7   53.85

2,362 2,371 2,334 2,322 2,311 2,271 2,250 2,090 2,020

0.2277 0.2101 0.2332 0.2184 0.2389 0.2272 0.2511 0.2519 0.2862

10% 20% 45%

to increase the flame speed. But the elevated density necessitates more thermal energy transfer from the reaction zone to raise the reactant temperature in the preheat zone. Since thermal diffusivity is inversely proportional to pressure, the net effect of increase in pressure leads to a reduction in flame speed. The same phenomena are observed for other compositions of species (H 2:CO:diluents), but are not shown here. Figure 4.8c shows the impact of pressure on adiabatic flame temperature for four different inlet conditions. In the fuel-lean cases, no influence of p is revealed. For stoichiometric flames, however, a difference of 4% (~100 K) in the adiabatic flame temperature for the two inlet temperatures indicates the importance of product dissociation at the higher flame temperature of the stoichiometric mixture at low pressure.

4.3 Counterflow Diffusion Flames Figure 4.9 shows the schematic of the counterflow configuration for studying nonpremixed flames. The physical model considered here is an axisymmetric laminar diffusion flame, stabilized somewhere near the stagnation plane of opposing syngas and air streams. This type of configuration is frequently used to investigate nonpremixed combustion, as it is approximately one-dimensional and because the residence times within the flame zone can be varied with ease. The flow is characterized by a strain rate, εs, which represents a characteristic velocity gradient for the flow. An increase in εs leads to a decrease in the flow residence time in the flame zone. A critical strain rate, εsext, represents the limit when the flame is quenched. This key parameter (εsext ) is tracked in simulations of nonpremixed syngas flames using the detailed modeling of counterflow flame (DMCF) code (Darabiha et al., 1988; Darabiha 1992). The code incorporates detailed chemical kinetic mechanisms and multispecies transport through the CHEMKIN package (Kee et al., 1986).

115

Fundamental Combustion Characteristics of Syngas

500 φ = 0.5, T = 300 K φ = 0.5, T = 600 K φ = 0.5, T = 700 K φ = 1.0, T = 300 K φ = 1.0, T = 600 K φ = 1.0, T = 700 K

Flame Speed (cm/s)

400

300

200

100

0

10

20

30 Pressure (bar)

40

50

1E–06

0.20

8E–07

HCO, 50 bar CO, 50 bar CO2, 50 bar HCO, 10 bar CO, 10 bar CO2, 10 bar

0.15

0.10

0.05

0.00 0.05

6E–07

4E–07

HCO Mass Fraction

CO and CO2 Mass Fractions

(a) 0.25

2E–07

0.06

Position (cm)

0.07

0 0.08

(b)

Figure 4.8  Effect of pressure on (a) flame speed at different inlet temperatures and equivalence ratios, (b) species profiles through the flame front, and (c) adiabatic temperature at inlet temperatures and equivalence ratios (40:40:20 H2/CO/H2O mixture).

116

Synthesis Gas Combustion: Fundamentals and Applications

Normalized Adiabatic Temperature

1

0.99

0.98

0.97

Tad, φ = 1, T = 300 K Tad, φ = 1, T = 600 K

Tad, φ = 0.5, T = 300 K

0.96

0.95

Tad, φ = 0.5, T = 600 K

10

40

20 30 Pressure (bar) (c)

50

Figure 4.8  (Continued.)

Flame Stagnation plane

Syngas

Air y x

Z = Zst

Figure 4.9  Schematic view of a counterflow diffusion flame.

Fundamental Combustion Characteristics of Syngas

117

4.3.1 H2/CO Mixtures Figure 4.10 shows the variation of the maximum temperature with the strain rate. The pressure is set to 1 bar and the incoming flow temperatures of the opposing streams are 300 K. Five different H2/CO compositions (5:95, 25:75, 50:50, 75:25, and 95:5) were treated and simulations were performed until the extinction limit of the strain rate (εsext ) was reached. As the strain rate increases, the maximum temperature decreases monotonically and eventually reaches a point where the flame is quenched. The temperature at εsext is referred to as the quenching temperature. For very low strain rates, Tmax is in the range of 2200 to 2350 K. For all H2/CO compositions, the quenching temperatures are found to be between 1300 and 1350 K. The higher value of εsext with increasing H2 concentration implies a stronger resistance to flame quenching. The value progressively increases from 6000 s–1 for 5% H2 to about 27,500 s–1 for 95% H2. The heat-release rate per unit flame area ( q s ) is defined by

q s =

 Ns   h W ω  dy k k k  −∞   k =1 

∫ ∑ +∞

(4.1)

where hk, Wk, and ω k represent the mass enthalpy, molecular weight, and reaction rate of species k, respectively. Figure 4.10b shows the heat-release rate as a function of the strain rate for five different H2/CO compositions. For a nonquenched flame, the heat-release rate decreases slightly with the increase in the percentage of H2.

4.3.2 CO2, H2O, and N2 Dilution Figure 4.11a shows the maximum temperature as a function of strain rate for different dilution concentrations of H2O. The value of Tmax decreases with the increase in dilution, but the corresponding εsext is slightly reduced. The dilution does not affect the heat-release rate per unit flame area (Figure 4.11b.) Figure 4.12 shows the variation of maximum temperature and heat-release rate with the strain rate, for 20% dilution with each of N2, H2O, and CO2. The curves with different diluting species overlap completely, implying that any diluent would result in the same level of maximum temperature and heat-release rate. The corresponding flame structures, however, are dissimilar. Figure 4.13a and b exhibits the flame structure in terms of major species at strain rates of 1000 s–1 and 10,000 s–1, respectively. All three 20% dilution cases are shown. The main products of H2/CO/diluent air combustion are H2O and CO2. For dilution with N2 or CO2, the mass fraction of H2O starts from zero on both sides of the counterflow burner and attains a peak value of ~0.22 in the flame zone. A similar trend is observed for CO2 concentration with diluent N2 or H2O, except it has a much lower peak (~0.025) in the flame zone. For dilution with H2O, the peak value of H2O mass fraction is ~0.24. The increase in the strain rate (Figure 4.13b) lowers the peak in the H2O concentration to ~0.2 for dilution with CO2 or N2. Figure 4.13a shows that on approaching the flame from the syngas (right) side, the CO2 mass fraction decreases to a local minimum followed by a feeble rise to a local

118

Synthesis Gas Combustion: Fundamentals and Applications 2400 05% H2, 95% CO

Maximum Temperature (K)

2200

25% H2, 75% CO 50% H2, 50% CO

2000

75% H2, 25% CO 95% H2, 05% CO

1800 1600 1400 1200

0

10000 20000 Strain Rate (1/s) (a)

3.50

Heat-Release Rate (102 W. cm–2)

3.00 2.50 2.00 1.50

05% H2, 95% CO

1.00

25% H2, 75% CO 50% H2, 50% CO

0.50

75% H2, 25% CO 95% H2, 05% CO

0.00

0

10000 Strain Rate (1/s) (b)

20000

Figure 4.10  Effect of strain rate on (a) maximum temperature and (b) heat-release rate for five different H2/CO mixtures.

maximum, before it decreases to zero on the oxidizer side. Increasing the strain rate to 10,000 s–1 erases the local extrema (Figure 4.13b). Figure 4.14a and b exhibits the flame structure in terms of minor species at strain rates of 1000 s–1 and 10,000 s–1, respectively, for syngas diluted with 20% CO2. The two sets of peaks on either side of the diffusion flame indicate two distinct zones of chemical activity, which somehow lead

119

Fundamental Combustion Characteristics of Syngas 2400

Maximum Temperature (K)

2200 2000 1800 1600 1400

1200

50%

102

103 Strain Rate (1/s)

0%

104

105

Heat-Release Rate (102 W. cm–2)

(a) 5 4 3

0%

2

50%

1

101

102

103 Strain Rate (1/s) (b)

104

105

Figure 4.11  Effect of strain rate on (a) maximum temperature and (b) heat-release rate for different levels of H2O dilution.

to a unique peak in the temperature profile (the temperature profile is not shown). On the oxidizer side, O2 and H2 first react to form HO2 and H2O2, and subsequently form the transient species of OH, H, and O. On the syngas side, only hydrocarbon and oxygenated hydrocarbon species are present, namely, CH2O, HCO, and CH4. The entire diffusion flame structure thus exhibits two distinct chemical zones. The phenomenon makes it difficult to reduce the detailed kinetic scheme to simpler semi-global mechanisms. Figure 4.14b shows that the increase in the strain rate considerably reduces the

120 2400

3.0

2200

2.5

2000

2.0

1800

1.5

1600

1.0

1400

0.5

1200

0

5000

10000 Strain Rate (1/s)

15000

Heat-Release Rate (102 W. cm–2)

Maximum Temperature (K)

Synthesis Gas Combustion: Fundamentals and Applications

0.0 20000

Figure 4.12  Effect of strain rate on maximum temperature and heat-release rate for 20% dilution with N2 (–), H2O (), and CO2 (▲) dilution.

OH concentration and promotes the formation of HO2. Switching the diluent from H2O to CO2 or N2 changes the flame structure slightly (not shown here).

4.3.3 Pressure Effects Figure 4.15a shows the heat-release rate q s as a function of the strain rate for different H2/CO compositions at three different pressures (1, 15, and 30 bar). The inlet temperature is fixed at 300 K. Results are compared with those for a CH4-air counterflow diffusion flame at the same operating conditions. At a given pressure, the heat-release rate increases with the strain rate, with the slopes of all the curves being ~0.5. The heat-release rate for the CH4-air flame is equivalent to that for the 50:50 H2/CO flame. However, εsext is much shorter for the CH4-air flames. Thus, switching from methane to syngas in a device that encounters diffusion flames may lead to a more stable and resistant flame with a slightly higher flame temperature (less than 10% at a given pressure). Based on Figure 4.15a, the heat-release rate per unit flame area may be described by a simple expression,

( )

q s = αεs

1/ 2



(4.2)

where the coefficient α depends on pressure and syngas composition. Figure 4.15b shows that α varies linearly with pressure. The H2/CO composition is 50:50 and the stain rate is 20 s–1. Figure  4.15c shows the effect of three different strain rates, εs = 20, 900, and 15,000 s–1, on the heat-release rate in a pressure range of 1 to 40 bar. The inlet temperature is fixed at 300 K. The heat-release rate is plotted as a function of pεs, the

121

Fundamental Combustion Characteristics of Syngas 0.4

CO2 H2O

CO

H2O dil.

0.3 Mass Fractions

εs = 1000 s–1

CO2 dil. N2 dil. O2

0.2

0.1

0.0

–2

0.4

CO2 H2O

0 Position (cm) (a)

1

εs = 10000 s–1

2

CO

H2O dil.

0.3 Mass Fractions

–1

CO2 dil. N2 dil. O2

0.2

0.1

0

–1

–0.5

0 Position (cm)

0.5

1

(b)

Figure 4.13  Spatial distributions of major species (O2, CO, CO2, and H2O) for a 20% dilution with N2, H2O (■), and CO2 () at two strain rates: (a) 1000 s–1 and (b) 10,000 s–1.

product of pressure and strain rate. A functional relationship of q s ~ pεs is exhibited, except near point A (pεs ~ 15,000 bar/s), which represents the flame extinction limit close to 1 bar. The correlation is similar to the situation observed in a H2-air or H2-O2 counterflow diffusion flame (Ribert et al., 2007). In the case of 20% dilution with H2O, CO2, or N2 species, the same correlation remains valid.

122

Synthesis Gas Combustion: Fundamentals and Applications

12

H × 103 O × 103 OH × 103 HO2 × 106 H2O2 × 106 CH4 × 106 CH2O × 106 HCO × 106

εs = 1000 s–1

Mass Fractions

10 8 6 4 2 0

–2

–1

0

1

2

Position (cm) (a) 30

H ×103 O × 103 OH × 103 HO2 × 106 H2O2 × 106 CH4 × 106 CH2O × 106 HCO × 106

εs = 10000 s–1

Mass Fractions

25 20 15 10 5 0

–1

–0.5

0 Position (cm) (b)

0.5

1

Figure 4.14  Spatial distributions of minor species for 20% dilution with CO2 for two different strain rates.

123

Fundamental Combustion Characteristics of Syngas

Heat-Release Rate (102 W. cm–2)

30 bar 15 bar 101

1 bar 100

10–1 1 10

25% H2, 75% CO 50% H2, 50% CO 75% H2, 25% CO 100% H2 100% CH4 102

103

104

105

106

40

50

Strain Rate (1/s) (a)

εs = 20 s–1

α Parameter

400

300

200

100

0

0

10

20

30

Pressure (bar) (b)

Figure  4.15  (a) Variations of heat-release rate with strain rate at different pressures. (b) Effect of pressure on α parameter at a strain rate of 20 s–1. (c) Heat-release rate as a function of product of pressure and strain rate.

124

Synthesis Gas Combustion: Fundamentals and Applications

Heat-Release Rate (102 W. cm–2)

102 εs = 20 s–1 εs = 900 s–1 εs = 15000 s–1 101

100

10–1 101

102

103

p . εs

104

105

106

(c)

Figure 4.15  (Continued.)

4.4 Strained Counterflow Premixed Flames A twin laminar counterflow premixed flame was also simulated, using the DMCF code. The same composition of premixed syngas was used on both sides. The pressure was set to 1 bar and the inlet temperature was 300 K. Figure  4.16a and b shows the variation of Tmax and q s with the strain rate for different compositions. As the H2 quantity in the mixture decreases, the strained premixed flames show less resistance to quenching. This may be due partly to the fact that the level of dilution increases as the quantity of H2 is reduced. The same trend is also observed for q s. As compared to syngas diffusion flames (Figure 4.10a), strained premixed flames exhibit lower values of εsext and hence are easier to quench. For the former case, q s increases with the strain rate (Figure  4.10b). In contrast, for the latter case, q s decreases with the increase in strain rate. These observations have also been made for classical fuels such as methane (Im et al., 1996), and syngas combustion follows the same trend.

4.5 Conclusions The characteristics of syngas combustion have been investigated. Laminar flames in premixed and counterflow diffusion configurations, as well as the ignition characteristics in homogeneous and perfectly stirred reactors (PSRs), were studied in detail by treating various syngas compositions at different operating conditions. The ignition time delay is found to be strongly dependent on the pressure and the inlet/preheated temperature. The lower value of ignition delay for the PSR than for the homogenous reactor can be attributed to the perfect mixing of incoming reactant mixture with a

125

Fundamental Combustion Characteristics of Syngas

2400

Maximum Temperature (K)

2200 2000 1800 1600 1400

1200

50% H2 + 50% CO 40% H2 + 40% CO + 20% H2O 30% H2 + 30% CO + 40% H2O 30% H2 + 30% CO + 20% H2O + 20% CO2 30% H2 + 30% CO + 20% H2O + 20% N2 102

103 Strain Rate (1/s) (a)

104

Heat-Release Rate (102 W . cm–2)

3

2

1 50% H2 + 50% CO 40% H2 + 40% CO + 20% H2O 30% H2 + 30% CO + 40% H2O 30% H2 + 30% CO + 20% H2O + 20% CO2 30% H2 + 30% CO + 20% H2O + 20% N2 102

103 Strain Rate (1/s)

104

(b)

Figure 4.16  Effect of strain rate on (a) maximum temperature and (b) heat-release rate for a premixed syngas flame in a counterflow configuration.

developing pool of radicals in the former case. At low temperature, the PSR extinction time is much lower than its ignition delay. Consequently, the critical residence time to maintain combustion is much shorter (~106–7 times at 700 K) than that for autoignition at low temperature, which indicates that even a tiny ignition source would enhance combustion tremendously. The estimated ignition delays at low temperatures, however, are not in good agreement with experimental data. Many factors,

126

Synthesis Gas Combustion: Fundamentals and Applications

such as physical mixing, different operating conditions, and catalytic surfaces, could be responsible for the departure of predictions from experimental results. The laminar flame speed for premixed syngas systems increases with H2 concentration. The flame speed exhibits a strong dependence on pressure, inlet/preheated temperature, and equivalence ratio. The maximum flame speed occurs at a fuel-rich condition, whereas the maximum flame temperature occurs near the stoichiometric value. The dilution of syngas with N2, CO2, or H2O lowers the flame speed. For a given dilution percentage and equivalence ratio, the flame speed and temperature are highest with N2, followed by H2O and then CO2. For counterflow diffusion flames, extinction limits of strain rate εsext are reported for various syngas compositions. The extinction limit increases with H2 concentration. The dilution of syngas does not affect the extinction limit, but it does alter the diffusion flame structure. Two zones of chemical activity are observed for the diffusion flame, one on the syngas side and the other on the air side. The heat-release rate per unit flame area correlates well with the product of the square root of pressure and strain rate. Compared to diffusion flames, premixed counterflow flames exhibit lower values of εsext and hence are easier to quench.

Acknowledgments This work was sponsored partly by the National Energy Technology Laboratory, U.S. Department of Energy, Grant DE-FG26-07NT43069, and partly by the Pennsylvania State University. The technical drawing service provided by Xiaodong Chen is gratefully acknowledged.

References Brabbs T. A., Belles F. E., and Brokaw R. S. (1970). Shock-tube measurements of specific reaction rates in branched-chain H2-CO-O2 systems. Proc. Combust. Inst. 13:129. Chaos M. and Dryer F. L. (2008). Syngas combustion kinetics and applications. Combust. Sci. Tech. 180:1053. CHEMKIN 4.0. (2004). San Diego: Reaction Design. Darabiha N. (1992). Transient behavior of laminar counterflow hydrogen air diffusion flames with complex chemistry. Combust. Sci. Tech. 86:163. Darabiha N., Giovangigli V., Candel S., and Smooke M. (1988). Extinction of strained premixed propane-air flames with complex chemistry. Combust. Sci. Tech. 60:267. Davis S. G., Joshi A. V., Wang F., and Egolfopoulos F. (2005). An optimized kinetic model of H2/CO combustion. Proc. Combust. Inst. 30:1283. Dean A. M., Steiner D. C., and Wang E. E. (1978). A shock tube study of the H2/O2/CO/Ar and H2/N2O/CO/Ar systems: Measurement of the rate constant for H+N2O = N2+OH*. Combust. Flame 32:73. Dryer F. L. and Chaos M. (2008). Ignition of syngas/air and hydrogen/air mixtures at low temperatures and high pressures: Experimental data interpretation and kinetic modeling implications. Combust. Flame 152:293. Gardiner W. C., Mallard W. G., McFarland M., Morinaga K., Owen J. H., Rawlins W. T., Takeyama T., and Walker B. F. (1972). Elementary reaction rates form post-inductionperiod profiles in shock-initiated combustion. Proc. Combust. Inst. 14:61.

Fundamental Combustion Characteristics of Syngas

127

Giles D. E., Som S., and Aggarwal S. K. (2006). NOX emission characteristics of counterflow syngas diffusion flames with airstream dilution. Fuel 85:1729. GRI-MECH. http://www.me.berkeley.edu/gri_mech/. Im H. G., Bechtold J. K., and Law C. K. (1996). Response of counterflow premixed flames to oscillating strain rates. Combust. Flame 105:358. Kee R., Dixon-Lewis G., Warnatz J., Coltrin M., and Miller J. (1986). A Fortran computer code package for the evaluation of gas phase multicomponent transport properties. Technical Report SAND86-8286 UC401, Sandia National Laboratories. Kee R. G., Grcar J. F., Smooke M. D., and Miller J. A. (1992). A Fortran program for modeling steady laminar one-dimensional premixed flames. Technical report, Sandia National Laboratories. Kim, T. J., Yetter R. A., and Dryer F. L. (1994). New results on moist CO oxidation: High pressure, high temperature experiments and comprehensive kinetic modeling. Proc. Combust. Inst. 25:759. Li J., Zhao Z. W., Kazakov A., Chaos M., Dryer F. L., and Scire J. J. (2007). A comprehensive kinetic mechanism for CO, CH2O, and CH3OH combustion. Int. J. Chem. Kinet. 39:109. Lieuwen T., McDonell V., Petersen E., and Santavicca D. (2006). Fuel flexibility influences on premixed combustor blowout, flashback, autoignition and stability. Paper GT2006-90770 presented at ASME Turbo Expo, Barcelona, Spain. Lieuwen T., McDonell V., Santavicca D., and Sattelmayer T. (2008). Burner development and operability issues associated with steady flowing syngas fired combustors. Combust. Sci. Tech. 180:1169. Markides C. N. and Mastorakos E. (2005). An experimental study of hydrogen autoignition in a turbulent co-flow of heated air. Proc. Combust. Inst. 30:881. McLean I. C., Smith D. B., and Taylor S. C. (1994). The use of carbon monoxide/hydrogen burning velocities to examine the rate of the CO+OH reaction. Proc. Combust. Inst. 25:749. Mittal G., Sung C.-J., and Yetter R. A. (2006). Autoignition of H2/CO at elevated pressures in a rapid compression machine. Int. J. Chem. Kinet. 38:516. Natarajan J., Nandula T., Lieuwen T., and Seitzman J. (2005). Laminar flame speeds of synthetic gas fuel mixtures. Paper GT2005-68917 presented at the ASME Turbo Expo, Reno, Nevada. Noble D. R., Zhang Q., Shareef A., Tootle J., Meyers A., and Lieuwen T. (2006). Syngas mixture composition effects upon flashback and blowout. Paper GT2006-90470 presented at the ASME Turbo Expo, Barcelona, Spain. Peschke W. T. and Spadaccini L. J. (1985). Determination of autoignition and flame speed characteristics of coal gases having medium heating values. Report EPRI AP-4291, Electric Power Research Institute. Petersen E. L., Kalitan D. M., Barrett A. B., Reehal S. C., Mertens J. D., Beerer D. J., Hack R. L., and McDonell V. G. (2007). New syngas/air ignition data at elevated pressure and comparison to current kinetics models. Combust. Flame 149:244. Qin X., Kobayashi H., and Niioka T. (2000). Laminar burning velocity of hydrogen-air premixed flames at elevated pressure. Exp. Thermal Fluid Sci. 21:58. Ribert G., Zong N., Yang V., Pons L., Darabiha N., and Candel S. (2008). Counterflow diffusion flames of general fluids: Oxygen/hydrogen mixtures. Combust. Flame 154(3):319–330. Rightley M. L. and Williams F. A. (1995). Analytical approximations for structures of wet CO flames with one-step reduced chemistry. Combust. Flame 101:287. Rightley M. L. and Williams F. A. (1997). Burning velocities of CO flames. Combust. Flame 110:285. Scholte T. G. and Vaags P. B. (1959). Burning velocities of mixtures of hydrogen, carbon monoxide and methane with air. Combust. Flame 3:511.

128

Synthesis Gas Combustion: Fundamentals and Applications

Shilling N. Z. and Lee D. T. (2003). IGCC—Clean power generation alternative for solid fuels: GE Power Systems. PowerGen Asia. Sivaramakrishnan R., Comandini A., Tranter R. S., Brezinsky K., Davis S. G., and Wang H. (2007). Combustion of CO/H2 mixture at elevated pressures. Proc. Combust. Inst. 31:429. Sun H., Yang S. I., Jomaas G., and Law C. K. (2007). High-pressure laminar flame speeds and kinetic modeling of carbon monoxide/hydrogen combustion. Proc. Combust. Inst. 31:439. Sung C.-J. and Law C. K. (2008). Fundamental combustion properties of H2/CO mixtures: Ignition and flame propagation at elevated pressures. Combust. Sci. Tech. 180:1095. Turns, S. (2000). An introduction to combustion. 2nd ed. New York: McGraw-Hill. Walton S. M., He X., Zigler B. T., and Wooldridge M. S. (2007). An experimental investigation of ignition properties of hydrogen and carbon monoxide mixtures for syngas turbine applications. Proc. Combust. Inst. 31:3147. Wang W. and Rogg B. (1993) Reduced kinetic mechanisms and their numerical treatment. I. Wet CO flames. Combust. Flame 94, 271. Yetter R. A., Dryer F. L., and Rabitz H. (1991a). A comprehensive reaction mechanism for carbon monoxide/hydrogen/oxygen kinetics. Combust. Sci. Tech. 79:97. Yetter R. A., Dryer F. L., and Rabitz H. (1991b). Flow reactor studies of carbon monoxide/ hydrogen/oxygen kinetics, Combust. Sci. Tech. 79:129. Zhang Q., Noble D. R., Meyers A., Xu K., and Lieuwen T. (2005). Characterization of fuel composition effects in H2/CO/CH4 mixtures upon lean blowout. Paper GT2005-68907 presented at the ASME Turbo Expo, Reno, Nevada.

Combustion 5 Turbulent Properties of Premixed Syngas Robert K. Cheng Contents 5.1 Introduction................................................................................................... 129 5.2 General Description of Turbulence Length and Time Scales....................... 131 5.3 Classification of Premixed Turbulent Flames and Turbulent-Flame Interactions.................................................................................................... 134 5.4 Heat-Release Rate and Turbulent Flame Speed............................................ 142 5.5 Effects of Syngas and Hydrogen on Turbulent Flame Speed........................ 154 5.6 Premixed Turbulent Combustion Properties at High Temperatures and Pressures................................................................................................. 157 5.7 Modeling Considerations for Syngas and Hydrogen Flames........................ 161 5.8 Conclusions.................................................................................................... 163 Acknowledgment.................................................................................................... 165 References............................................................................................................... 165

5.1 Introduction Lean premixed combustion is the technological foundation for the ultralow NOX industrial burners and for the dry-low-NOX (DLN) method employed in state-ofthe-art low-emissions gas turbines. Though effective, the lean premixed combustion method is hampered by undesirable side effects, such as flame instability and combustion oscillations. These problems can be exacerbated by the variations in the combustion properties of syngases. Therefore, understanding how the basic properties of premixed turbulent flames change with fuel compositions is essential to the development of fuel-flexible combustors that can burn a variety of syngases derived from coal gasification. Turbulence occurs naturally in all flow systems of practical interest and is one of the most significant unresolved problems of engineering physics. In combustion devices, turbulence is generated by the shearing fluid motions when the fuel and oxidizer streams mix with each other, or when the reactants flow through narrow internal passages. Once generated, turbulence can persist for an extended period before being dissipated by viscous forces. The coupling of turbulence with combustion involves complex thermal, fluid, and chemical processes that occur simultaneously 129

130

Synthesis Gas Combustion: Fundamentals and Applications

over a wide range of spatial and temporal scales. Many of these processes have not been well characterized or fully understood. Nevertheless, premixed turbulent flame models have been developed based on classic turbulence scaling laws and conceptual arguments on the reactive scalar structures and laminar flame behaviors. Turbulence has two roles in a premixed combustor. In the premixer, it promotes molecular mixing of the fuel and oxidizer streams into a homogeneous mixture for delivery to the burner. In the burner, turbulence produced in the wakes of the swirl vanes and the bluff bodies, together with residual turbulence in the flow of the reactants, controls most aspects of the flame. These aspects include the heat release rate, flame properties, flame evolution, and emissions. The most important role of turbulence is to increase the mean heat release rate. For premixed combustion, the rate of increase is often expressed in terms of a turbulent flame speed that is much higher than the laminar flame speed. Elucidating the underlying fluid-mechanical and chemical processes responsible for the increase in mean heat release by turbulence has been the central focus of fundamental research on premixed turbulent combustion. The studies involve measurements of velocity and scalar fields by laser diagnostics, development of turbulent flame models, numerical simulations of turbulence-flame interactions, and characterization of combustion dynamics. Due to the diversity of the subjects, there is a wealth of knowledge and information available in combustion textbooks and review articles (e.g., Libby and Williams, 1994; Peters, 2000; Bilger et al., 2005). The fundamental issue with syngas combustion is associated with the significant variation in their compositions that changes the laminar flame speed, heat release ratio, local fuel consumption rate, and flame front instability mechanisms. The responses of premixed turbulent flames to these changes are not well characterized or understood. Because of the presence of high hydrogen concentration in some syngases, the effects can be substantial and may cause a significant increase in the turbulent flame speed. This is a particular concern to manufacturers of industrial heaters and DLN gas turbines because of the increased propensity for the premixed turbulent flame to flash back. Despite decades of extensive research since Damkohler (1940) introduced the basic modeling concept for premixed turbulent flames, many issues regarding the turbulent flame speed and its relationship to the heat release rate are still unresolved. Several recent reviews called attention to this situation (Lipatnikov and Chomiak, 2002, 2005; Driscoll, 2008) and elaborated on the theoretical and computational approaches to address the problems. This chapter offers an alternate perspective to show that the generalization of Damkohler’s original premise has led to inconsistencies and discrepancies in the experimental data, and their sometimes contradictory implications. Consequently, the collective studies available in the scientific literature cannot give meaningful insights into the fuel effects unless the differences associated with the flame geometry and the experimental methods are scrutinized and carefully considered in the data interpretation. This chapter begins with an overview of the fundamental processes of premixed turbulent flames and the physical models for their analysis. This is followed by a review of the many approaches for defining and measuring the turbulent flame

Turbulent Combustion Properties of Premixed Syngas

131

speeds. The effects of burning syngases are addressed in the context of the characteristics of lean hydrogen premixed turbulent flames. The changes due to high temperatures and pressures are also reviewed. To close, general comments on the characteristic features of lean hydrogen turbulent flame structures and their implications on modeling of the heat release are given.

5.2 General Description of Turbulence Length and Time Scales Turbulence is a property of the flow and not of the fluid. It is sustained by receiving energy from the momentum of the bulk flow through shearing motions at the wall regions or around obstacles, such as swirl vanes or fuel spokes. Due to its random and chaotic nature, statistical methods are required to describe turbulence behavior (Hinze, 1959). The classic statistical turbulence model considers the flow as a chaotic array of eddies of widely different sizes being convected by the mean flow. The largest eddy sizes are on order of the flow passages or the characteristic dimensions of the combustion device. The smallest sizes are on the order of the characteristic length scale associated with molecular transport by viscosity. The wide range of eddy sizes are generated by vortex stretching in a cascading process through which the large eddies break down and transfer their energy to the smaller ones. Eventually, the energy contained in the smallest eddies is dissipated by viscous stresses. The basic property of turbulence can be described in terms of the spatial and temporal statistical distributions of the velocity vector using Reynolds decomposition of the turbulent fluid motion into a fluctuating velocity component, ui , about a mean velocity, Ui . The subscript i denotes the three orthogonal velocity components on the x, y, and z axes. Here, x is the principal axis in the direction of the mean flow. The distribution function for a given stochastic variable, that is, the time-varying velocity component at a point, is expressed by means of its probability density function (pdf), from which the statistical moments can be deduced. The first two moments of the velocity pdf are the mean and the variance. The Reynolds-averaged Navier-Stokes (RANS) equations for Ui and ui contain a correlation turbulent stress tensor that represents turbulence production via a transfer of energy from the mean flow to the turbulent motions. The normal terms of the stress tensor are the root mean square (rms) fluctuating velocity components usually abbreviated as u′, v′, and w′ in combustion literature. The cross-terms are called shear stresses or Reynolds stresses, which are the cross-correlations between two of the three fluctuating velocity components, for example, ui uj . These cross-correlations are deduced from the joint pdfs of the two fluctuating velocity components. The presence of the stress tensor in RANS represents the first closure problem for turbulence modeling. Most commercial RANS codes for turbulent combustion use the κ-ε model to close the shear tensor, with κ being the turbulent kinetic energy and ε being its dissipation by eddy viscosity. The κ-ε model is based on the assumption that the turbulent transport processes are diffusive in nature so that the eddy viscosity relates to the shear forces through the mean velocity gradients. Inherent in this approach is that the turbulence is locally isotropic. Turbulence isotropy, which means

132

Synthesis Gas Combustion: Fundamentals and Applications

1.0

fx (r,t)

0.8

0.6

0.4

0.2

0

2

4

6

8 10 r (mm)

12

14

16

18

Figure  5.1  Two-point cross-correlation of isotropic turbulence produced by a perforated plate. The line is the exponential fit. Integration of the exponential fit gives an integral length scale l x of 2.5 mm, approximately the same as the hole size of the perforated plate.

that all three fluctuation components are equal to zero Reynolds stresses, rarely exists in complex combusting flows. Therefore, the modeling constants in RANS CFD calculations need to be well calibrated against combustion experiments. Dissipation is central to turbulence theory. It invokes the eddy cascade hypothesis to describe the rate at which energy is transferred from the large eddies to the small ones. For free shear flows without wall effects, dissipation is invariant within the inertial subrange of turbulence. The invariant dissipation process provides a theoretical basis for the development of scaling laws for turbulence scales that represent the size and strength of the turbulent eddies. Comparison of turbulence scales with the characteristic scales of premixed laminar flames is the foundation of premixed turbulent combustion theories and models. Experimentally, turbulence length scales are deduced from the cross-correlation coefficient, R( x , r , t ) = ui ( x , t )ui ( x + r , t ), of the velocity fluctuations ui at two points separated by a distance r. A typical shape of the normalized correlation function fi (r,t) = R(r,t)/u 2j (t ) is shown in Figure 5.1, where it approaches 1 at r → 0 and decays to 0 at large r. The decay range represents eddy sizes within the inertial subrange, which contains most of the turbulent kinetic energy. The length scale associated with these eddies is the integral length scale li and is defined by ∞



li (t ) =

∫ f (r , t)dr i

0

(5.1)

133

Turbulent Combustion Properties of Premixed Syngas

3

Dissipation

Energy cascade

Generation

Log E (k)

5

k lt

lk

Energy Spectrum

Figure 5.2  Typical turbulence energy spectrum.

The integral time scale, ti = li/ui′, with ui′ being the rms of the velocity fluctuation, is proportional to the mean turnover time of these eddies. Invoking the assumption of turbulence isotropy gives the Kolmogorov length scales lk at which the turbulent eddies are influenced by viscosity. The basic premise is that the rate of energy transfer from larger eddies at the integral length scale is equal to the dissipation rate of energy at the Kolmogorov scale. A Fourier transform of the isotropic two-point cross-correlation function yields the turbulent kinetic energy spectrum, E(k). It represents the density of kinetic energy per unit wave number, k, which is the inverse of the eddy size. Figure 5.2 shows a typical shape of the energy spectrum for isotropic turbulence on a log-log scale. The spectral distribution at the small wave numbers is in general not universal because it corresponds to the large eddies associated with flow instabilities at the inflow boundary. The maximum of E(k) is usually found at wave numbers corresponding to the integral length scale. The decrease in E(k) at larger wave number represents dissipation within the inertial subrange. The –5/3 slope of decay is a characteristic feature of the dissipation of homogeneous isotropic turbulence, and is derivable from analysis of the energy density distribution in wave number space. The drop-off at k > lk is due to viscous effects. The energy cascade spectrum, or the turbulence spectrum, relates the integral length scale and the Kolmogorov scale to turbulence intensities, and it has been the subject of extensive experimental and theoretical studies. Measuring the two point correlations presents some difficulties, especially at small r on the order of the Kolmogorov scale lk . The situation is much worse in combustion flows where the hostile environments prohibit the use of hot-wire anemometry, which has been the mainstay for verifying turbulence theories.

134

Synthesis Gas Combustion: Fundamentals and Applications

To circumvent these difficulties, the use of autocorrelation function at one point has become the prevalent method to measure the turbulence spectrum and the turbulence length scales. By invoking Taylor’s hypothesis, the turbulent eddies in a homogeneous turbulent flow with Ui >> ui can be perceived as being transported across the point of observation at a rate equal to the mean velocity along x. The fluctuation of the velocity at that point with time will be nearly identical to the instantaneous distribution of the velocity along x so that the autocorrelation function has the same distribution as the two-point cross-correlation function in the principal flow direction. The integral time scale ti, that is, the time integration of the autocorrelation function, can be used to deduce the integral length scale through simple space-time transformation, li = Ui ti. ∞



ti ( x ) =

∫ R(x, t, t + ∆t)dt

(5.2)

0

Analogously, the Fourier transform of the autocorrelation function yields the turbulent frequency spectrum. The integral of the turbulent frequency spectrum is the mean square of the velocity fluctuation. Taylor’s hypothesis, despite the limitation of its application to only the principal flow axis, is clearly a much simpler and convenient approach to measure turbulence length scales. In fact, the turbulence length scales reported in the combustion literature are mostly the longitudinal integral scale, lx, obtained from the autocorrection function.

5.3 Classification of Premixed Turbulent Flames and Turbulent-Flame Interactions The interaction between a discrete vortex and a planar laminar premixed flame is the simplest physical model of the premixed turbulent flame (Figure 5.3). Here, the laminar flame is considered to be a thin interface separating the reactants and products. This interface propagates into the reactants at a displacement flame speed, sD, which is close to the laminar flame speed, SL . When the size of the vortex is large compared to the thickness of this interface, the planar flame front becomes wrinkled as it burns through the vortex. The degree of flame wrinkling depends on the ratio between sD and the turbulence intensity of the single vortex, defined by the eddy size and its turnover velocity. If the flame front propagation rate is unchanged while it burns through the vortex, that is, fuel consumed by a unit area on the flame surface remains constant, wrinkling of the flame front increases the flame surface density per unit volume and thus the volumetric heat release rate. Wrinkling is therefore the primary process through which turbulence increases the mean heat release rate of a premixed flame. When the flame interacts with multiscale turbulence, its flame wrinkle structures become more complex in response to the range of turbulent eddy sizes. Typical wrinkled flame structures at low turbulence are shown in Figure 5.4 by a laser tomographic image of a premixed CH4/air flame at equivalence ratio ϕ = 0.7 (laminar flame speed SL = 0.196 m/s). This flame was generated in a low-swirl burner that

135

Turbulent Combustion Properties of Premixed Syngas

Products

SD

Vortex

Fl

am

ef

ro

nt

Reactants

Figure 5.3  Model of a planar laminar flame interacting with a discrete vortex. SD is the displacement flame speed of the front. (With permission.)

20 mm

Products Flame

Reactants

Co-flow air

Figure 5.4  Laser sheet tomographic image of a premixed turbulent flame in the wrinkled flamelet regime. The cold reactants appear bright due to Mie scattering from oil aerosol introduced in the flow. The oil aerosol burned and evaporated in the hot products to outline the topology of the wrinkled flame. (With permission.)

supplied a flow of reactants at a bulk flow velocity U0 = 5 m/s, turbulence intensity of u′ = 0.32 m/s, and integral length scale of lx = 3.5 mm (Cheng et al., 2002). In laser tomography, the reactants are seeded with an oil aerosol that burns and evaporates at the flame front. When illuminated by a thin laser sheet, the wrinkled flame is outlined by the sharp boundary between the reactants, which appeared bright due

136

Synthesis Gas Combustion: Fundamentals and Applications C=1 Reaction zone

Preheat zone

Products

Reactants SL,0 C=0

dL Thermal thickness lf ~ 1 mm

Figure 5.5  Inner structure of a laminar flame.

to Mie scattering from the aerosol, and the products, which appeared dark due to absence of light scattering sites. The tomographic image of Figure 5.4 shows that the flame wrinkles have different sizes but are all on the order of 5lx . The topologies of the convex and concave wrinkles (relative to the reactants) are different. Whereas the convex wrinkles are rounded, the concave wrinkles are sharp and form into cusps. Formation of the flame cusps is a consequence of the flame fronts merging into each other as they propagate in the concave region. This occurs only when the eddy turnover time, estimated by lx /u′, is long compared to the flame residence time dL/sD, where sD is the local displacement flame speed and dL is the reaction zone thickness of the laminar flame defined by dLSL = ν, with ν being the fluid viscosity. The reaction zone thickness, dL , is one of two characteristic length scales of a laminar flame that is a convenient scaling parameter for premixed turbulent combustion. The other characteristic length scale is the preheat zone thickness, lf , and it is sometimes referred to as the thermal thickness. Figure 5.5 shows a schematic of the inner structure of a laminar flame to differentiate the two. The preheat zone is a chemically inert region where the temperature of the reactants increases due to heat transfer from the chemically active reaction zone in which fuel is consumed and radicals are depleted by chain-breaking reactions. For a CH4/air flame at atmospheric temperature and pressure, the preheat zone thickness, lf , is on the order of 1 mm. The corresponding reaction zone thickness is smaller by about an order of magnitude. Damkohler (1940) introduced the flame wrinkling concept as a theoretical foundation for premixed turbulent combustion. The formation of the flame cusps at low turbulence was one of the first characteristics to be identified. This phenomenon illustrates that the features of the flame wrinkles are dependent on the ratios of the turbulence scales and the flame scales. Subsequent elaboration on the flame wrinkling concept led to the classification of premixed turbulent flames into various regimes (Borghi, 1985; Peters, 1986; Abdel-Gayed et al., 1989; Poinsot et al., 1990). Many authors proposed regime diagrams that are expressed on phase spaces defined by the nondimensional parameters for the initial turbulence and flame conditions. Figure 5.6 shows a regime diagram by Peters (2000) that is expressed in terms of

137

Turbulent Combustion Properties of Premixed Syngas 103

Broken reaction zones

u´/SL

102

101

Re t

1

10–1 10–1

=

100

e: gim e Re than n o l Kaδ = 1 n Z smal ss ctio e Rea ce scal hickne n i Th ulen ame t n b erio Tur rmal fl Crit e s h m t illia l = lm ov-W m i l 1, K Ka = Wrinkled Flamelet Regime: Turbulence scale > chemistry scale Laminar Turbulence does not affect reaction zone flames 100

101

102

103

104

lx/dL

Figure 5.6  Regime diagram for premixed turbulent flames according to Peters (2000).

u′/SL and lx /dL . The nondimensional numbers associated with the two parameters are the turbulent Reynolds number,

Ret = (u′/SL)(lx/dL)

(5.3)

the Damköhler number representing the ratio of the turbulent time scale, tt , and the flame time scale, tf :

Da = (lx/dL) (u′/SL) –1 = tt /tf

(5.4)

and the Karlovitz number for the ratio of the reaction zone thickness to the Kolomogorov dissipation scale:

Ka = (u′/SL )3/2 (lx /dL) –1/2 = dL/lk

(5.5)

The boundaries between the regimes are defined by the lines Ret = 1 and Ka = 1. The Ret = 1 line separates the laminar flames from the turbulent flames. The Ka = 1 line in the turbulent flame region is known as the Klimov-Williams criterion, and it separates the wrinkled flamelet regime below from the thin reaction zone regime above. Ka = 1 means that the Kolmogorov scale of the incident turbulence is equal to the reaction zone thickness of the laminar flame. For flames with initial condition of Ka < 1, the smallest relevant turbulent eddies are larger than the reaction zone thickness. The role of turbulence is to wrinkle the flame front without affecting its inner chemical structure. The flame in Figure 5.4 is an example of a flame in the wrinkled flamelet regime. Flames with Ka > 1 are within the thin reaction zone regime where

138

Synthesis Gas Combustion: Fundamentals and Applications

10 mm Flame 1 u´/SL = 3.45 Parallel contours

c > 0.8 c = 0.7 c = 0.6 c = 0.5 c = 0.4 c = 0.3 c = 0.2 c < 0.2 Flame 6 u´/SL = 18.7 Complex topology and pockets

Figure  5.7  Two-dimensional Rayleigh scattering images of ϕ = 0.7 CH4/air flames with moderate, Ka = 1.3 (left), and intense, Ka = 17.1 (right), turbulence. (From Shepherd et al., 2002. With permission.)

the smallest turbulent eddies are smaller than the reaction zone thickness. These smallest eddies may influence the reactive-diffusive structure while the large eddies wrinkle the flame fronts. Peters also proposed an additional criterion of Kaδ = 1, with Kaδ defined by an inner-layer length scale lδ that replaces dL in Equation 5.5. lδ characterizes the fuel consumption layer within the flamelet. The physical meaning of Kaδ > 1 is that the smallest turbulent eddies can penetrate into the inner layer to disrupt the reactive-diffusive structures and locally quench the flamelet. Therefore, the region above the Kaδ = 1 line is called the broken reaction zone regime. In Figure 5.7, the wrinkled structures of flames at Ka ≈ 1 and Ka > 1 are shown by the two planar laser-induced Rayleigh scattering images from Shepherd et al. (2002). These lean CH4/air flames of ϕ = 0.7 were generated in a low-swirl burner with isotropic turbulence intensity, u′/U0, approaching 25%. Rayleigh scattering is the inelastic light scattering from the gas molecules, and its intensity is proportional to the gas density. Here, the field of view (22 mm × 20 mm) is smaller than in Figure 5.4 to zoom in on a portion of the flame brush so that the density distribution within the thin flame fronts can be resolved. The two small gray-scale images at the top left corners are the raw data where the higher-density reactants appear bright. These images showed that the increase in Ka from 1.3 to 17.1 produces highly convoluted flamelets with fine wrinkles. The analyzed images show the contours of the instantaneous reaction progress variable, c, as defined in Figure 5.5. Shepherd et al. (2002) applied statistical methods to determine if the higher turbulence intensity broadened the preheat zone according to the conjecture for flames in the thin-reaction zone regime. Using the spaces between the instantaneous c contours to quantify flame front broadening, the statistical results from 100 Rayleigh scattering images did not give strong evidence to show that significant broadening occurred at high Ka. Other authors have used Rayleigh scattering or similar methods, and found some isolated preheat zone broadening in burners with more complex turbulence structures, such as a piloted jet Bunsen flame

Turbulent Combustion Properties of Premixed Syngas

139

(e.g., Chen and Bilger, 2002). In a recent review, Driscoll (2008) concluded that there is sufficient experimental evidence to show that the flame front broadening phenomenon caused by small eddies penetrating into its preheat zone is a rare event even for flames with Ka >> 1. This suggests that the Klimov-Williams criterion should be interpreted as a threshold above which the penetration of small-scale turbulence eddies into the thermal layer becomes statistically plausible. The main implication is that the wrinkled flamelet model for premixed turbulent flames should be valid for most practical combustion systems that burn natural gas (Shepherd et al., 2002). Flame wrinkling also induces stretch on the flame front due to the variations in the flame curvatures and the local velocity divergence. The process can be modeled as locally stretched laminar flames so that the turbulent flame can be considered as an ensemble of stretched flamelets (Law and Sung, 2000). Flame stretch is defined by Williams (1985) as the time rate of change of flame area per unit area, A.

a=

1 dA A dt

(5.6)

How a laminar flame reacts to curvature and strain is dependent on its thermal/ diffusive and hydrodynamic instability mechanisms. Whether or not the flame front is inherently stable or unstable can dampen or promote flame wrinkle formation. There is a large body of work on the effect of stretch on laminar flames to support the development of the stretched flamelet model (Law and Sung, 2000). From these studies, a general expression has been derived for the local displacement flame speed, sD, that includes first-order correction terms for small curvatures and strains.

S D = S L 0 − S L 0 LK c − LK s

(5.7)

Here, SL0 is the unstretched laminar flame speed, Kc is the flame curvature with positive values for flame fronts that are convex to the reactants (in units of mm–1), Ks is strain imposed by the local velocity gradients at the flame front, and L is the Markstein length on the order of dL . The ratio of the Markstein length and the reaction zone thickness, L/dL , is the Markstein number, Ma, which is a physicochemical property of the reactants. The expression for Ma derived from high activation energy asymptotics shows that it is a function of the reduced activation energy (the Zeldovitch number), the heat release ratio of the mixture τ = Tad /T0 – 1, with Tad being the adiabatic flame temperature, and the Lewis number, Le, representing the ratio of the thermal and mass diffusivity. The Lewis number characterizes the tendency of a laminar flame to become unstable or more stable due to an imbalance in the diffusion of species and heat. This stabilizing or destabilizing mechanism has a significant influence on the properties of the premixed turbulent flames. For the lean methane/air flames, the Lewis numbers are close to unit. This means near-equal diffusivity of species and heat within the flame fronts, and the Markstein length is small with a value of near zero. Under these conditions, sD on the wrinkle methane/air flamelets is relatively insensitive to curvature and strain. For lean

140

Synthesis Gas Combustion: Fundamentals and Applications H2

CH4

φ = 0.8, Le ≈ 1

C3H8

φ = 0.75, Le = 1.85

10 mm

U0 = 15 m/s U´/lx = 210 sec–1

U0 = 5 m/s u´/lx = 57 sec–1

φ = 0.3, Le = 0.33

Figure 5.8  OH-PLIF images of lean H2, CH4, and C3H8 flames with moderate (top) and intense (bottom) turbulence.

p­ ropane/air flames, the heavier fuel molecules diffuse more slowly than the air molecules such that Le > 1 and Ma > 0. Consequently, sD of the flame fronts with positive curvatures (convex to the reactants) and stretch rate is reduced, and sD the flame fronts with negative curvatures (concave to the reactants) and compressive stretch is enhanced. This is a stabilizing mechanism that counteracts the effects of turbulence and retards the formation of flame wrinkle. For lean H2/air flames, the highly diffusive H2 molecule in air means that Le < 1 and Ma < 0. Therefore, sD of the concave flame front is increased while the convex flame front is reduced. This is a destabilizing mechanism that augments flame wrinkling induced by turbulence. The thermal/diffusive effect, also commonly referred to as the Lewis number effect, on premixed turbulent flame wrinkle structures is shown in Figure  5.8 by the planar laser-induced fluorescence images of the OH radicals (OH-PLIF) of six lean H2, CH4, and C3H8 flames at two bulk velocities, U0, of 5 and 15 m/s. These flames were generated in a low-swirl burner with turbulence produced by a perforated plate (same as in Figure 5.4). The laminar flame speeds SL for the three H2, CH4, and C3H8 mixtures are nearly the same, but the values of Le are respectively 0.33, 1, and 1.85. OH radicals are formed at the flame front and their concentrations decay slowly in the hot products. The bright regions on the OH-PLIF images are the hot products, and the boundary between the bright and dark region outlines the wrinkled flame structures. The three cases at the top are at the lower velocity, U0 = 5 m/s, where the OH-PLIF image for the lean H2 flame shows highly convoluted flame fronts whose topologies are distinctly different than the less wrinkled CH4 and C3H8 flames. The H2 wrinkles are smaller and characterized by elongated and deep inlets instead of sharp flame cusps. In comparison, the flame wrinkles of the CH4 and C3H8 flames are significantly less corrugated. At the higher velocity, U0 = 15 m/s, the H2

141

Turbulent Combustion Properties of Premixed Syngas 1 H2 C3H8

Normalized Probability

0.8

0.6

0.4

0.2

0 –4

–2 0 2 Flame Front Curvature (1/mm)

4

Figure 5.9  Normalized probability density functions of flame front curvature for a H2/air flame (ϕ = 0.3, Le = 0.33) and a C3H8/air flame (ϕ = 0.75, Le = 1.85), both at u′/sL = 1.16. (From Goix and Shepherd, 1993. With permission.)

flame becomes even more convoluted and breaks up into islands of reactants and products. These flame structures are not found in the CH4 flame, though the flame wrinkles become smaller at the higher U0. In comparison, the C3H8 flame structures are less wrinkled than the CH4 flame and show that the flame is not as sensitive to turbulence as the CH4 and H2 flames. Due to the random and chaotic nature of the turbulent flames, the degree of flame wrinkling varies from one instance to the next. Statistically, the flame wrinkle structures can be more or less convoluted than shown in Figure 5.8. To study the statistical distribution of the flame wrinkle structures, the local flame curvature Kc is often used to quantify the flame wrinkle scale. Figure 5.9 compares the probability density functions of Kc from Goix and Shepherd (1993) for a H2/air (ϕ = 0.30) and a C3H8/ air (ϕ = 0.75) turbulent flame with u′/SL = 1.6 generated by a stagnation flow burner. These pdfs show that Kc has both positive (convex) and negative (concave) values, with the most probability values at zero, that is, locally flat flamelets. Therefore, the mean value of Kc is not very meaningful because it is very small and near zero. The differences between the two flames are found at large curvatures corresponding to the small flame wrinkles. The more convoluted H2/air flame has higher probabilities of flame fronts with larger positive and negative curvatures than the C3H8/air flame. The Kc pdf for the H2/air flame is skewed toward the positive curvature, while the distribution for the C3H8 flame is more Gaussian in shape. This shows that the topology of the flame wrinkles is affected by the Lewis number. The comparison of the Kc pdfs in Figure 5.9 shows that the statistical distributions of the flame curvatures for the flames with Le < 1 and Le > 1 at low turbulence are not significantly different despite the readily noticeable differences in the overall

142

Synthesis Gas Combustion: Fundamentals and Applications

topology of the flame wrinkles. The high probabilities of flame fronts with small curvatures mean that the contribution from the second term in Equation 5.7 is small. Because the strain effect, that is, the third term of Equation 5.7, is also small at low turbulence levels (i.e., u′/SL ≈ 1), the wrinkled laminar flame model that assumes SD = SL0 everywhere on the flame surface provides a good description of flames with Le ≈ 1. At high turbulence levels typical of those encountered in DLN gas turbines and industrial burners, the strain and curvature effects become nontrivial and the stretched laminar flame model of Equation 5.6 is required. Obviously, the wrinkled laminar flame model would not be appropriate for flames with Le < 1 even at low turbulence; this topic is covered in Section 5.7. Due to its relevance to premixed turbulent flames, there is a significant amount of research on stretched laminar flame theory to model the aerodynamic effects on flame properties. Many review articles are available in the scientific literature (e.g., Law and Sung, 2000) on the effects of stretch on the laminar flame speed, reaction rates, flame front instability mechanisms, and extinction limits. Experimentally, stretched laminar flame studies have been performed using burners that produce statically stretched or curved flames. How a laminar flame responds to dynamic stretch and curvature, as in a real turbulent flow, remains unexplored. For H2 flames, due to their inherently unstable nature, a stationary and stable laminar H2 flame can only be generated by subjecting the flame to very high stretch rates or by burning at high equivalence ratios where the Lewis number is near unity. Therefore, detailed knowledge on the fundamental properties of unstretched and stretched lean H2 laminar flames is still lacking.

5.4 Heat-Release Rate and Turbulent Flame Speed The idealized double-infinite one-dimensional planar turbulent flame brush of Figure 5.10 is a convenient model to illustrate the relationship between the turbulent heat release rate and the increase in flame front surface area due to flame wrinkling. Assuming that the turbulence in the reactants is isotropic, and the mixture has a high Damkohler number with Le = 1, the local displacement speed, sD, of the wrinkled flame fronts is constant and equal to SL . Within a streamtube of cross section area AT , flame wrinkling causes the flame front surface area AL to be larger than AT . The mass flow rate into the streamtube is then m = ρr sL AL = ρr ST AT , where ρr is the density of the reactants and ST is the mean velocity of the reactants flowing into the streamtube. ST increases above SL in direct proportion to the increase in flame surface area:

AL ST = AT S L

(5.8)

The flame surface area ratio, AL/AT , quantifies the increase in the heat release rate due to the flame wrinkling process. Its counterpart, the ST /SL ratio, shows that the flow velocity in the streamtube is also increased to compensate for the higher fuel consumption rate of the larger flame surface area. If the turbulent reactants are in a

Turbulent Combustion Properties of Premixed Syngas

143

SD Thin wrinkled flame sheet

Reactants U = ST

Products

u´ = v´ = w´

Flame brush

Figure 5.10  Idealized one-dimensional double-infinite planar turbulent flame.

stationary frame of reference, ST is the propagating speed of the planar turbulent flame brush through the mixture. Therefore, ST is known as the turbulent flame speed. The relationship between flame surface area and the turbulent flame speed is the original concept introduced by Damkohler (1940), who identified AL/AT and ST /SL as the parameters for the mean heat release rate of premixed turbulent flames. Damkohler also developed scaling laws for ST . For large-scale turbulence, the interaction with the flame fronts is kinematic in nature. In the limit of large u′/SL the ST increase linearly with u′ as

ST u′ = 1+ SL SL

(5.9)

For small-scale turbulence that can modify the preheat zone, ST scales with the length scale ratio lx /dL as well as with u′/SL:

ST ≈ SL

u′ l x SL d L

(5.10)

Subsequent studies attempted to merge the two limiting cases of Damkohler’s ST scaling laws. The outcome is an expression in the general form of n



 u′  ST = 1+ K   SL  SL 

(5.11)

Measurements of ST to obtain empirical values for the correlation coefficients K and n in Equation 5.11 remain a major focus of experimental studies. Earlier works

144

Synthesis Gas Combustion: Fundamentals and Applications

show that K is proportional to the heat release rate and n is on the order of 0.7 to 1. The deviation of the exponent n from unity is often referred to as the bending effect. This bending is considered an important problem of premixed turbulent combustion because it indicates that the turbulent flame has a nonlinear response to the turbulence intensity. Despite extensive efforts, a universal turbulent flame speed correlation has yet to be found (Bradley, 1992; Lipatnikov and Chomiak, 2002). The results are, in fact, quite scattered on the ST /SL versus u′/SL plot, with the constants K and n varying with experimental configuration and fuel type. The reasons for the large inconsistencies in the ST data will be discussed in the latter part of this section. The other parameter for the turbulent heat release, AL/AT , is most often expressed in terms of the flame surface density, Σ, which is the local flame surface area per unit volume. Σ is convenient for modeling because it can be interpreted as a characteristic length scale of the wrinkled flamelets (Vervisch and Veynante, 2000). In the species conservation equation, the local heat release rate can be closed by a balance equation for Σ, expressed in terms of the tangential strain rate on the flame surface, flame propagation at the displacement flame speed, and the curvature effects. These are the same basic processes in the stretched flamelet model of Equation 5.7. Σ relates to ST through the flame surface integral in the direction locally normal to the turbulent flame brush η (Bray and Cant, 1991): +∞



ST = S D

+∞

∫ Σ ∂η = S I ∫ Σ ∂η L0 0

−∞

(5.12)

−∞

Here, a generalized form of the displacement flame speed, sD, is used. It is the unstretched laminar flame speed, SL0, modified by a stretch factor, I0. Numerical simulations have shown that I0 is about unity for small Markstein numbers (Bell et al., 2002; Hawkes and Chen, 2006). Experimental verification of Equation 5.12 is difficult, however, because measuring the flame surface density, Σ, and the stretch factor, I0, requires statistical analysis of two- or three-dimensional information of the flame wrinkle structures, as well as instantaneous local flow divergence information. Earlier experimental studies focused on conditions with I0 ≈ 1, and used the flame crossing length as a one-dimensional representation of Σ. The flame crossing length, together with later two-dimensional studies of Σ, has shown that the distributions of Σ have the same general shape (Shepherd, 1996):

Σ = 4 Σmax c (1 − c )

(5.13)

where c is the mean progress variable (or reactedness). It represents the probability of encountering the products, and has the values of 0 in the reactants and 1 in the products. Σmax is the maximum value of Σ found somewhere near c = 0.5. The distributions of c in the coordinate η normal to the flame brush also assume the form of an error function (Lipatnikov and Chomiak, 2002). This allows the flame surface integral to be further reduced to

Turbulent Combustion Properties of Premixed Syngas

145

+∞



ST = S D

∫ Σ ∂η = I

0

Σmax δT

(5.14)

−∞

where δT is the turbulent flame brush thickness. Despite the universality of the Σ(c) distribution, experimental verification of Equation 5.14 has proven to be challenging. The data obtained thus far show that the flame surface integral gives values that are much smaller than ST . The main reason for the inconsistency between ST and the flame surface integral is that the equality of Equation 5.8 is unambiguous and physically meaningful only for the idealized onedimensional planar premixed turbulent flame. Unfortunately, this one-­dimensional system cannot be realized in practice. In laboratory burners and in practical com­ bustion systems, the shape, structures, and dynamics of the turbulent flames are ­controlled by many factors, including the boundary conditions, the size of the burner, the geometry of the flame stabilizer, and the velocity distributions at the inlet. In all cases, the flame structures evolve in space and time. In adapting the basic concept of Figure 5.10 to laboratory burners, many different approaches have been taken without careful consideration of the influences from the boundary conditions, and from the burner and flame geometries. This has contributed to a large scatter and inconsistency in the published results. To illustrate, consider the simple rod-stabilized v-flame configuration of Figure 5.11a, where the twin flame brushes are stabilized by the small recirculation zone in the wake of the rod, and propagate into isotropic turbulence produced by a grid placed upstream. The initial boundary condition stipulates that there is no flame wrinkle at the anchoring point. The wrinkles begin to develop and grow away from the stabilizer as the flame frees itself from the recirculation zone and interacts with turbulence in the freestream. The twin flame brushes broaden due to the development of the wrinkles and turbulent diffusion. This means that the values of Σ, as well as the flame brush thickness, vary with distance downstream. The flame brushes are oblique to the flow so that the flowlines through the flame are not normal to the mean flame brush orientations as marked by the mean c contours. The turbulent flame thickness obtained along the flowlines is much larger than the thickness defined by the direction η normal to the c contours (Cheng and Shepherd, 1991). Obviously, integrating Σ through the two different paths gives very different results. Additionally, due to the growth in the flame brush thickness, the flame surface integral is a local value that changes with the distance from the stabilization point. This clearly shows that the flame wrinkle structure and the heat release of a turbulent flame are not solely dependent on the turbulence characteristics in the reactants (Driscoll, 2008). The measurement of ST for the v-flame is equally problematic. Here, ST is defined as the velocity component normal to the oblique stationary turbulent flame brush as |U| Sinθ, where |U| is the magnitude of the velocity vector in the reactants and θ is the small relative angle between the velocity vector and the c contour chosen to represent the mean turbulent flame brush orientation (Smith and Gouldin, 1979). The value of ST defined in this manner is unique only for flames with uniform

146

Synthesis Gas Combustion: Fundamentals and Applications

Products

80 60 40 20

Reactants –40 –20 20 40 0 Transverse Distance (mm)

110

60

Stagnation Plate

90

c–= 0.50

c–= 0.95

– c = 0.05

70 50 –60

–40

–20 0 20 40 Transverse Distance (mm)

60

120 Axial Distance × (mm)

Vertical Distance (mm)

(b) Stagnation burner

0 –60

(c) Low-swirl burner

c–= c–= 0.9 0.5 5 0 c–= 0. 05

Vertical Distance (mm)

(a) V-flame

100

Flame Zone

Flowlines from Velocimetry

100 80 60

– c = 1.0

40

– c=0

20

60

c–=

5

80

.50

100

c–= 0

120

c–= 0.0

Vertical Distance (mm)

(d) Conical flame

140

0. 95

–50 0 50 Transverse Distance (mm)

40 20 0 –60 –40 –20 0 20 40 60 Transverse Distance (mm)

Figure  5.11  Experimental configurations for stationary premixed turbulent flames and their overall flow field patterns outlined by the flowlines.

147

Turbulent Combustion Properties of Premixed Syngas

flame brush thickness. This means that the c contours are parallel and the flame angle is independent of which c contour is chosen to represent the mean turbulent flame brush orientation. The condition cannot be satisfied in a v-flame because the flame brush thickness increases downstream of the flame stabilizer, and the c contours are divergent. The flame brush angle defined at the leading edge of the flame brush is therefore larger than the angle defined at the trailing edge. Consequently, the value of ST varies depending on which c contours are used to define θ. The variation becomes larger as the flame brush thickness increases away from the flame stabilizer. Other features of the v-flame brush and the flow field also contribute to the inconsistencies in the ST results. The fact that the flame brush is slightly curved, and the flame angle is not constant, shows that ST is a local function that varies with the distance downstream. Even more inconsistencies are introduced when the ST data are correlated with the turbulence intensity u′. Because ST determined for the v-flame is a local value, it should be correlated with the local u′ measured at the same location. But some investigators used the u′ value measured at the inlet of the reacting or the nonreacting flows. This less than precise approach disregards the fact that turbulence evolves with distance downstream due to a combination of natural turbulence decay and production by the fluctuating pressure field generated by the turbulent flame. The v-flame is a good example to show that the large inconsistencies in the ST data are caused by the use of an overly simplistic model for the complex flame configuration, and by a lack of strict discipline in the experimental methods. Unfortunately, these problems are prevalent in laboratory studies of ST in all burners. To begin to address these problems, a group of researchers through a series of workshops (Cheng and Gouldin, 2004) has proposed a set of guidelines on the acceptable approach to measure ST from different laboratory experiments. The foundation of the guideline is the recognition that there are four different ways to define ST. The attributes of the laboratory flame configuration and its flow field dictate which definition applies. The four definitions are:

  ST,LD = Local Displacement Speed = U f − Ur = |U| Sinθ



ST,GD = Global Displacement Speed = 

∂ rf  − Ur ∂t

(5.15)

(5.16)

+∞



ST,LC = Local Consumption Speed = S L 0 I 0

∫ Σ ∂η

(5.17)

−∞



ST,GC = Global Consumption Speed = 

m r ρr AT

(5.18)

Equation 5.15 shows that the turbulent flame speed for the v-flames is called the local displacement speed and is a special case where the propagating velocity of the

148

Synthesis Gas Combustion: Fundamentals and Applications

 flame brush with respect to the frame of reference is stationary so that U f = 0. As discussed above, the relative angle θ is very small for v-flames (about 9 to 11 degrees for the flowlines in Figure 5.11a) and is a source of large uncertainties. To minimize the uncertainties, laboratory burners that produce detached locally normal flames such as the stagnation flow burner (Figure 5.11b) and the low-swirl burner (Figure 5.11c) have been developed for measuring ST,LD. The wrinkled flames generated in these burners are free to fluctuate in response to the inflow turbulence because they are not held by a flame holder. The flame brush is locally normal to the approach flow at the centerline, and by definition the mean axial velocity entering the flame brush is ST,LD. The situation is analogous to defining the stretched laminar flame speed from the classic opposed flow burner. In the detached flame configurations, the velocity in the reactants decelerates toward the flame brush due to mean flow divergence. Therefore, ST,LD cannot be inferred and requires direct measurement of the velocity statistics by velocimetry methods (Cho et al., 1986; Liu and Lenze, 1988; Chan et al., 1992; Kostiuk et al., 1993; Bedat and Cheng, 1995). Consequently, the ST,LD results for these burners are more precise and consistent. Away from the centerline, the inflow velocity vectors are not normal to the flame brushes. Measurements of ST,LD at the off-center locations require the definition of a flame brush orientation, that is, same procedure as for the v-flames, but the uncertainties are much less because θ is close to normal. But almost all the ST,LD data for stagnation burners and low-swirl burners have been obtained at the centerline due to their relevance to the stabilization processes in these configurations. The global displacement speed, ST,GD of Equation 5.16, is the average of ST,LD over the entire turbulent flame brush. It is meaningful only for a system where the propagating flame brush has no edge. This condition is satisfied in an expanding flame kernel initiated by a spark in a combustion vessel (Figure 5.12a). ST,GD is the averaged expanding of the flame kernel, r f , relative to the velocity  rate of the radius  of the reactants, Ur . In practice, Ur is often assumed to be zero, even though analyses have shown that the expanding flame induces fluid motions ahead of its front at (a) Single flame kernel

(b) Twin flame kernel

Flame sheet

Ignitor

Ignitor

Products R

Reactants U∞ = 0.0, u´ – c = 0.0 0.5 1.0 Point Spherical Flame

Fuel/Air Twin Spherical Flames

Figure 5.12  Experimental configurations for transient premixed turbulent flames.

149

Turbulent Combustion Properties of Premixed Syngas

velocities that can be nontrivial. Another source of uncertainty is the choice of the flame surface to represent rf . The situation is analogous to the uncertainties associated with the choice of a mean flame surface for a v-flame. Wrinkling of the expanding turbulent flame kernel generates a thickened flame brush, and rf defined at the leading edge of the flame brush is larger than r f at the trailing edge. Because the flame brush thickness increases as the flame kernel expands as it interacts with turbulence, the rate of expansion at the leading edge is greater than the rate at the trailing edge. Moreover, the flame kernel cannot respond to the full turbulence spectrum until it reaches a certain size, and the ST,GD determined during the developing stage is not consistent. Uncertainties also arise when the flame kernel is not spherical. The different approaches to extract r f from the nonspherical flame kernel (sometimes analyzing only half of the flame kernel, as in Kido et al. (2002)) bring about more inconsistencies in the collective data set. The spark-initiated expanding turbulent flame kernel is of significance because it was the only way to measure turbulent flame speeds at high u′/SL . Therefore, an overwhelming majority of the turbulent flame speed data at high u′/SL have been measured in this configuration. The data set reported in the combustion literature consists of a mix of ST,GD and ST,LD. Some of the data were ST,LD, measured by the approaching speeds of the simultaneously ignited twin flame kernels method shown in Figure 5.12b (Abdel-Gayed et al., 1987), or by the transit time of a part of a large expanding flame kernel between two observation points (Shy et al., 2000). Not surprisingly, the data have significant scatter when shown on the ST /SL versus u′/SL plot, but they all have a consistent trend, as shown in Figure 5.13 from Bradley (1992). At low u′/SL , ST,LD increases linearly before bending at higher u′/SL and eventually leveling off. Each data set terminates at a u′/SL value where the turbulent flame kernels cease to propagate after ignition. The termination point is often referred to as 20

18

6000

16 12 20

04

KLe = 5.3 (Poinsot et al.) 1.90 3000 2000

000

250

25

500

30

1000 KLe = 6.0

35

KL

40

0

0

0.3

0.4

20 2 10 e) = R LI (L

5

3000 0.6

0 1.9 50

0.1

4

e=

0.0 7

5

KLe = 0.00 13 0.0050 0.011 0.02 0.030 0.04 0 0.05 3 0.1 0 KL

ut/ul

10

e=

5 0.2

3

3

1 0.2

4000 3000

8

0

15

14

KLe = 1.00

500

10 u´k/ul

250

0

.0 e1 KL 500

6000 5000

1000

2 0 e) = 10 R LI (L

ch en Qu

15

20

Figure  5.13  Correlation of turbulent burning speeds, ut. (From Bradley, 1992. With permission.)

150

Synthesis Gas Combustion: Fundamentals and Applications 25 Previous results [11] Present results

20

ST/SL

15

10

Domain of Bradley’s date [15]

5

0

0

2

4 q´/SL

6

8

0.00

2.82

5.64 u´/SL

8.46

11.28

Figure  5.14  Correlation of the local displacement flame speeds, ST,LD, of a low-swirl burner. (From Bedat and Cheng, 1995. With permission.) Previous results refer to Chan et al. (1992) and Bradley’s data refer to Bradley (1992).

the quenching limit, implying that the wrinkled flame kernel is quenched by intense turbulence. This seems to be a misnomer because the spark energy can have a large influence on whether or not a stable flame kernel can be sustained after ignition. In an extensive review, Lipatnikov and Chomiak (2002) called attention to the large discrepancies in these data and emphasized the need to understand the bending effect of turbulent flame speed and its quenching limit. More recent studies of ST,LD in stagnation burner (Liu and Lenze, 1988) and in low-swirl burner (Bedat and Cheng, 1995) have produced data for stationary flames at the same u′/SL levels as the transient flame kernel experiments. Figure 5.14 shows that these ST,LD results have a continuous linear increasing trend with u′. The magnitudes of the ST,LD and their rate of increase with u′ are higher than those reported by Bradley (1992). The differences illustrate again that the turbulent flame speed is very sensitive to the flame configuration and boundary conditions. The flame brush in a stagnation burner or a low-swirl burner is detached and has no anchoring point. It is free to interact with the turbulent eddies and can be considered fully developed. The ST,LD measured at the centerline is therefore a steady-state value. A nonbending ST,LD trend means that the fully developed flame brush has no lag in its response to the increase in turbulence intensity. The expanding flame kernel, on the other hand, is a developing flame that may or may not have reached a steady state during the time of the measurement. Therefore, the size of the combustion chamber or, in Bradley’s case, the distance between the two ignition points has

Turbulent Combustion Properties of Premixed Syngas

151

an influence on the measured results. If the chamber is not sufficiently large, the flame may not have reached a quasi-steady state or fully responded to the turbulent eddies before impinging on the vessel walls or merging with each other. The situation is similar in a v-flame in that the flame brushes are developing away from the stabilizer, and ST,LD measured at different downstream locations are not consistent because the flame structures are still evolving. The fact that nonlinear ST,LD correlations are found in the expanding flame kernel experiments, as also in the oblique region of v-flames and conical Bunsen-type flames (Figure 5.11d), suggests that the bending effect is a characteristic of developing turbulent flame brushes. This aspect has been considered by Lipatnikov and Chomiak (2002), who showed analytically that flame development can mimic the bending effect of premixed turbulent flame speed. The local consumption flame speed ST,LC of Equation 5.17 and the global consumption flame speed ST,GC of Equation 5.18 are the counterparts of ST,LD and ST,GD. ST,LC is the flame surface integral and its definition does not require a choice of a mean flame surface. It is often considered to be a less ambiguous way to quantify the mean heat release rate than ST,LD and ST,GD. But as discussed earlier, when the flame is oblique to the flow, the value of ST,LC is sensitive to the choice of the integration path. The global consumption flame speed ST,GC defined in Equation 5.18 is an average value obtained from the mass flow rate, m r , through a mean flame surface area, AT . This definition is valid only for flame configurations such as Bunsen-type conical flames (Figure  5.11d), where all the reactants flow through AT and cannot escape around the edges of the flame brush. In a recent extensive review, Driscoll (2008) argued that the quantities ST,LD, ST,LC, and ST,GC are not expected to be equal because the flame wrinkle structures are influenced by the flame geometry and the boundary conditions.* He stressed that their significance is to provide quantitative features of the premixed turbulent flames for validating models and simulations—the justification being that models utilizing the differential equation for Σ (Duclos et al., 1993; Peters, 1999; Hawkes and Cant, 2001; Pitsch and Duchamp de Lageneste, 2002) and time-dependent three-­dimensional direct numerical simulations (DNSs) with full chemistry (Bell et al., 2005, 2007) have the capability to address the effects associated with flame geometry and the boundary conditions. The turbulent flame speed is just one of the many experimentally quantifiable features of the premixed turbulent flame that can be used to compare with the modeling or the computational results. Even if the choice of the integration pathway or the flame surface c contours is arbitrary, the comparison is meaningful as long as the same definition is applied to both the experiments and computations. Based on this reasoning, he proposed that the integration path for ST,LC be locally normal to the turbulent flame brush, as well as the use of the c = 0.5 contour to prescribe AT for the determination of ST,GC from a category of “envelop flames” that fully enclose the reactants, for example, a Bunsen flame. The ST,LC data reported in the combustion literature are still limited, and most of the results summarized by Driscoll (2008, Table 4) were inferred by assuming *

Driscoll focused on steady flames and ST,GD­ is not included in his review.

152

Synthesis Gas Combustion: Fundamentals and Applications

Normalized Burning Rate

14 12

SC from ∑ SC from m ST,LD

10 8 6 4 2 0

0

2

4

6 u´/SL

8

10

12

Figure 5.15  Comparison of the local displacement flame speed, ST,LD, the consumption flame speed based on the flame surface area integral and the consumption flame speed cal• culated based on mass balance m. (From Shepherd and Cheng, 2001. With permission.)

ST,LD = I0 ΣMax δT . One of the earlier attempts to integrate Σ through v-flames and stagnating flames produced results that were not very illuminating, as they all clustered around ST,LC /SL,0 of 1.6 to 1.8 regardless of flame configuration, fuel type, and equivalence ratio (Shepherd, 1996). Subsequent use of a low-swirl burner produced more meaningful results that show an increasing trend of ST,LC with u′ (Shepherd and Cheng, 2001). In this study, ST,LC were determined by integrating Σ derived from two-dimensional OH-PLIF images of four methane/air flames with u′/SL,0 varied from 2.5 to 7.5. To verify the physical significance of ST,LC, the mass consumption rate within the flame brush was also estimated by performing a conditional mass flux balance on the flow of reactants into and out of a control volume at the centerline. The ST,LC and the mass consumption rate were also compared with ST,LD measured at the centerline. All three sets of data are shown in Figure 5.15 (from Shepherd and Cheng, 2001) to illustrate that the ST,LC is directly comparable to the results from conditional mass flux balance. This confirms that ST,LC is a meaningful quantity that is consistent with the mean fuel consumption rate. In comparison, the ST,LD data are 3.75 times higher than the ST,LC data. The difference can be explained by the substantial radial outflow that occurred within the turbulent flame brush due to flow divergence. Consequently, only a small fraction of the reactants entering the leading edge of the flame brush at a velocity equal to ST,LD were consumed within the control volume. This provides a physical explanation as to why the two quantities are expected to be different in realistic systems. A much smaller ratio of ST,LD /ST,LC = 1.667 was reported by Lawn and Schefer (2006), who utilized a quartz diffuser cone to stabilize a series of normal flames burning CH4 and a fuel mixture consisting of 75% CH4 and 25% H2. The diffuser burner shares a common feature with the low-swirl burner, in that both configurations

Turbulent Combustion Properties of Premixed Syngas

153

exploit mean flow divergence to stabilize detached flames. The main difference is the diffuser cone produces a lower rate of divergence than the low-swirl burner. The cone also encloses the flame brush and imposes side boundary conditions on the flame and its flow field. Therefore, it seems reasonable to conjecture that the smaller ST,LD /ST,LC ratio in the diffuser burner is a consequence of the restricted radial outflow. The definition of the global consumption flame speed ST,GC (Equation 5.18) is an extension of the classic flame cone height method for measuring laminar flame speed in a Bunsen flame. It is valid only for envelope flames, where the reactants cannot escape without burning. Because it is defined against a mean flame surface area AT that is arbitrary in a thickened turbulent flame brush, there is no direct relationship between ST,GC and the local consumption flame speed, ST,LC. Therefore, ST,GC and ST,LC should be treated as different but equally acceptable ways to characterize the turbulent flame. ST,GC measured in axisymmetric Bunsen burners shows both linear (Smallwood et al., 1995) and bending (Kobayashi, 2002; Kobayashi et al., 2005) trends with increasing u′. Though the definitions used by the two groups are slightly different, with Smallwood et al. using the c = 0.05 contour for AT and Kobayashi et al. choosing c = 0.1, the difference seems to be too small to have caused the large discrepancy in the ST,GC trends. To find a physical explanation for the discrepancy, Filatyev et al. (2005) investigated ST,GC in a rectangular slot burner that produced two-dimensional envelop flames. Three turbulence generators were used to vary the turbulence intensity. To be consistent with Equation 5.14, the ST,GC was defined by using the c = 0.5 contour close to where Σmax occurs. These authors concluded that the bending and linear trends are associated with whether or not u′ and U0 are varied independently. If u′ is varied by using different turbulence generators, the ST,GC data obtained at the same U0 show significant bending. But if u′ is varied by increasing U0 as is done in most other studies, the set of ST,GC data at a relatively high turbulence level of u′/U0 = 0.2 shows a clear linear trend. Based on this observation, the authors argued that ST,GC in Bunsen-type flames is also dependent on U0 and the size of the burner, W. From the discussion on the various turbulent flame speeds, it is obvious that the processes controlling the heat release rate of a premixed turbulent flame cannot be encapsulated by a single parameter that can be applied unequivocally to different flame configurations. Therefore, acknowledging the need to use different definitions of the turbulent flame speeds for different flame configurations will help to scrutinize some of the confusing and most often contradictory findings. As pointed out by Driscoll (2008), models and numerical simulations have advanced to the point that their validation requires detailed statistical measurements of the velocity and the scalar fields to characterize the effects associated with the boundary conditions. The turbulent flame speed is just one of the many quantifiable flame properties that can be deduced from the experimental data for direct comparison. It is useful as long as a consistent definition is applied to the experiments and the computational results. This implies that the physical meaning of the turbulent flame speed is of a lesser significance. As to the application of the turbulent flame speed data in heat release models and in estimating the dynamic flame behavior such as flame flashback, the large inconsistencies in the results indicate that an all-inclusive flame geometry

154

Synthesis Gas Combustion: Fundamentals and Applications

neutral approach will not produce meaningful insights. A more useful approach is to scrutinize the data, and select those obtained in flames whose geometry is consistent with the burner configuration being considered. For industrial burners and gas turbine combustors that rely on flame holding by a bluff body or flow recirculation, only the turbulent flame speed data obtained in anchored flames are relevant. Similarly, data from expanding flame kernel experiments apply only to reciprocating engine analyses, and those from low-swirl burners to be used for burners and combustors that employ this flame stabilization method. Additionally, disregarding legacy turbulent flame speed data and selecting those that are accompanied by accurate characterizations of the initial boundary conditions and turbulence characteristics will help to produce more meaningful and consistent insights.

5.5 Effects of Syngas and Hydrogen on Turbulent Flame Speed The effects of syngas on the premixed turbulent flame processes are associated with H2, whose presence increases the turbulent flame speed and causes imbalance of heat and mass diffusion at the flame fronts. In their extensive review, Lipatnikov and Chomiak (2005) provided detailed descriptions of the physical mechanisms, and the theoretical and modeling treatments of the effects of molecular transport on premixed turbulent flames. The thermal/diffusive effects on stretch fact I0 are also discussed in the review of Driscoll (2008). Because syngas compositions vary significantly, these effects can be dominant in syngas flames with high H2 fuel concentration. On the other hand, the presence of inert diluents in syngases may enhance or counter the effects of H2. There are only a few basic studies of lean premixed turbulent flames using syngases, and many fundamental aspects remained unexplored. Therefore, the discussion in the section is focused on results from studies of lean H2 flames and flames burning H2 blended with other hydrocarbons. Hydrogen is much more reactive than hydrocarbons. Its laminar flame speed at stoichiometry is about five times higher than that of CH4. It is often assumed that their turbulent flame speeds would differ by the same amount. This is a concern to combustion engineers because it implies that the syngas flames have a much higher propensity to flashback. Experimental data on the turbulent flame speeds of H2 flames (Table  1 in Lipatnikov and Chomiak, 2005) show that they are generally higher than in hydrocarbon. But the increases have not been quantified because of the uncertainties and significant scatter in the data set. To gain some insights, the ST,LD results measured in normal flames, that is, those generated in stagnation burners and low-swirl burners, are examined here. As discussed in Section 5.4, ST,LD in these configurations are more consistent because they required the use of laser velocimetry methods to measure the velocity statistics. Liu and Lenze (1988) were the first to report ST,LD of H2/CH4 flames in a stagnation burner. Their objective was to investigate the effect of increasing SL on ST,LD. Changing the H2 fuel percentages from 0 to 56% enabled them to vary SL from 0.16 to 0.72 m/s. Their ST,LD data showed linear increases with u′ with no significant

155

Turbulent Combustion Properties of Premixed Syngas H2 0.5 H2/0.5 CO2 0.75 H2/0.25 N2 CH4 0.6 CH4/0.4 N2 0.5 CH4/0.5 CO2 0.6 CH4/0.4 H2 0.75 H2/0.25 CH4 0.5 CH4/0.5 H2 SG1 SG2 SG3

70 60

ST/SL

50 40 30

Correlation for H2 ST/SL = 1 + 3.15 u´/SL

Correlation for CH4 ST/SL = 1 + 1.73 u´/SL

20 10 0

0

5

10

15 u´/SL

20

25

30

Figure 5.16  ST,LD data for H2, CH4, and syngases. (From Cheng et al., 2009. With permission.) Syngas compositions are SG1 – 0.2 H2/0.4 CO/0.4 CH4, SG2 – 0.3 H2/0.3 CO/0.4 CO2, and SG3 – 0.6 H2/0.4 CO.

bending. The correlation coefficient, K in Equation 5.11, increased from 2.2 to 4 with increasing H2 concentration. Though these authors did not focus their study on the Lewis number effects, their results provided the first evidence to show that H2 has a direct effect on the ST,LD correlation. Goix and Sheperd (1993) used the stagnation flame configuration to investigate the Lewis number effects. Although they did not report ST,LD, their analysis of the flame surface area shows that the pure H2 flame has a larger area increase than the CH4, C2H2, and C3H8 flames, to suggest a higher ST,LD for H2. In a recent study, Littlejohn and Cheng (2007) investigated the fuel effects on a low-swirl injector* by comparing the flow fields and ST,LD of lean premixed turbulent flames burning several fuel blends of CO2 and N2 diluted hydrocarbons and a fuel blend of 0.6 CH4/0.4 CH4. The ST,LD data obtained for all mixtures were found to be consistent with the correlation ST,LD /SL = 1 + 1.73 (u′/SL) determined for CH4. But close examination of the two data points for the 0.6 CH4/0.4 H2 flames shows them to be higher than the CH4 results. In a subsequent paper, Cheng and Littlejohn (2008) extended their investigation to H2 and N2 diluted H2 flames and found a different ST,LD correlation coefficient K for H2. The data are shown in Figure 5.16, where the H2 flames have a higher correlation coefficient, K = 3.15, than the CH4 flames, K = 1.73. The higher K value for H2 flames is consistent with the trend shown by the results of Liu and Lenze. The most recent study in a low-swirl injector includes ST,LD data for three simulated syngases (Littlejohn et al., 2008). The few data points are *

The low-swirl injector is a version of the low-swirl burner of Bedat and Cheng (1995) that uses a vane swirler instead of tangential air jets to impart swirl.

156

Synthesis Gas Combustion: Fundamentals and Applications 30

+3

.15

L

u´/ SL C S H4 , T/ S C3 H L= 1 + 3,& 1.7 C2 3 u H4 ´/ S

H2 φ = 0.7 H2 φ = 0.9 CH4 φ = 0.7 CH4 φ = 0.9 CH4 φ = 0.98 C3H8 φ = 0.7 C3H8 φ = 1.14 C3H8 φ = 1.5

ST/SL

H ST /2 SL

=1

20

10

0

0

5

10

u´/SL

15

20

25

Figure 5.17  ST,GC for H2, CH4, and C3H8 from flame kernel experiments of Kido et al. (2002).

also plotted in Figure 5.16, and they are between those of CH4 and H2. The ST,LD data sets for each syngas show linear dependency on u′, but with different values of the correlation coefficient K. The results from the normal flame studies suggest that the primary effect of increasing H2 concentration in the fuel is to increase the ST,LD correlation constant K by about a factor of 2. More extensive turbulent flame speed data for H2 are available for the expanding flame kernel configuration. The study by Kido et al. (2002) is most often quoted because these investigators measured the turbulence intensity and length scales, while other studies inferred these quantities. The SL,GD was determined by examining the developments of the flame wrinkle structures on the lower half of the expanding flame kernels. Eight different H2, CH4, and C3H8 mixtures were used, all having SL,0 = 0.15 m/s. Their results are plotted in Figure 5.17 and show their magnitudes to be much lower than data from the low-swirl burner. For all mixtures, the nonlinear bending effect is clearly seen. The fastest-burning speeds are found for the lean H2 flames and for the rich C3H8 flames, and the results are consistent with the destabilizing Lewis number effects. Their rate of increase for the fastest-burning lean H2 mixture in the linear region corresponds to K ≈ 1, and the difference between the ST,GD /SL for the lean H2 and CH4 mixtures is about a factor of 2 at u′/SL < 10. Wu et al. (1990) used a Bunsen burner to study H2 jet flames of 0.8 < ϕ < 1.8 and found that the thermal/diffusive unstable flames at ϕ = 1.0 and 0.8 have higher ST,GC than the thermal/diffusive stable flames at ϕ = 1.8. The ST,GC for all mixtures show linear correlation with u′. The highest rate of increase in ST,GC with u′ for flames at ϕ = 1.0 correspond to K ≈ 1. This value is similar to the results of Kido et al. for lean H2 flames. Despite the significant differences in the experimental configurations and the methods of analyses, the most interesting implication of these studies is that the lean H2 turbulent flame speeds are all about twice as high as the lean CH4 flame speeds.

Turbulent Combustion Properties of Premixed Syngas

157

But the magnitudes of the turbulent flame speeds are different and strongly dependent on the flame configuration. Those in detached normal flames are about three times higher than in expanding flame kernels and Bunsen flames. The 2:1 turbulent flame speed ratio between the H2 and CH4 flames, if verified by more extensive studies, can be very useful for estimating the effects associated with the increasing H2 concentrations in syngases.

5.6 Premixed Turbulent Combustion Properties at High Temperatures and Pressures Many industrial burners for high-temperature applications operate at initial temperatures of up to 1300 K. Preheating the reactants is an effective way to increase system efficiency by recovering waste heat from the exhaust (Katsuki and Hasegawa, 1998). Heat recovery is done either by using a recuperator to preheat the air or by recirculating a portion of the flue gases into the combustion chamber. High temperature expands the combustion limits of the fuel mixtures and is particularly useful for utilizing weaker fuels, such as some syngases with high concentrations of inerts. Power generating equipment, such as gas turbines and internal combustion engines, operate at high temperatures and pressures. Their inlet conditions can be estimated by the adiabatic compression of air. At pressures of 10 to 20 atm, the corresponding air temperatures are 575 to 700 K. High pressures and temperatures affect premixed flames by altering the heat and molecular transport processes, and thus the laminar flame speed and the reaction zone thickness. Laboratory studies and theoretical analyses have shown that the laminar flame speed varies with temperature and pressure according to m



n

T  P S L = S L ,STP      T0   P0 

(5.19)

where SL,STP is the laminar flame speed at standard temperature and pressure (STP). For hydrocarbons, the exponent m is on the order of 2 and n is –0.5. These values show that SL decreases with pressure and increases with temperature. The two effects counter each other such that the SL at typical gas turbine conditions is about 18 to 23% higher than at STP. Studies also show that the Markstein length and the reaction zone thickness decrease with increasing pressure, implying that the premixed flame fronts are less stable at high pressures. Laboratory studies of stationary premixed turbulent flames at high temperatures and pressures require a substantial investment in equipment, infrastructure, and operational support. There are only a few facilities worldwide that are devoted to basic research, and equipped with optical access for the application of laser diagnostics. The most extensive body of work on stationary premixed turbulent flames at high temperatures and pressures is by Kobayashi and his collaborators (Kobayashi et al., 1997, 2002, 2005, 2007; Kobayashi and Kawazoe, 2000; Lee et al., 2000; Kobayashi, 2002). Their experimental setup consists of a pressurized chamber on

158

Synthesis Gas Combustion: Fundamentals and Applications 40

0.1 MPa 0.5 MPa 1.0 MPa 2.0 MPa 3.0 MPa Glder (0.1 MPa)

ST/SL

30

20

10

0

0

5

10

15

u´/SL

Figure 5.18  ST,GC for conical Bunsen CH4 flames at pressures 0.1 < P < 2.0 MPa. (From Kobayashi et al., 2002. With permission.)

the order of 1 m3 with a small Bunsen burner of 20 mm diameter placed within. The chamber volume is sufficiently large compared to the flame, so that the configuration is essentially an open flame in a pressurized and preheated environment. Their initial studies (Kobayashi et al., 1997; Kobayashi and Kawazoe, 2000; Lee et al., 2000; Kobayashi, 2002) were focused on the effects of pressure on CH4 and C3H8 flames at standard temperature. The bulk flow velocities of the experiments were relatively low at 2 < U0 < 3 m/s, with turbulence intensity at 10%. Hot-wire anemometry was used to characterize the turbulent nonreacting pressurized flow. The integral length scale and the turbulence intensities were found to decrease slightly with increasing pressure up to 3.0 MPa. The Schlieren images of the pressurized flames showed more wrinkled flame fronts and sharper flame cusps than flames at atmospheric pressure. Lee et al. (2000) investigated the pressurized flames with OH-PLIF, and quantified the pressure effects on increasing the degree of flame wrinkling and the flame surface density Σ . The ST,GC for the pressurized flames were deduced from the OH-PLIF images by the cone surface method, with the flame angle prescribed by the c = 0.5 contour (Kobayashi, 2002). The results (Figure 5.18) showed that pressure increases ST,GC /SL , and all data have the same characteristic bending as the STP flames. Kobayashi also observed an intrinsic instability behavior in the pressurized laminar flames. They hypothesized that this instability may explain the flames’ tendency to become more wrinkled when interacting with the turbulent eddies at high pressure. Kobayashi’s investigations of preheated and pressurized CH4/air flame were reported in two papers (Kobayashi et al., 2005, 2007). The experimental conditions were 2 < U0 < 3 m/s, 0.1 < P 1.0 MPa, and T = 293 and 573 K. Statistical details of the preheated and pressurized nonreacting turbulent flows were reported in the first paper. The turbulence spectra showed that the inertial subrange was not

Turbulent Combustion Properties of Premixed Syngas

159

altered by elevated pressure and temperature. The flame wrinkle scales obtained from OH-PLIF were analyzed using the fractal method. The analysis showed that the inner cutoff scale, εi, that is, the smallest relevant flame wrinkle size, correlated with the characteristic length scale of the intrinsic flame instability found on the pressurized laminar flames. The SL,GC data for the preheated and pressured flames were reported in the second paper. Here, the c = 0.1 contour was used to define the flame surface. The results confirmed the persistence of the bending effect at high temperature and pressure. An interesting trend that is shown by the data but was not elaborated in the paper is that the SL,GC obtained at the same pressure but at different temperatures are well correlated. This allowed the authors to propose a power law to correlate SL,GC for Bunsen flames that included only the pressure dependency: ST,GC /SL = K((P/P0)/ (u′/SL ))0.38. The exponent of 0.38 is the same for SL,GC data defined at the c = 0.1 and c = 0.5 contours. Therefore, the use of different contours changes only the constant K from 2.9 (for SL,GC defined at c = 0.5) to 5.04 (for ST,GC defined at c = 0.1). This is the first paper to show that the choice of the c contour affects the magnitude of the SL,GC data but not their overall correlation trend. Preheated and pressurized Bunsen flames at conditions relevant to gas turbines were investigated by Griebel et al. (2007). The diameter of the burner (25 mm) and the turbulence generators (perforated plates) was similar to the one used by Kobayashi. However, the flame was confined in a quartz enclosure (75 mm diameter and 325 mm in length) and stabilized by the outer recirculation zone formed at the sudden expansion plane at the entrance to the quartz enclosure. The preheat temperatures of 673 to 773 K and pressures at 0.5 to 1.44 MPa were similar to the conditions explored by Kobayashi. But the bulk velocity of 30 < U0 < 60 m/s was much higher, and the equivalence ratios of the CH4/air flames 0.43 < ϕ < 0.56 were lower. The inlet turbulence characteristics were measured by particle image velocity (PIV) in the isothermal flows, and OH-PLIF images of the flames were used to deduce the mean c contours for analyzing mean flame features such as flame length, flame brush thickness, and growth rate. Their results showed that pressure had no effect on the flame length and the flame brush thickness. Lowering ϕ from 0.56 to 0.43 elongated the flame by a factor of 2, but the increase in U0 from 30 to 60 m/s only elongated the flame by 11%. Due to the relatively small changes in mean flame properties with pressure, the ST,GC deduced from the c = 0.05 contour showed no pressure dependency. This conclusion is in direct contrast to the conclusion from the study by Koyayashi et al. Griebel et al. also reported a bending trend of SL,GC /SL with increasing u′/SL . Close inspection of the two sets of data shows that at the same u′/SL level, the Kobayashi SL,GC /SL data are twice as high as the data of Griebel et al. These two studies clearly show that the properties and behaviors of pressurized premixed turbulent flames produced by the same type of burner can be very different. Therefore, the insights gained from each study cannot be generalized and applied until the causes of the differences are identified. Halter et al. (2007) investigated the effects of H2 addition (20%) to CH4 Bunsen flames at low velocities and pressures without preheat (0.1 < P < 0.5 MPa and U0 = 2.1 m/s). The objective was to characterize the changes in Σ and ST,GC . They found

160

Synthesis Gas Combustion: Fundamentals and Applications

that the flame becomes shorter when H2 is added. But the SL,GC and the magnitudes of Σ showed only slight increases. The ST,GC data also indicated pressure dependent, as in Kobayashi’s studies, but these authors did not vary u′ to investigate its correlation trend. Recently, Cheng et al. (2009) studied a low-swirl injector (LSI) with H2 and natural gas at STP and at gas turbines conditions (0.1 < P < 0.8 MPa, 298 < T < 580 K, 18 < U0 < 60 m/s). The flow fields and the ST,LD of the flames at STP were measured by PIV. The preheated and high-pressure experiments were performed to determine the effect of H2 on flame position, flashback limit, and NOX emissions. As discussed above, the LSI generates a detached flame that propagates within a divergent flow produced at the near-field region adjacent to the nozzle exit. The flame lift-off position, xf , is a critical parameter that is unique to this configuration. From PIV studies of CH4/air flames at STP, it has been shown that xf is not sensitive to U0 and ϕ when U0 is much larger than SL (Cheng et al., 2008). This is associated with the coupling between the self-similar divergent flow and the linear correlation of ST,LD with u′, which can be expressed in a simple analytical model. As illustrated by the flame luminosity images of Figure 5.19, the LSI flames position at high inlet temperatures and pressures exhibit the same invariant behavior. However, xf was found to decrease with increasing H2 fuel concentration. This a consequence of the higher ST,LD correlation constant K for the H2-rich flames. The xf data for the STP flames in Figure 5.19 illustrate that their magnitudes and trend are consistent with the model prediction. The xf data obtained at gas turbine conditions also show the same trend. Moreover, the values of xf and the extent of flame shift are the same as the STP flames. This implies that the coupling between the self-similar divergent near-field and the linear ST,LD dependency on u′ is also relevant at gas 20 m/s 0.8 MPa

50

40 m/s 0.8 MPa

60 m/s 0.8 MPa

40

x*f mm

30

20

STP 1 atm 2 atm 4 atm 8 atm Model

10

0

0

20

40

60

80

100

H2% in CH4

Figure 5.19  Effects of H2 on flame brush positions, xf, of a low-swirl injector at STP and gas turbine conditions, and comparison with model prediction. (From Cheng et al., 2009. With permission.)

Turbulent Combustion Properties of Premixed Syngas

161

turbine conditions. In view of the highly nonlinear nature of the flame and turbulent flow processes, the consistent behavior of these flames at STP, and at high temperatures and pressures, is quite unexpected. The analytical model can provide a foundation to gain more insights into the processes at high temperatures and pressures. The laboratory studies discussed in this section show that the knowledge on premixed turbulent flames at high temperatures and pressures is still very limited. It is premature to make general conclusions on how changes in fuel compositions, especially high H2 concentration, alter the behaviors of preheated and pressurized premixed turbulent flames. One area of research that will be very beneficial is the measurements of the flame flow fields. For example, Griebel et al. (2007) commented that their flames were most likely controlled by shear-layer-type turbulence and not by the grid turbulence through the burner nozzle. If this can be verified by velocity measurements, it would help to explain why their results are different than those from Kobayashi et al. and Halter et al. In the case of the low-swirl injector, direct measurement of SL,LD for preheated and pressurized flames will provide the much needed data for comparison with the STP results, and to verify the analytical model.

5.7 Modeling Considerations for Syngas and Hydrogen Flames The discussion in Section 5.4 illustrated that a primary effect of the thermal/diffusive imbalance in lean H2 flames is the destabilization of the flame fronts that increases the degree of flame wrinkling. Due to the high diffusivity of H2, the wrinkling process also causes significant changes in the flame front structures and their local fuel consumption rate. An indication of this phenomenon is shown by the OH-PLIF image of lean H2 flame of Figure 5.8, where the OH concentrations are higher in the convex regions of the wrinkled flame than in the concave regions. In contrast, the OH concentrations are more uniformly distributed on the thermal/diffusive neutral CH4 flame and the thermal/diffusive stable C3H8 flame. Lewis number effects on the wrinkled flame fronts are mostly explored by direct numerical simulations (DNSs) (Chen and Im, 2000; Im and Chen, 2002), because measuring the local fuel consumption rate and the inner structures of the thin wrinkled flames presents major experimental challenges. In a recent paper, Bell et al. (2006) conducted two-dimensional time-dependent DNS studies of lean H 2, CH4, and C3H8 flames using detailed chemical kinetic schemes (Figure 5.20). The use of adaptive mesh refinement enabled them to increase the size of their computational domains to the physical dimensions of the laboratory experiments, so the results are directly comparable. The DNS results of the local fuel consumption rates, sc, for the three mixtures within a subregion of the computational domain (1.2 cm in width) are shown below the OH-PLIF images in Figure 5.20. The numerical results show that the fuel consumption rate is highly nonuniform on the H2 flame, with gaps appearing in the concave regions. This feature is very different than those of the CH4 and the C3H8 flames, where the fuel consumptions rates have no discontinuities.

162

Synthesis Gas Combustion: Fundamentals and Applications (a) Propane

(b) Methane

(c) Hydrogen

OH-PLIF (a) Propane

(b) Methane

(c) Hydrogen

DNS Local Heat-Release Rate (a) Propane

2.5

2.0

(b) Methane

2.5

Scloc

Scloc

1.5

1.5

1.5

1.0

1.0

1.0

0.5

0.5

0.5

0.0

–4

–2

0

2

4

0.0

(c) Hydrogen

2.0

2.0

Scloc

2.5

–4

–2

0

2

4

0.0

DNS Local Fuel Consumption Speed

–4

–2

0

2

4

Figure 5.20  OH-PLIF images (top), simulated local heat release rates (middle), and crosscorrelation of local fuel consumption speed with local flame curvature (bottom) on the flame fronts of lean propane (left), methane (middle), and hydrogen (right) turbulent premixed flames. (From Bell et al., 2006. With permission.)

Correlations of the local sc with the mean flame curvature for the three flames are shown at the bottom row of Figure 5.20. In the case of the thermal/diffusively stable C3H8 flame, sc correlates negatively with the mean curvature, with higher values in the negative concave regions and lower values in the convex regions. The sc of the CH4 flames has a very weak negative correlation with mean curvature because it is thermal/diffusively neutral. In contrast, sc of the H2 flame is very sensitive to the mean curvature with a strong positive correlation. At zero mean curvature, that is, a locally flat flame, sc has a very large range of values. Toward the large negative mean curvatures, the sc values are locally very close to zero or extinct. The mechanism that caused the extinction events in the H2 flame were also examined, by comparing the H2 concentrations along a flowpath through an extinction point and an adjacent flowpath through a strongly burning region. For the flowpath going through the extinction point, 97% of the H2 is diffusively transported out of the fluid and into the adjacent flowpaths, where burning occurs at a higher fuel/air ratio.

Turbulent Combustion Properties of Premixed Syngas

163

Therefore, the extinction event is not due to heat loss associated with high strain, but to depletion of fuel by preferential diffusion. This mechanism is very different than the classic approach to characterize the local variations in the fuel consumption rate by the Markstein number, which only represents the response of the flame to curvature and strain. Consequently, the stretched flamelet model is not valid for lean H2 mixtures or lean syngases with high H2 fuel constituencies. Chen and Bilger (2004) also found “superadiabaticity” in the convex regions of the wrinkled lean H2/air flames generated in a Bunsen burner. The authors applied combined two-dimensional OH-PLIF and two-sheet Rayleigh scattering method to resolve the three-dimensional flame front orientations, and the three-dimensional scalar dissipations rates at the fronts. The objective was to evaluate the parameters in the conditioned moment closure methods for premixed turbulent combustion (Klimenko and Bilger, 1999). Their results show that the local OH mole fractions correlate with the mean flame curvature, but not with the local gradient of the progress variable. These findings suggest a change in the internal flame front structures. They also reported that the conditional mean progress-variable dissipation rates were higher than the value for an unstretched laminar flame. The higher than expected dissipation rate or “superadiabaticity” is consistent with the DNS results, showing high sc at the planar region of the wrinkled flames. In their review of the molecular transport effects on premixed turbulent flames, Lipatnikov and Chomiak (2005) concluded that the stretched flamelet library concept has not yet been shown to be effective in predicting the Lewis number effects. They argued that the important physical mechanisms of premixed turbulent flame propagation are associated with the processes at the leading edge of the flame brush. The rationale is that the flame brush is sustained at the leading edge by the “critically curved flames,” that is, flames that can withstand the turbulence disturbances without quenching, and by the flame fronts that propagate into vortex tubes. This concept seems to be consistent with the processes of lean H2 flames at Le < 1, where burning is more intense at the convex parts close to leading edges, while burning at the concave trailing edge is strongly affected by processes at the leading edge. As this and other concepts are still evolving, the question on the mathematical model for lean H2 premixed turbulent flame processes remains open.

5.8 Conclusions The framework for theoretical analysis and modeling of the premixed turbulent flames is based on comparing the turbulence length and time scales with the characteristic length and time scales of premixed laminar flames. This is the foundation for the flame regime concept from which the wrinkled flamelet regime has been shown to be valid for hydrocarbon flames in most practical situations. The turbulent flame is perceived to consist of wrinkled flamelets that are curved laminar flames being stretched by the turbulent eddies. For lean hydrocarbon flames, stretch induces slight modification of the local flame speeds and fuel consumption rates of the wrinkled flamelets, without inducing significant changes to their inner structures. The degree of wrinkling is, to the first order, directly proportional to the mean heat release rate and the turbulent flame speed. While the theoretical and experimental studies of

164

Synthesis Gas Combustion: Fundamentals and Applications

stretched laminar flames have made significant progress, many basic issues concerning the relationships among the turbulence intensity, mean reaction rate, and turbulent flame speed remain unresolved. This is due to the significant influences of the burner geometry, and also the initial and boundary conditions on the turbulenceflame interactions. Consequently, universal premixed turbulent flame behaviors should not be expected. Understanding how the premixed turbulent flame characteristics change with flame geometry and boundary conditions is important to the interpretation and analysis of experimental data. For certain cases, e.g. the Bunsen-type burners and the low-swirl burner, the application of flame geometry specific analyses has shown to produce more consistent results. The presence of hydrogen in syngases introduces a multitude of complexities to the premixed turbulent flame processes. The flame speeds of syngases with high hydrogen concentrations are faster than those of hydrocarbons. In addition, some syngas flames may be thermal/diffusively unstable at lean conditions. These processes are also influenced by the presence of other fuel species and by heavy diluents. Therefore, much of the knowledge on stretched and wrinkled flamelets for hydrocarbon fuels may not be relevant to many syngases. The overall effects of hydrogen in syngases can be inferred from experimental and computational studies of premixed turbulent flames burning blended hydrogen-hydrocarbon fuels and pure hydrogen. The results from several flame configurations all seem to indicate that the turbulent flame speeds of lean hydrogen flames are about a factor of 2 faster than those of hydrocarbon flames, even though their magnitudes are significantly different for different flame geometries. A few lean syngas turbulent flame speed data are found to be between those of methane and hydrogen. Direct numerical simulations show that lean hydrogen flames are more wrinkled than the hydrocarbon flames and have locally higher fuel consumption rates due to the thermal/diffusive instabilities. The highly nonuniform fuel consumption rates imply that the stretched flamelet model is not applicable to the lean hydrogen premixed turbulent flames. But alternate modeling approaches have yet to be developed. Due to the large variations in the syngas fuel compositions, their flame properties under lean conditions will change from thermal/diffusively neutral to thermal/diffusively unstable with increasing hydrogen fuel concentration. These changes present significant challenges to modeling of syngas combustion in industrial burners and power systems. Therefore, research on thermal/diffusive effects on laminar and turbulent syngas flames will be necessary to characterize and resolve these complex issues. Experimental research on fundamental properties of premixed turbulent flames at high temperatures and pressures is rare due to the scarcity of the facilities. Studies of CH4 Bunsen-type flames conducted at low velocities showed that the flames at elevated pressures become more wrinkled, and their turbulent flame speeds increase. However, the turbulent flame speeds measured in a similar flame configuration with a tight enclosure at velocities close to gas turbine conditions do not show pressure dependency. A study in a low-swirl injector with natural gas and hydrogen found consistent flame behavior at STP and at high temperatures and pressures. The difference in these findings shows, again, the dominant effects of flame geometry and boundary conditions that should be carefully considered when interpreting the results. At present, the basic properties of premixed turbulent flames at heated and

Turbulent Combustion Properties of Premixed Syngas

165

pressurized conditions remain largely unexplored. There is an urgent need for data on the turbulent flame speeds and other scalar measurements to quantify the topology of the flamelets, the internal flamelet structures, and the heat release rates. These experiments are very difficult to perform. But with the proper choice of flame geometry having well-defined boundary conditions, and consistent analysis methods, even a small amount of data will lead to useful insights into the properties and behaviors of natural gas and syngas flames relevant to heating and power systems.

Acknowledgment Support of this work was provided by the U.S. Department of Energy, Fossil Energy, under Contract DE-AC03-76F00098.

References Abdel-Gayed, R. G., Bradley, D., and Lawes, M. (1987). Turbulent burning velocities: A general correlation in terms of straining rates. Proc. R. Soc. London A 414:389. Abdel-Gayed, R. G., Bradley, D., and Lung, F. K.-K. (1989). Combustion regimes and the straining of turbulent premixed flames. Comb. Flame 76:213. Bedat, B., and Cheng, R. K. (1995). Experimental study of premixed flames in intense isotropic turbulence. Comb. Flame 100:485. Bell, J. B., Cheng, R. K., Day, M. S., and Shepherd, I. G. (2006). Numerical simulation of Lewis number effects on lean premixed turbulent flames. Proc. Comb. Inst. 31:1903. Bell, J. B., Day, M. S., and Grcar, J. F. (2002). Numerical simulation of premixed turbulent methane combustion. Proc. Comb. Inst. 29:1987. Bell, J. B., Day, M. S., Grcar, J. F., Lijewski, M. J., Driscoll, J. F., and Filatyev, S. A. (2007). Numerical simulation of a laboratory-scale turbulent slot flame. Proc. Comb. Inst. 31:1299. Bell, J. B., Day, M. S., Shepherd, I. G., Johnson, M. R., Cheng, R. K., Beckner, V. E., Lijewski, M. J., and Grcar, J. F. (2005). Numerical simulation of a laboratory-scale turbulent V-flame. Proc. Natl. Acad. Sci. USA 102:10006. Bilger, R. W., Pope, S. B., Bray, K. N. C., and Driscoll, J. F. (2005). Paradigms in turbulent combustion research. Proc. Comb. Inst. 30:21. Borghi, R. (1985). On the structures and morphology of turbulent premixed flames. In Recent advances in the aerospace sciences, ed. F. Casci, 117–138. Plenum Press: New York. Bradley, D. (1992). How fast can we burn? Comb. Inst. Proc. 24:279–285. Bray, K. N. C., and Cant, R. S. (1991). Some application of Kolmogorov turbulence research in the field of combusiton. Proc. R. Soc. London A 434:217. Chan, C. K., Lau, K. S., Chin, W. K., and Cheng, R. K. (1992). Freely propagating open premixed turbulent flames stabilized by swirl. Proc. Comb. Inst. 24:511. Chen, J. H., and Im, H. G. (2000). Stretch effects on the burning velocity of turbulent premixed hydrogen/air flames. LBNL Report 28:211. Chen, Y.-C., and Bilger, R. W. (2002). Experimental investigation of three-dimensional flamefront structure in premixed turbulent combustion. I. Hydrocarbon/air Bunsen flames. Comb. Flame 131:400. Chen, Y.-C., and Bilger, R. W. (2004). Experimental investigation of three-dimensional flamefront structure in premixed turbulent combustion. II. Lean hydrogen/air Bunsen flames. Comb. Flame 138:155. Cheng, R. K., and Gouldin, F. C. (2004). Experimental database for premixed turbulent flames. International Workshop on Premixed Turbulent Flames. http://eetd.lbl.gov/aet/­ combustion/workshop/database/flame-config.html.

166

Synthesis Gas Combustion: Fundamentals and Applications

Cheng, R. K., and Littlejohn, D. (2008). Laboratory study of premixed H2-air & H2-N2-air flames in a low-swirl injector for ultra-low emissions gas turbines. J. Eng. Gas Turb. Power 130:31503. Cheng, R. K., Littlejohn, D., Nazeer, W. A., and Smith, K. O. (2008). Laboratory studies of the flow field characteristics of low-swirl injectors for application to fuel-flexible turbines. J. Eng. Gas Turb. Power 130:21501. Cheng, R. K., Littlejohn, D., Strakey, P., and Sidwell, T. (2009). Laboratory investigations of low-swirl injectors with H2 and CH4 at gas turbine conditions. Proc. Comb. Inst. 32:3001–3009. Cheng, R. K., and Shepherd, I. G. (1991). The influence of burner geometry on premixed turbulent flame propagation. Comb. Flame 85:7. Cheng, R. K., Shepherd, I. G., Bedat, B., and Talbot, L. (2002). Premixed turbulent flame structures in moderate and intense isotropic turbulence. Comb. Sci. Tech. 174:29. Cho, P., Law, C. K., Hertzberg, J. R., and Cheng, R. K. (1986). Structures and propagation of turbulent premixed flames stabilized in a stagnation flow. Int. Comb. Symp. 21:1493. Damkohler, G. (1940). The effects of turbulence on the flame velocity in gas mixtures. Elektrochem. 46:610. Driscoll, J. F. (2008). Turbulent premixed combustion: Flamelet structure and turbulent burning velocities. Prog. Energy Comb. Sci. 34:91–134. Duclos, J. M., Veynante, D., and Poinsot, T. (1993). A comparison of flamelet models for premixed turbulent combustion. Comb. Flame 95:101. Filatyev, S. A., Driscoll, J. F., Carter, C. D., and Donbar, J. M. (2005). Measured properties of turbulent premixed flames for model assessment, including burning velocities, stretch rates, and surface densities. Comb. Flame 141:1. Goix, P. J., and Sheperd, I. G. (1993). Lewis number effects on turbulent premixed flame structure. Comb. Sci. Tech. 91:191. Griebel, P., Siewert, P., and Jansohn, P. (2007). Flame characteristics of turbulent lean premixed methane/air flames at high pressure: Turbulent flame speed and flame brush thickness. Proc. Comb. Inst. 31:3083. Halter, F., Chauveau, C., and Gokalp, I. (2007). Characterization of the effects of hydrogen addition in premixed methane/air flames. Int. J. Hydrogen Energy 32:2585. Hawkes, E. R., and Cant, R. S. (2001). Implications of a flame surface density approach to large eddy simulation of premixed turbulent combustion. Comb. Flame 126:1617. Hawkes, E. R., and Chen, J. H. (2006). Comparison of direct numerical simulation of lean premixed methane-air flames with strained laminar flame calculations. Comb. Flame 144:112. Hinze, H. (1959). Turbulence. New York: McGraw-Hill. Im, H. G., and Chen, J. H. (2002). Preferential diffusion effects on the burning rate of interacting turbulent premixed hydrogen-air flames. Comb. Flame 131:246. Katsuki, M., and Hasegawa, T. (1998). The science and technology of combustion in highly preheated air. Proc. Comb. Inst. 27:3135. Kido, H., Nakahara, M., Nakashima, K., and Hashimoto, J. (2002). Influence of local flame displacement velocity on turbulent burning velocity. Proc. Comb. Inst. 29:1855. Klimenko, A. Y., and Bilger, R. W. (1999). Conditional moment closure for turbulent combustion. Prog. Energy Comb. Sci. 25:595. Kobayashi, H. (2002). Experimental study of high-pressure turbulent premixed flames. Exp. Therm. Fluid Sci. 26:375. Kobayashi, H., Hagiwara, H., Kaneko, H., and Ogami, Y. (2007). Effects of CO2 dilution on turbulent premixed flames at high pressure and high temperature. Proc. Comb. Inst. 31:1451. Kobayashi, H., Kawahata, T., Seyama, K., Fujimari, T., and Kim, J.-S. (2002). Relationship between the smallest scale of flame wrinkles and turbulence characteristics of highpressure, high-temperature turbulent premixed flames. Proc. Comb. Inst. 29:1793.

Turbulent Combustion Properties of Premixed Syngas

167

Kobayashi, H., and Kawazoe, H. (2000). Flame instability effects on the smallest wrinkling scale and burning velocity of high-pressure turbulent premixed flames. Proc. Comb. Inst. 28:375. Kobayashi, H., Nakashima, T., Tamura, T., Maruta, K., and Niioka, T. (1997). Turbulence measurements and observations of turbulent premixed flames at elevated pressures up to 3.0 MPa. Comb. Flame 108:104. Kobayashi, H., Seyama, K., Hagiwara, H., and Ogami, Y. (2005). Burning velocity correlation of methane/air turbulent premixed flames at high pressure and high temperature. Proc. Comb. Inst. 30:827. Kostiuk, L. W., Bray, K. N. C., and Cheng, R. K. (1993). Experimental study of premixed turbulent combustion in opposed streams. 2. Reacting flow field and extinction. Comb. Flame 92:396. Law, C. K., and Sung, C. J. (2000). Structure, aerodynamics, and geometry of premixed flamelets. Prog. Energy Comb. Sci. 26:459. Lawn, C. J., and Schefer, R. W. (2006). Scaling of premixed turbulent flames in the corrugated regime. Comb. Flame 146:180. Lee, G. G., Huh, K. Y., and Kobayashi, H. (2000). Measurement and analysis of flame surface density for turbulent premixed combustion on a nozzle-type burner. Comb. Flame 122:43. Libby, P. A., and Williams, F. A. (1994). Turbulent reacting flows. London: Academic Press. Lipatnikov, A., and Chomiak, J. (2005). Molecular transport effects on turbulent flame propagation and structure. Prog. Energy Comb. Sci. 31:1. Lipatnikov, A. N., and Chomiak, J. (2002). Turbulent flame speed and thickness: Phe­nom­ e­nology, evaluation, and application in multi-dimensional simulations. Prog. Energy Comb. Sci. 28:1. Littlejohn, D., and Cheng, R. K. (2007). Fuel effects on a low-swirl injector for lean premixed gas turbines. Proc. Comb. Inst. 31:3155. Littlejohn, D., Cheng, R. K., Noble, D. R., and Lieuwen, T. (2009). Laboratory investigations of low-swirl injector operating with syngases. J. Eng. Gas Turb. Power, in press. Liu, Y., and Lenze, B. (1988). The influence of turbulence on the burning velocity of premixed flames at different laminar burning velocities of CH4-H2 mixtures. Proc. Comb. Inst. 11:747. Peters, N. (1986). Laminar flamelet concepts in turbulent combustion. Proc. Comb. Inst. 25:1231. Peters, N. (1999). The turbulent burning velocity for large-scale and small-scale turbulence. J. Fluid Mech. 382:101. Peters, N. (2000). Turbulent combustion. Cambridge, UK: Cambridge University Press. Pitsch, H., and Duchamp de Lageneste, L. (2002). Large-eddy simulation of premixed turbulent combustion using a level-set approach. Proc. Comb. Inst. 29:2001. Poinsot, T., Veynante, D., and Candel, S. (1990). Diagrams of premixed turbulent combustion based on direct simulation. Paper presented at the Twenty Third Symposium (International) on Combustion, Orleans, France. Shepherd, I. G. (1996). Flame surface density and burning rate in premixed turbulent flames. Proc. Comb. Inst. 26:373. Shepherd, I. G., and Cheng, R. K. (2001). The burning rate of premixed flames in moderate and intense turbulence. Comb. Flame 127:2066. Shepherd, I. G., Cheng, R. K., Plessing, T., Kortschik, C., and Peters, N. (2002). Premixed flame front structure in intense turbulence. Proc. Comb. Inst. 29:1833. Shy, S. S., I, W. K., and Lin, M. L. (2000). A new cruciform burner and its turbulence measurements for premixed turbulent combustion study. Exp. Therm. Fluid Sci. 20:105. Smallwood, G. J., Gulder, O. L., Snelling, D. R., Deschamps, B. M., and Gokalp, I. (1995). Characterization of flame front surfaces in turbulent premixed methane/air combustion. Comb. Flame 101:461.

168

Synthesis Gas Combustion: Fundamentals and Applications

Smith, K. O., and Gouldin, F. C. (1979). Turbulence effects on flame speed and flame structure. AIAA J. 17:1243. Vervisch, L., and Veynante, D. (2000). Interlinks between approaches for modeling turbulent flames. LBNL Rep. 28:175. Williams, F. A. (1985). Combustion theory. Addison-Wesley: Reading, MA. Wu, M. S., Kwon, S., Driscoll, J. F., and Faeth, G. M. (1990). Turbulent premixed hydrogen air flames at high Reynolds-numbers. Comb. Sci. Tech. 73:327.

Formation 6 Pollutant and Control Kevin J. Whitty, Hongzhi R. Zhang, and Eric G. Eddings Contents 6.1 Introduction................................................................................................... 169 6.2 Nitrogen Oxides (NOX).................................................................................. 171 6.2.1 NOX Formation Mechanisms............................................................. 171 6.2.2 Effect of Combustion Conditions on NOX Emissions........................ 173 6.2.3 NOX Control Technologies................................................................. 175 6.3 Sulfur Species................................................................................................ 179 6.4 Carbon Monoxide.......................................................................................... 180 6.5 Volatile Organic Compounds........................................................................ 182 6.6 Trace Elements.............................................................................................. 185 6.7 Particulate Matter.......................................................................................... 186 6.8 Carbon Dioxide.............................................................................................. 187 6.9 Conclusions.................................................................................................... 188 References............................................................................................................... 189

6.1 Introduction As with any fuel, the combustion of syngas can produce gaseous pollutants such as nitrogen oxides (NOX), sulfur dioxide (SO2), carbon monoxide (CO), volatile organic compounds (VOCs), particulate matter, and trace species such as mercury and other metals. The amount of emissions generated depends on the properties of the syngas as well as the type and operating conditions of the combustor. Currently, the most common use of syngas in combustion systems is for production of power in gas turbines in an integrated gasification combined cycle (IGCC) arrangement, with the syngas resulting from gasification of coal, petcoke, oil, or other fossil fuels. Many IGCC systems exist today, and available data indicate that emissions from IGCC systems are generally less than those from conventional combustors such as pulverized coal (PC)–fired boilers and circulating fluidized bed (CFB) boilers (Table 6.1). A compilation of published emissions data from several IGCC systems is presented in Table 6.2, and shows consistently low values for emissions of key pollutants.

169

170

Synthesis Gas Combustion: Fundamentals and Applications

Table 6.1 Typical Emission Data in lb/MMBtu from Power Plants of Different Technologies Technology

NOX

SO2

Particulate

CO

Mercury

PC-fired plant (with advanced pollution controls) Atmospheric CFB (with SNCR) IGCC

<0.15 0.09 0.09

0.2 0.4 0.08

<0.03 0.011 <0.015

0.12 0.11 0.04

4.4 × 10–6 4.0 × 10–6 3.2 × 10–6

SNCR = selective noncatalytic reduction. Source: Ratafia-Brown et al. (2002); Amick (2004).

Table 6.2 Emissions from IGCC Power Plants IGCC Plant and Location TECO Polk Power Station, Florida Wabash River, Indiana NUON/Demkolec, Buggenum, Netherlands ELCOGAS, Puertollano, Spain Dow Chemical LGTI, Louisiana (1987–1995) Texaco Cool Water, California) 1984–1988)

NOX SO2 Particulate CO VOC Mercury lb/MMBtu lb/MMBtu lb/MMBtu lb/MMBtu lb/MMBtu lb/MMBtu 0.04

0.12

<0.004

n/a

n/a

5.2 × 10–6

0.103 0.085

0.13 0.053

0.011 0.001

0.045 n/a

0.002 n/a

3.2 × 10–6 n/a

0.11

0.018

0.005

n/a

n/a

n/a

n/a

n/a

1.7 × 10–6

0.004

n/a

n/a

0.26

<0.15

0.07

0.07

<0.01 0.009

Source: U.S. DOE (2002); Amick (2004); Hornick (2006).

Although gas turbines burn more syngas than any other type of combustion system, use of waste- or biomass-derived syngas as a substitute for natural gas, coal, or oil in boilers and reciprocating engines has become more common, driven by both economics and a desire to reduce use of fossil fuels. Often, these combustion systems serve a single facility and are smaller, nonutility-scale systems with a total thermal input of less than 50 MW. Published data on emissions from syngas combustion in these types of systems are limited and difficult to summarize due to variations in syngas production systems, feedstock, extent and nature of syngas cleaning, and combustor configuration and operation. However, based on well-established understanding of combustion chemistry and results from laboratory experiments, it is possible to estimate pollutant formation during combustion of syngas. The following sections discuss different classes of pollutants, their production, and control methods.

Pollutant Formation and Control

171

6.2 Nitrogen Oxides (NOX) 6.2.1 NOX Formation Mechanisms NOX is a collective term to describe several oxides of nitrogen that are important in combustion systems. Nitric oxide (NO) is generally the most commonly emitted NOX species and poses minimal direct health or environmental risk at the concentrations found in the environment. When NO is released to the environment, however, it can be oxidized to nitrogen dioxide (NO2), primarily through photochemical reactions that also involve VOCs. This secondary oxidation reaction is slow relative to the primary NO formation reactions (that occur during combustion), with a time constant on the order of days (Seinfeld and Pandis, 1998). Nitrogen dioxide (NO2) is generally emitted in only small percentages from combustion devices; however, it poses a much more serious health and environmental risk than NO. One of the dominant risks associated with either direct NO2 emissions or production via secondary oxidation reactions is the formation of ozone due to reaction of NO2 with VOCs. Nitrous oxide (N2O) is generally a minor constituent of NOX except in certain combustion devices that operate at lower temperatures, such as fluidized bed combustors. One of the primary concerns with N2O emissions is the recognition that it is a significant greenhouse gas, with a global warming potential 298 times that of carbon dioxide (Houghton et al., 2001; IPCC, 2007). The primary NOX species of interest during the combustion of syngas is NO, and subsequent discussion will focus on NO formation chemistry. NO emissions during syngas combustion result primarily from two different mechanisms: fuel NO formation and thermal NO formation. A third mechanism, prompt NO formation, is responsible for a considerably smaller fraction of the overall NOX emissions, unless emissions levels are very low. Each of these mechanisms will be briefly described below. Fuel NO results from the oxidation of nitrogen-bearing species that have evolved from the fuel during gasification, and these species are primarily HCN and NH3. These species are readily converted to NO in the presence of oxygen at combustion temperatures, and thus can become significant sources for NO if the nitrogenbearing precursors are present in sufficient quantity. The precursor species can also be reduced to N2 under oxygen-starved conditions (Miller and Bowman, 1989), given sufficient residence time and temperature, and many control techniques for the minimization of fuel NO take advantage of this effect, as discussed later in this chapter. A simplified schematic of fuel nitrogen conversion is shown in Figure 6.1. Thermal NO arises from the oxidation of molecular nitrogen (N2) in the combustion air. The fundamental steps in this oxidation process are given by the extended Zeldovich mechanism (Bozzelli and Dean, 2000), as illustrated by the following three reactions:

O + N2 ↔ NO + N

(6.1)



N + O2 ↔ NO + O

(6.2)



N + OH ↔ NO + H

(6.3)

172

Synthesis Gas Combustion: Fundamentals and Applications

Fuel N

NH3 or HCN

OH

NHi

OH,

Fuel NO reactions

O2

NO

Prompt NO-N2 reacting with fuel fragments to form fuel N species

NO N2

O2, OH, O

NO

Thermal NO-N2 reacting with oxygen at high temperature

Figure 6.1  Simplified schematic of NO formation mechanisms.

These reactions involve radical species (O, N, H, OH) that are initially formed through decomposition or abstraction reactions. Due to the inherent stability of the N2 molecule, considerable energy is required to oxidize N2, and thus thermal NO is only formed in appreciable quantities at elevated combustion temperatures (>1370°C, 2500°F). If an oxy-fired combustion system is used, the formation of thermal NO can in principle be reduced, as most of the N2 is removed upstream of the combustor in an air separation device. However, practical systems will rarely use 100% oxygen, due to economic considerations, and thus some N2 will still be present (typically 2 to 5%). Prompt NO is thought to be formed by the reaction of N2 with fuel fragments (e.g., CH radical) in the very early part of the flame (Miller and Bowman, 1989). This rapid formation early in the flame is what gives rise to the label “prompt” NO. Recent studies (Bozzelli and Dean, 1995; Konnov et al., 2000), however, have indicated the possibility of a different radical, NNH, formed via reactions of N2 with a hydrogen radical (H), that may play a significant role in NO formation. Harrington et al. (1996) measured NO in hydrogen flames, under conditions where temperatures were low enough to preclude significant thermal NO formation, which provided evidence of the viability of a nonhydrocarbon route for prompt NO formation. Additional studies by Konnov and coworkers (2001), in which they compared kinetic calculations with experiments in a well-stirred reactor, focused on the conditions under which the NNH pathway is significant. Their findings indicated that this route was an important contributor to NOX formation over a wide temperature range (1500 to 2200 K), but primarily at short residence times (<1 ms). At longer residence times, particularly at the highest temperatures (>2100 K), the contributions due to thermal NO become significant. These studies provide evidence for multiple routes for prompt NO formation, and the particular route will depend on fuel composition. Although prompt NO reactions occur very quickly in the flame, only small amounts of prompt NO are typically formed (on the order of 10 to 20 ppm). However, since NOX emissions from modern gas turbine installations, or other systems in which syngas could be utilized, are now commonly regulated to very low NOX levels (5 to 10 ppm in many regions), turbine designs and control strategies must clearly address the contributions due to prompt NO. The implications for syngas combustion, as compared to conventional natural gas combustion, are that NO emissions will likely be dominated by formation through the NNH route, due to the high H2 concentrations and relatively low hydrocarbon content.

Pollutant Formation and Control

173

6.2.2 Effect of Combustion Conditions on NOX Emissions Emission levels of NOX during the combustion of syngas will be highly variable, and will depend on the particular combustion device and its operating conditions, as well as the level of fixed-nitrogen contaminants in the syngas. Numerous studies have been carried out in both laboratory-scale and full-scale systems. These studies will be summarized below. Giles et al. (2006) studied the NOX emission characteristics of laboratory­generated syngas mixtures in a counterflow diffusion flame. They studied syngas with and without the presence of hydrocarbons, as well as the impact of several diluents on the reduction of NOX. Although the specific NOX emission levels may not be representative of a full-scale device, the trends observed in this study can be quite beneficial to evaluations of performance in real systems that involve diffusion flames. The presence of methane in their test mixtures significantly increased the amount of prompt NO formed, although they noted that the bulk of the NO formed was thermal NO. The presence of methane was also found to reduce the peak flame temperatures. The NOX reduction effectiveness of diluents was found to be H2O > CO2 > N2, in order of decreasing effectiveness. Park et al. (2003) carried out another counterflow diffusion flame study to investigate the impacts of injecting both CO and CO2 on NO emissions. They found that CO2 provided some suppression of overall reaction rate due to its higher heat capacity and that CO provided some attenuation of prompt NO formation. The relative amounts of CO and H2 can have a significant impact on NOX emissions and flame stability. The laboratory-scale study of Kim and Choi (2001) showed that NOX emissions increased as the fuel composition changed from a CO:H2 ratio of 1:1 to 2:1, and that increased levels of CO also resulted in decreased flame stability. The work of Hasegawa et al. (2001), also in a laboratory combustor, examined the influence of the CO:H2 ratio on the conversion of NH3 to NOX. The conversion rate increased from approximately 60% to over 80% when the CO:H2 ratio increased from 0.43 to 2.33 under fuel-lean conditions, and the conversion rate decreased slightly with increasing pressure. If a small percentage of CH4 was present, no influence of the CO:H2 ratio on the conversion of NH3 to NO was observed. Weiland and Strakey (2007) investigated the effect of pressure on NOX emissions during the combustion of syngas for pressures ranging from atmospheric to 8 atm. They found NO levels increased with increasing pressure, as would be anticipated due to a corresponding increase in gas-phase reaction rates. They also investigated the impact of nitrogen (N2) dilution of the syngas and found that dilution had limited impact on NO emissions that were normalized to account for differences in residence time. This behavior is due to trade-offs between thermal dilution effects and enhanced diffusivity of key fuel species such as H2, which result in similar overall flame temperatures. Therefore, the use of nitrogen as a diluent would only reduce NO because of the decrease in the overall residence time. Charlston-Goch and coworkers (2001) used laser-induced fluorescence (LIF) to study NO formation in simulated syngas mixtures in a premixed flame over a range of pressures (1 to 11.9 atm), equivalence ratios (ϕ = 0.7 to 1.3), and CH4 concentrations

174

Synthesis Gas Combustion: Fundamentals and Applications 8.0

NO Concentration/ppm

6.0

1.00 atm 3.05 atm 6.10 atm 9.15 atm 11.9 atm

4.0

2.0

0.0

0.7

0.8

0.9

1.0

1.1

1.2

1.3

Equivalence Ratio (φ)

Figure 6.2  LIF-measured postflame front NO concentrations as a function of equivalence ratio in syngas-air flames (CH4 = 2.8%) for pressures from 1.0 to 11.9 atm. (From CharlstonGocha et al., 2001, used with permission.)

(0 to 14.9%). The simulated syngas mixture consisted of 17.5 vol% H2, 21.2% CO, 12.2% CO2, and 49.1% N2, and represented a syngas derived from air-blown coal gasification. CH4 was then added to this baseline syngas mixture. The results indicated a clearly increasing trend of NO with increasing pressure. However, the pressure effect was highly dependent upon flame stoichiometry. As shown in Figure 6.2, at a pressure of 1 atm the NO concentration varied approximately linearly between 2 and 6.5 ppm as the flame conditions were varied between fuel lean (ϕ = 0.7) and fuel rich (ϕ  =  1.3). As the pressure was increased, however, the effect of equivalence ratio on NO under fuel-rich conditions (ϕ > 1.0) tapered off until, at the highest pressure (11.9 atm), there was essentially no variation in NO concentration between ϕ = 1.0 and ϕ = 1.3. There continued to be a linear increase in NO for fuel-lean conditions, however (ϕ < 1). Charlston-Goch et al. (2001) also investigated the effect on NO formation of increasing the hydrocarbon content of the syngas mixture. Figure 6.3 shows a linear increase in NO concentration as hydrocarbons are introduced, due to the presence of CH species that can promote additional prompt NO formation. Shoffstall and Waibel published a series of reports on the utilization of low-Btu fuels from coal gasification as industrial process fuels (Shoffstall, 1977; Shoffstall and Waibel, 1977; Waibel et al., 1978; Waibel and Fleming, 1979). They performed experiments with medium- and low-Btu gases from five different coal gasification processes: the Lurgi oxygen, Winkler oxygen, Koppers-Totzek oxygen,

175

Pollutant Formation and Control 8.0

NO Concentration/ppm

6.0

1.00 atm 3.05 atm 6.10 atm 9.15 atm 11.9 atm

4.0

2.0

0.0

0.0

12.0 4.0 8.0 Fuel-CH4 Concentration (%)

16.0

Figure 6.3  LIF-measured postflame front NO concentrations as a function of CH4 addition for pressures from 1.0 to 11.9 atm. (From Charlston-Gocha et al., 2001. With permission.)

Wellman-Galusha air, and Winkler air processes. The heating value of these fuels ranged from 3.7 to 11 MJ/m 3 (100 to 300 Btu/ft3). The Lurgi oxygen and Winkler oxygen fuels had compositions that were approximately 40% H 2, 20 to 30% CO and CO2, 3 to 10% CH4, and 1% N2. The Koppers-Totzek fuel contained 53% CO, 35% H 2, 10% CO2, 1% CH4, and 1% N2. The processes that used air as an oxidizer during gasification, Wellman-Galusha and Winkler air, produced syngas (producer gas) with high concentrations of N2 and correspondingly low heating value. The compositions of these gases were 45 to 55% N2, 20 to 25% CO, 15% H 2, 7% CO2, and 1 to 3% CH4. Shoffstall and Waibel’s experiments found that NO emissions increased with increasing excess air and with an increase in the degree of fuel–air mixing. NO emissions from combustion of the different fuels at ambient and elevated fuel temperatures were closely correlated to adiabatic flame temperatures, which indicates that the NO was primarily formed by the thermal NO mechanism. Some experiments were performed with fuel contaminants. Results showed that the level of NO emissions increased with increasing fuel-nitrogen and fuel-sulfur contents.

6.2.3 NOX Control Technologies NOX can be controlled by methods applied within the combustion device (so-called in-furnace control) as well as by methods applied after the combustion device (postcombustion control). In-furnace control methods include low NOX burners and

176

Synthesis Gas Combustion: Fundamentals and Applications

injectors, air staging, flue gas recirculation, water dilution, and reburning. The specific in-furnace control methodology to be used depends upon the primary NO formation mechanism(s). The suitability of each of the methods for different applications will be discussed in subsequent paragraphs. Postcombustion control methods include selective noncatalytic reduction (SNCR) and selective catalytic reduction (SCR), both of which require the injection of a reagent such as ammonia or urea. In addition, many engine applications utilize a postcombustion catalytic converter, and these have been demonstrated to be effective with applications involving syngas components (Akansu et al., 2004). In general, however, in-furnace control is preferred if very high NO reductions are not required, as this generally requires no significant additional footprint, external equipment, or additional reagents. For high levels of NO reduction (>85%), in most cases an SCR catalyst bed with suitable reagent injection is required, although this approach typically incurs the greatest capital and operating expenditures. In-furnace control is primarily targeted at either fuel NO or thermal NO. Fuel NO formation from syngas combustion can be controlled by minimizing contact between fixed nitrogen species present in the fuel (primarily HCN and NH3) and oxygen (O2) in the oxidizer stream. The use of air staging or two-stage combustion is the most common approach for initially limiting contact with oxygen, and this can be accomplished by either physical air staging (removing the oxidizer from the combustion zone and introducing it downstream) or aerodynamic staging (limiting early mixing between the fuel and oxidizer streams through various burner or injector designs). If the syngas fuel stream passes through a cleaning step prior to combustion, the fuel nitrogen species may be largely removed, and thus control of fuel NO formation may not be necessary. A recent demonstration of the successful application of this technology to syngas combustion was provided by Hasegawa and Tamaru (2007). Thermal NO is dominated by temperature effects and is less sensitive to specific gas composition variations. Thus, control technologies developed and proven for various gas-fired applications can be easily applied to the combustion of syngas. Thermal NO can be readily controlled by limiting the peak flame temperatures in the primary combustion zone, and two common methods are flue gas recirculation (FGR) and water injection. FGR reduces peak flame temperatures by diluting the combustion products, and water injection extracts heat from combustion products by vaporization of the water (as well as by providing some dilution). In typical FGR applications, 50 to 75% NOX reduction is achievable with the use of 10 to 20% by weight recirculated flue gas. The flue gas can be recirculated externally by injecting exhaust gases into the combustion zone, or the flue gas can be recirculated internally through the use of eduction techniques to draw furnace gases into the burner to dilute the near-burner combustion process (Sarofim and Flagan, 1976; Feese and Turns, 1998; Milani and Nelli, 1992). The use of water injection has limited application due to efficiency penalties associated with the addition of the water. However, for combustion systems where the penalties can be tolerated, or where the additional flue gas mass may be beneficial for heat transfer purposes, the method can be quite effective. Sarofim and Flagan (1976) reported a 90% reduction in NOX using injection of water at a level comparable to 15% by weight of the air and fuel fed to the burner.

177

Pollutant Formation and Control

There are also indications that FGR may be more effective for NOX reduction with syngas than for more conventional hydrocarbon fuels. A recent study by Smith and Bartley (2000) illustrated the benefits of using FGR on stoichiometric mixtures of natural gas and syngas, as opposed to natural gas alone. The syngas–natural gas mixture resulted in a 77% NOX reduction beyond that obtained with natural gas for similar levels of FGR in a single-cylinder Caterpillar research engine. Another potentially useful attribute of syngas in terms of thermal NO control is the typically high H2 content, which can greatly increase the overall burning rate of the fuel if the H2 content is high enough. The formation of thermal NO is kinetically limited, requiring high temperatures such that the kinetic rates are fast enough to form significant amounts of NO within the residence time of the combustion device. If the H2 content of the fuel significantly increases the overall burning rate, then it may be possible to reduce the residence time of the combustor and thereby reduce thermal NO formation (GE Energy, 2005). Reburning, or staged fuel injection, is another NOX reduction technique initially suggested by Wendt and coworkers (1973), in which fuel is injected into the combustion products, downstream of the primary combustion zone, as shown in Figure 6.4, where SR denotes the stoichiometric ratio (air input/air required for complete combustion). This fuel addition creates a local reducing environment in the area around and just downstream of the point of injection, which results in the conversion of NO to N2. Air is injected downstream of the reburn zone to produce fuel-lean conditions overall in the furnace. This air, referred to as burnout air, oxidizes the remaining fuel fragments after the reburn zone. Although many in-furnace control technologies target specific formation mechanisms, reburning is applied slightly downstream of the primary formation zone and can thus be used to reduce NO formed in the flame, regardless of the dominant formation mechanism. The feasibility of this method, however, is specific to the particular combustion device. Still, there are certain advantages in the application of reburning. For example, the peak flame temperature in the primary combustion zone can be reduced if a portion of the primary fuel is used as the reburn fuel. This approach can also distribute the fuel heat release if such an effect is desirable in the combustion system. Reburn fuel injection

Burnout air injection

NO reduction

Some NO reformation

SR typically > 1.1

SR < 1.0

SR > 1.1

Primary combustion zone

Reburn zone

Burnout zone

NO formation

Figure 6.4  Configuration for the application of reburning for NOX control. SR = stoichiometric ratio.

178

Synthesis Gas Combustion: Fundamentals and Applications

The principal mechanism involved in traditional reburning is the reaction of hydrocarbon radicals from the reburn fuel with NO to form HCN. Subsequently, the HCN is converted to an ammonia fragment (NHi) and then to N2 under fuel-rich conditions, as shown in Figure 6.1. Reburning has been proven with a range of hydrocarbon fuels in gaseous, liquid, and solid states, including natural gas, coal, wood, and oil (Chen et al., 1986, 1989; Bales, 1995; Stapf et al., 1998; Smoot et al., 1998). For the special case of reburning using syngas as the injected reburn fuel, the primary reactants would be CO and H2 (although the small amount of hydrocarbons typically present could still provide some NO reduction). Nonhydrocarbon fuel reburning with species such as CO and H2 has been shown to provide some effectiveness (Bortz and Offen, 1987; Chen et al., 1986; Rutar et al., 1996; Lissianski et al., 2002; Wu et al., 2004). Chen et al. (1986) found that reburning with H2 and CO produced notable NOX reduction, although it was less effective than hydrocarbon fuels. Furthermore, reburning with H2 led to a steady NOX decrease as the reburn zone stoichiometric ratio (SR) was reduced from 1.10 to 0.70. The reduction effectiveness did not proceed through a maximum at a reburn zone SR near 0.90, which is a characteristic observation with hydrocarbon fuels. In the reburn zone, the NO concentration drops to significantly lower levels for hydrocarbon fuels than for CO or H2 due to reactions that convert NO in the presence of hydrocarbons to fixed nitrogen species (HCN, NH3). These fixed nitrogen species are then reoxidized to some extent to NO upon addition of burnout air such that the net NO reduction is not as great as that achieved in the actual reburn zone. For reburning with CO and H2, the NO levels achieved in the reburn zone are not as low as those with hydrocarbon fuels. However, there is limited production of fixed nitrogen species to reconvert into NO, so NOX levels remain unchanged as the gas passes through the burnout air addition zone. Because equilibrium favors greater concentrations of fixed nitrogen species at lower reburn zone stoichiometries, the reburning effectiveness of hydrocarbon fuels goes through a maximum at a stoichiometric ratio of approximately 0.9. The effectiveness then decreases at lower reburn zone stoichiometries due to subsequent oxidation of the increased levels of fixed nitrogen species. This effect is not observed with H2 and CO due to the limited fixed nitrogen species production, and thus increasingly higher reductions are obtained at lower stoichiometries. This effect of reburning can be predicted readily by means of the CHEMKIN plug-flow reactor model to simulate the combustion, reburn, and burnout zones. Figure 6.5 shows the results for CH4, H2, and CO reburning. The calculations were performed using the GRI mechanism (Smith et al., 1999), with a reburn zone initial temperature of 1760 K, a 100 K/s quench rate, and a reburn zone residence time of 0.4 s. Varying amounts of CH4, CO, and H2 were added as reburn fuels to achieve different reburn zone stoichiometries. The final stoichiometry after the burnout air addition was 1.15. Additional kinetics calculations indicated that the CO and H2 reburning effectiveness was not as sensitive to the reburn zone temperature as CH4. Bortz and Offen (1987) compared the reburning performance of two different types of syngas with that of natural gas for a coal-fired primary flame in a 500,000 Btu/h

179

Pollutant Formation and Control 100%

CH4 reburning

90%

H2 reburning

% Reduction of NOX

80%

CO reburning

70% 60% 50% 40% 30% 20% 10% 0%

SR = 0.6

SR = 0.8

SR = 0.9

SR = 1.0

Reburn Zone SR

Figure 6.5  Kinetics calculations demonstrating reburning effectiveness of various species as a function of reburn zone stoichiometry.

(150 kW) furnace. The two syngas fuels were commercial coal gasification products from two distinct gasification processes, denoted as the Lurgi low-Btu gas (LBG) and Koppers-Totzek medium-Btu gas (MBG), respectively. The simulated gases contained roughly 30% H2 and 10% CO2 without any methane. The difference between the two was the amount of CO and N2 they contained. LBG contained 16% CO and 45% N2, as compared to 55% CO and 1.0% N2 for MBG. Reburning with the two syngas fuels was found to reduce NOX emission reductions in the range of 20 to 30%, while under the same conditions, natural gas reburning resulted in a 60 to 70% NOX reduction. After some optimization, it was determined that if the primary com­bustion zone was operated at a stoichiometric ratio between 1.00 and 1.05, with the thermal input from the reburn fuel accounting for 15 to 20% of the total firing rate, then NOX reductions of 57 and 63% could be achieved with the use of MBG and LBG, respectively. For the same conditions, natural gas produced a 78% reduction in NOX. In summary, NOX emissions from the combustion of syngas can have contributions from fuel, thermal, and prompt NO mechanisms. The relative contribution of each depends on the syngas composition and combustion conditions. NOX species can be readily converted to an inert gas species (N2). This conversion can be achieved using both combustion modifications and postcombustion controls. As a result, a wide range of options exist for the control of NOX emissions, and each combustion application should be evaluated to determine the most suitable means of control.

6.3 Sulfur Species The primary sulfur component in syngas is hydrogen sulfide, which may exceed a concentration of 2% for sulfur-rich fuels. Smaller concentrations of other reduced

180

Synthesis Gas Combustion: Fundamentals and Applications

sulfur compounds such as carbonyl sulfide, mercaptans, dimethyl sulfide, dimethyl disulfide, and carbon disulfide may also exist (Ratafia-Brown et al., 2002). Prior to combustion in a gas turbine, nearly all sulfur compounds must be removed from the syngas. Current gas cleaning technologies generally achieve 95 to 99.9% removal of these compounds (Kathmann and Huder, 2001; Ratafia-Brown et al., 2002). The target for future applications is to achieve almost zero concentration of sulfur species in the gas. Because the gas entering the turbine is nearly sulfur-free, emissions of sulfurous compounds from IGCC systems are typically lower than from conventional power generation technologies (e.g., PC-fired boilers) with postcombustion SO2 removal systems (Table 6.1). Available measurements from IGCC systems indicate that, on a constant power output basis, SO2 emissions are less than half, and can be as little as 3% as much as, those from pulverized coal-fired boilers (Amick, 2004; Hornick, 2002; Ratafia-Brown et al., 2002). Because gas turbine combustors are highly efficient and hydrogen sulfide is very reactive with oxygen, there is essentially no risk for emission of unburned hydrogen sulfide in well-­performing turbines. Interestingly, unexpectedly high concentrations of SO3 and sulfate aerosol precursors have been measured from aircraft engine exhaust (Streets, 1963; Lukachko et al., 1998) and may be enhanced by turbulent flow fields within turbine blades or by catalysis of SO2 oxidation by the turbine blade material (Streets, 1963; Lukachko et al., 1998; Sorokin et al., 2004). Similar behavior may occur in syngas-fired gas turbines, but overall sulfate aerosol emissions would be very low due to the low sulfur content of the clean syngas. For syngas fired in boilers, kilns, and internal combustion engines, sulfur species in the syngas are oxidized primarily to SO2. Some of the SO2 will undergo further oxidation to form SO3. In systems with inefficient mixing of the air and syngas, fuel-rich regions or flow streams may exist. Under such circumstances, very little oxygen is available for complete combustion to SO2. Any reduced sulfur species remaining after all oxygen is consumed exits the system in a reduced form (e.g., H2S). Consequently, the partitioning of gaseous sulfur emissions between oxidized species (SO2 and SO3) and reduced species depends on the combustor performance and gas mixing. For syngas produced from high-sulfur fuels through gasification and without syngas sulfur removal, the flue gas created upon combustion contains a high concentration of sulfur and needs to be treated by postcombustion methodologies.

6.4 Carbon Monoxide Carbon monoxide in syngas combustion products has two primary sources: unburned syngas CO, resulting from inefficient mixing that yields equivalence ratios outside the ignition range, and incomplete combustion of hydrocarbon species in the syngas. Carbon monoxide can also result from decomposition of lubricating oils when syngas is burned in reciprocating engines (Li and Karim, 2005). Experience from coal-fired power plants indicates that final CO emissions from gas turbine-based IGCC systems are lower than those from conventional combustion-based systems (Table 6.1). Burning mixtures of hydrogen and CO in piston engines has also been determined to produce much lower emissions of CO than either gasoline or natural gas under similar conditions (Mustafi et al., 2006).

Pollutant Formation and Control

181

The flammability limits of major components in syngas have been tabulated in the review by Chomiak et al. (1989), which provides a boundary for the lower ignition limit that is usually slightly higher than the lower flammability limit due to the effect of flame shapes on flammability. For example, Li and Karim (2005) reported a lean operational limit in their engine emission study of a methane fuel at ϕ = 0.6 in comparison with the lower flammability limit at ϕ = 0.53 reviewed by Chomiak et al. The lean operational limit for CO was found to be ϕ = 0.45 in the engine study in comparison with ϕ = 0.34 in the review. Pure hydrogen gas has a lower flammability limit, similar to methane on a volume percent basis, but a much higher upper limit (75%) than hydrocarbon fuels. Carbon monoxide has a higher flammability limit (74%), similar to hydrogen (Chomiak et al., 1989). Therefore, syngas with dominant carbon monoxide and hydrogen fractions presents a much wider ignition range than typical ranges for conventional hydrocarbon fuels of natural gas and oils. The extension of the ignition range toward a lower lean limit after the addition of hydrogen to other fuels is discussed for a gas engine fired with a process gas (Gruber and Herdin, 1997). In another study, the addition of 30 vol% hydrogen to a methane fuel significantly reduced the operational limit in a spark ignition (SI) engine from an equivalence ratio of 0.6 to 0.48 (Li and Karim, 2005). The new lean limit is even lower than the methane flammability limit at ϕ = 0.53. The expanded operational range effectively lowers the probability of unburned gas eddies of improper mixing during the combustion of synthesis gas, resulting in low emissions of unburned CO, even in systems with poor mixing judged by the standard of conventional designs. The above-mentioned SI engine was fired with pure hydrogen gas in order to characterize the CO and CO2 emission from the decomposition of engine lubricating oils. As shown in Figure 6.6, the hydrogen-fueled engine experiment indicated a strong linear dependence of CO emission on the equivalence ratio, which was observed by Li and Karim (2005). The CO emission from the decomposition of lubricating oils in a SI engine fueled with hydrogen gas at a compression ratio of 7 running at a constant speed of 900 rev/min can be estimated based on the equivalence ratio by the linear regression fitting [CO] in ppm = 521.76 × ϕ – 10.298, which yields a CO concentration of 210 ppm at ϕ = 0.6 and 360 ppm by extrapolation to ϕ = 0.88. In comparison, at ϕ = 0.88, the same engine fueled with mixtures of H2 and CH4 under similar operation conditions but with a slightly higher compression ratio yields CO emissions of 590 ± 40 ppm, calculated from results covering a variation of the hydrogen content from zero to 50% in the fuel. Therefore, in engines fueled with methane-rich syngas, a significant portion of the CO emission comes from the decomposition of engine oils, which indicates a very high degree of complete combustion with these gaseous fuels. The gaseous fuel emits less CO than conventional fuels used in SI engines. For example, in the United States the average allowable CO concentration for new vehicles is on the order of 2000 ppm, according to the U.S. Environmental Protection Agency (EPA) standard of 3.4 g/mile and assuming 25 miles/gallon gas mileage. In summary, incomplete combustion of syngas fuel CO and hydrocarbons contri­ butes little to CO emissions in a reasonably well-designed and well-operated combustion device. The hydrogen content in syngas fuels extends the operational limit and facilitates the oxidation of other organic fuel fractions by providing elevated

182

Synthesis Gas Combustion: Fundamentals and Applications Temperature in °C (Ditaranto et al., 2007)

100

300

400

500

600

700

800 100

H2 and CH4 fuel, CR = 8.5, Tin = 22°C, ST =15°CA BTDC, 900 rpm, H2/H2 + CH4 = 0.15%, 6 data points (Li-Karim, 2005) H2 Fuel CR = 7, Tin = 22°C ST = 15°CA BTDC N = 900 rpm (Li-Karim, 2005)

y= 521.8x – 103.0 R2= 0.911 10

10

1 0.0

CH3OH-H2 Fuel Monolith catalytic reactor (Ditaranto, 2007) 0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

(CO) in ppm (Ditaranto et al., 2007)

(CO) in ppm (Li and Karim, 2005)

200 1000

1 1.0

Equivalence Ratio (Li and Karim, 2005)

Figure  6.6  CO emission data from three experiments are compared to each other. Diamonds represent the CO emission from a SI engine fueled with hydrogen gas as a function of the equivalence ratio. Squares represent the CO emission in the same engine under similar conditions but fired with CH4 and H2 mixed fuels at various mixing ratios with a fuel equivalence of 0.88. Triangles represent the CO emission from a monolith catalytic reactor burning the anode off-gas of a methanol fuel cell.

combustion temperatures. For turbine and boiler applications, CO emissions from syngas combustion are low, as evidenced by the data in Tables 6.1 and 6.2. In diesel or SI engines, however, decomposition of lubricating oils can contribute significantly to CO emissions.

6.5 Volatile Organic Compounds Incomplete combustion products include volatile organic hydrocarbons, their radicals, and polycyclic aromatic hydrocarbons (PAHs). Combustion of gaseous fuels yields lower emissions of VOCs than liquid fuels. Particulate matter, including tars, is usually removed before firing gaseous fuels in gas turbines, boilers, engines, and other combustors. However, this may not be true in small gasifiers operating on residual fuels that often lack sophisticated gas scrubbing. Natural gas has been increasingly used in the transportation sector, displacing the more polluting liquid fuels. Syngas is even cleaner in terms of VOC emissions since it usually contains a smaller amount of hydrocarbon compounds. Emissions of VOCs from low calorific heating value gases depend exclusively on minor fractions of paraffinic species, for example, methane in most gases, and to a lesser extent on the chemical and thermal properties of the main components. CO in synthesis gas is not an efficient carbon source for VOCs. Therefore, it is insufficient to study the VOC emissions from the combustion of syngas including only major fuel fractions with the assumption that the carbon source of VOCs comes mainly from CO. Major reaction pathways of VOC have been discussed elsewhere for laminar

183

Pollutant Formation and Control

+H-OH

+O

+HO

2-O

2-OH

CO

CO2

2-OH

CH2O

+H

+HX-X

CH3

+H

+HO -O 2 2

H–

HCO

CH4

CH2 = CH.

C2H2

C2H4

C2H5

C2H6

Figure 6.7  Formation pathways of the vinyl radical (CH2 = CH•) in a modeled premixed flame of a generic synthesis gas that does not include a 2% methane fraction.

premixed flames of small paraffins (Marinov et al., 1996) and composite liquid fuels (Zhang et al., 2008). Park and coworkers (2003) discussed a set of very different pathways in an opposed-flow diffusion flame burning hydrogen gas with the addition of carbon oxides on the fuel side. To gain a better understanding of VOC formation and the influence of methane, premixed combustion of syngas with or without methane was modeled using the Utah surrogate mechanisms, which have been critically validated with experimental species profiles of a premixed flame with synthetic natural gas (Zhang et al., 2007). A generic synthesis gas composition of 40% CO, 25% H2, 25% H2O, and 10% CO2 was used, and 1% H2 and 1% H2O were replaced with 2% CH4 when the effect of hydrocarbon-containing fuels on VOC emissions was considered. The major formation pathways of the vinyl radical are mapped out in Figures 6.7 and 6.8 for the fuels without or with 2% methane, respectively. The vinyl radical is probably the most reactive hydrocarbon radical in flames and facilitates many important pollutant formation reactions (Zhang et al., 2008). The arrows indicate the carbon flow between pairs of species, and the thickness of an arrow is proportional to the corresponding reaction rate on a logarithmic scale. If the generic gas does not contain methane, the CO fraction is the main carbon source for VOC. CO has three resonantly stabilized structures between − C≡O+ ↔ C=O ↔ +C-O −. The triple bond of CO undergoes a cascading hydrogenation that forms C–H sigma bonds on the more reactive carbon atom. As seen in Figure 6.7, the majority of CO forms CO2, and only a small fraction of CO undergoes hydrogenation (HCO), which is the first C–H sigma bond formation at a rate that is two orders of magnitude lower than that of the CO2 formation. The rates of forward and reverse reactions between CO and HCO are balanced. A second C–H sigma bond formation yields formaldehyde CH2O at a rate that is five orders of magnitude lower than the

184

Synthesis Gas Combustion: Fundamentals and Applications +X-XH

CH2 = CH.

CH2O

+

H-C

CO

H -O

+ O2

+HO -O 2 2

H–

HCO

H

C2H2

3

CH2CO +

+

CH

CO

C2H4

+X-HX

XXH

+

C2H5

+

CO-H

HCCO

+

O

2 -H

CO

-H OH

+H-OH

CH3

+ C 2-

+HO -OH 2 +O -O 2

-H

2

-H

CO2

1CH /3CH 2 2

CH4

C2H6

Figure 6.8  Formation pathways of the vinyl radical (CH2 = CH•) in a modeled premixed flame of a generic synthesis gas that includes a 2% methane fraction.

CO2 formation. The formation of the first two C–H sigma bonds is comparatively easier than formation of the third sigma bond because the first π-bond given up in hydrogenation is a pair of lone electrons donated by oxygen. Only about 0.1% of formaldehyde undergoes further hydrogenation either directly or via intermediate CH2OH and yields the CH3 radical. The estimated formation rate of the vinyl radical from the methyl radical is negligible, probably at the noise level. Therefore, the combustion of syngas without methane yields almost no VOC emissions. When 2% methane by volume is added to the generic synthesis gas, methane becomes the sole carbon source of VOC, as seen in Figure 6.8. The formation of the conjugate methyl radical (CH3) yields C2 species via combination, and C2 species undergo consecutive dehydrogenation for the formation of the vinyl radical, the rate of which is about five orders of magnitude lower than that of methane consumption. Methane is a clean fuel, although it produces considerably higher VOC emissions than CO and H2 fuels. Figure  6.8 shows that the cascading hydrogenation of CO toward the CH3 radical is ineffective, as the readily formed CH3 radical from methane also favors an oxidation pathway of equal importance to methyl combination, and this alternate pathway for methyl radical leads to oxygenate species and will ultimately release CO via dehydrogenation steps. In summary, the formation and control of VOC will depend upon fuel composition, as syngas with little or no hydrocarbon content initially is unlikely to produce

Pollutant Formation and Control

185

significant VOCs in the residence times of typical combustors. For fuels with significant hydrocarbons, the formation pathways for VOCs are more readily achieved. Under these conditions, it will be important to ensure sufficient mixing with the oxidizing medium at high enough temperatures to facilitate destruction of the VOCs. As mentioned in Section 6.4, a well-designed and well-operated combustion device should not have difficulty in achieving low levels of VOC. A notable exception could be with syngas combustion in SI engines, in which CO emissions are possible due to the decomposition of the lubricating oil.

6.6 Trace Elements The concentration of trace elements in syngas combustion flue gas depends on many factors, including the feedstock composition, type of gasification system, gasification temperature, system pressure, and configuration of the gas cleaning system. Coal contains many trace elements, including As, B, Ba, Be, Ca, Cd, Co, Cr, Cu, Fe, Hg, K, Mn, Mo, Na, Ni, Pb, Sb, Se, Si, Sn, V, and Zn, as well as chlorine and other halogens. Most of these components exist at very low concentrations, and many are not considered pollutants. Biomass materials generally contain a similar mix of components but tend to have higher concentrations of alkali metals (K, Na) and chlorine (Oakey et al., 2004). Unlike oxidizing combustion conditions, the most stable gas species under gasification conditions may be a reduced compound such as a sulfide. The volatility profiles of trace metals is also correspondingly different. Oakey et al. (2004) considered the volatility of metal species formed during biomass gasification and concluded that these species and their associated volatilities can be significantly different from those formed under combustion conditions, particularly for alkalis, arsenic, vanadium, and boron. Generally, the volatility of metals is higher under gasification conditions. For coal-based systems, several trace metals, including As, Se, Sb, Pb, and Hg, can remain volatile throughout the entire syngas cleaning and conditioning train (Frandsen et al., 1994; Helble et al., 1996). The same metals can remain volatile in biomass gasification syngas, which can also exhibit notable concentrations of Ba, Cd, K, V, and Zn (Oakey et al., 2004). Final concentrations will depend on the specific system configuration. In gas cleaning systems that cool the syngas to relatively low temperatures (below 350°C), the majority of metal species will condense and be removed in a scrubber or filter prior to syngas firing. Sorbent beds of, for example, activated carbon, may also be installed specifically to remove trace metals. Measurements of airborne trace metals from IGCC systems indicate that concentrations are generally equal to or lower than emissions from conventional power systems (Radian Corp., 1995; Ratafia-Brown et al., 2002). Mercury deserves specific discussion due to its adverse health effects and recent attention as an emission from coal-fired power plants. Mercury has a low boiling point (357°C) and is present in syngas predominantly as elemental mercury (Hg). Oxidized mercury (e.g., HgCl2) and mercuric sulfide (HgS) may also exist in smaller proportions. Much of the mercury is removed in syngas cleaning equipment of IGCC systems (Ratafia-Brown et al., 2002), but an analysis of mercury behavior in operating IGCC plants indicates that a significant fraction of the mercury in the coal can

186

Synthesis Gas Combustion: Fundamentals and Applications

be exhausted from the gas turbine, primarily as elemental mercury (Granite et al., 2002). As noted in Table 6.1, mercury emissions from IGCC plants are of the same magnitude as from conventional PC-fired boilers, on the order of 1 to 5 lb/1012 Btu. Currently, most IGCC systems do not include technologies to specifically remove mercury from the syngas. However, several effective sorbents, including activated carbon, are available and have been successfully employed in coal gasification systems, providing capture efficiencies in excess of 90% (Ratafia-Brown et al., 2002; Denton, 2004). Chlorine present in the feedstock is converted primarily to hydrogen chloride (HCl) during gasification. Most of this will be removed in systems with a wet scrubber (U.S. DOE, 2002). In power generation systems that include acid gas removal (AGR) systems for the synthesis gas, more than 99% chlorine removal can be achieved (U.S. EPA, 1995). The final form of chlorine that does exit the combustion system depends on combustion conditions and the concentration of metal species in the gas. Typically, chlorine will partition between metal chlorides and HCl. Alkali chlorides are especially stable under low-SO2 conditions. Fluoride is expected to behave in a manner similar to that of chlorine. In systems that do not involve intensive syngas cleaning, the fate of trace elements depends largely on the temperature profile of the syngas prior to combustion. If the temperature falls below the gas-liquid transformation temperature, trace metals condense and generally are not emitted with the combustion flue gas. If a wet scrubber or particle filter is present, the condensed material can be removed from the flue gas in this system. In the absence of such equipment, condensed metals or oxides of the metals may deposit on the process equipment, particularly cooler components, such as heat exchangers for cooling the raw syngas, and would require periodic removal. Volatile metals, notably As, Hg, Sb, Se, and V (Oakey et al., 2004), can be expected to remain in the syngas and produce mostly metal oxides as a result of combustion.

6.7 Particulate Matter Particulate matter in synthesis gas can be grouped into three broad classifications based on source: (1) inorganic material in the fuel, which is released either as condensed-phase material or as vapor that subsequently condenses; (2) for fluidized bed gasifiers, attrited bed material that has elutriated from the bed; and (3) polyaromatic hydrocarbon species that condense to form tars. Although tar concentrations in syngas can be quite high, particularly for biomass gasification systems (Milne, 1998), tars contribute little to combustion emissions because they are typically cracked or scrubbed in the syngas cleaning system, and any residual tars will be burned in the syngas combustor. The concentration of inorganic parti­ culates in raw syngas depends on the feedstock, gasifier configuration, and gasification conditions. The total inorganic content in a fuel is generally not a good indicator of syngas particulate levels, since much of the material is removed in the gasification process, either as slag in high-temperature gasifiers or as bed material in fluidized bed gasifiers. In IGCC systems, particle filters or scrubbers in the syngas cleaning system remove the bulk of particulate matter present in the

187

Pollutant Formation and Control

raw syngas (Ratafia-Brown et al., 2002). The concentration of inorganic material in the syngas is thus very low, as are resulting emissions from syngas combustion. Measured concentrations of inorganic particulates from IGCC plants are lower than those from corresponding PC-fired plants with postcombustion control by as much as one order of magnitude on an equal Btu basis (Hornick, 2002; Amick, 2004; Khan, 2006).

6.8 Carbon Dioxide Essentially all carbon contained in syngas components ends up as carbon dioxide when the syngas is burned, provided that there is sufficient air and mixing. Syngas components contributing to CO2 emissions include carbon monoxide, hydrocarbons, and of course, carbon dioxide in the syngas itself. The mass of carbon dioxide produced per volume of syngas can be calculated by the formulas below:



kg CO 2 per Nm 3 (0°C, 1 atm) of syngas = 1.963

∑( y n )

pounds CO 2 per scf (70°F, 1 atm) of syngas = 0.1137

i C,i

∑( y n ) i C,i

(6.4)

(6.5)

where yi is the volume fraction (mole fraction) of syngas component i and nC,i is the number of carbon atoms in each molecule of component i. To determine the mass of CO2 produced on a heating value basis (e.g., kg CO2 per MJ of syngas), the constants in the formulas above can be divided by the heating value (MJ/Nm3 or Btu/scf) of the syngas. If syngas is being used to displace natural gas, the impact on carbon dioxide emissions can be determined from Equations 6.4 and 6.5. First, the mass of CO2 emitted per unit heating value should be calculated for both the natural gas and the syngas, making sure to divide the formula constant by the heating value of the gas. Dividing the result for the syngas by the result for the natural gas will result in a factor representing the change in CO2 emissions on an equal heat input basis. For example, consider the natural gas and syngas represented in Table 6.3. From Equation 6.4 one can calculate CO2 emissions for natural gas:

1.963[0.97(1)CH4 + 0.03(2)C2H6 + 0.01(1)CO2] = 1.963(1.01) = 1.983 kg CO2/Nm3 gas

Dividing this result by the heating value of the natural gas (35.58 MJ/Nm3) results in CO2 emissions, on a heating value basis, of 0.0557 kg CO2 per MJ natural gas. Similar calculations for the syngas result in emissions of 1.276 kg CO2/Nm3 gas and 0.1042 kg CO2/MJ syngas. Thus, on an equal heating value basis, the syngas results in 0.1042/0.0557 = 1.87 times more CO2 emissions than the natural gas it would be

188

Synthesis Gas Combustion: Fundamentals and Applications

Table 6.3 Example Natural Gas and Syngas Properties Component CO CH4 C2H6 H2 CO2 N2 Heating value (LHV)

Units

Natural Gas

Syngas

vol% vol% vol% vol% vol% vol% MJ/Nm3

— 94  3 —  1  2 35.58

32  7  3 35 20  3 12.24

displacing. An increase in CO2 emissions when displacing natural gas is typical, and is a consequence of the much lower heating value of the syngas.

6.9 Conclusions Syngas is a unique fuel in terms of pollutant formation during combustion. Because it is formed by gasification of fuels such as coal, petcoke, heavy oil, and biomass, syngas may contain components such as metals and halides that are not typically present in gaseous fuels such as natural gas or liquefied petroleum gas. Combustion of syngas generally results in lower gaseous emissions than direct combustion of the fuel from which the syngas is produced. This is particularly true for IGCC systems, in which contaminants such as sulfur species, halides, and trace elements are efficiently removed from the syngas prior to its firing in the gas turbine. In systems without intensive syngas cleaning, for example, when the syngas is fired in a boiler or kiln, sulfur compounds, halogens, and trace metals in the syngas are emitted in much the same form as for conventional combustion systems (e.g., SO2, HCl, fly ash). Final concentrations of these pollutants are dictated by the efficiency of postcombustion pollutant control systems. NOX, CO, VOC, and PAH emissions are not directly tied to the presence of pollutants in the syngas, but instead depend on the design and conditions of the particular combustion system. Low-NOX combustion techniques typically employed on industrial systems, such as air staging, fuel staging (reburning), and flue gas recirculation, are effective in reducing NOX emissions from syngas combustion. Emissions of CO and VOCs depend largely on the effectiveness of air and syngas mixing in the combustion system. VOC production is also enhanced by hydrocarbon species in the syngas, which provide radicals that promote growth of hydrocarbon molecules. In applications where natural gas is wholly or partly displaced by syngas, emissions of CO2 will likely increase because syngas generally has a lower heating value and a correspondingly higher carbon-to-heating value ratio than natural gas. Under some circumstances, when the syngas contains high concentrations of CO and CO2, emissions of CO2 from syngas combustion can be double that of natural gas on an equal heat input basis.

Pollutant Formation and Control

189

References Akansu, S. O., Dulger, Z., Kahraman, N., and Veziroglu, T. N. (2004). Internal combustion engines fueled by natural gas-hydrogen mixtures. Int. J. Hydrogen Energy 29:1527. Amick, P. (2004). Wabash River coal gasification repowering project overview. Paper presented at the Workshop on Gasification Technologies, Indianapolis, June 8–9. Bales, F. E. (1995). The Use of wood as a reburning fuel in combustion systems. MSc thesis, University of Utah, Salt Lake City. Bortz, S. J., and Offen, G. R. (1987). Reburning with low and medium Btu gases. Paper presented at Proceedings of the 1987 Joint EPA/EPRI Symposium on Stationary Combustion NOX Control, New Orleans, March 23–26. Bozzelli, J. W., and Dean, A. M. (1995). O + NNH: A possible new route for NOX formation in flames. Int. J. Chem. Kinet. 27:1097. Bozzelli, J. W., and Dean, A. M. (2000). Combustion chemistry of nitrogen. In Gas-phase combustion chemistry, ed. W. C. Gardiner, 126, New York: Springer-Verlag. Charlston-Goch, D., Chadwick, B. L., Morrison, R. J. S., Campisi, A., Thomsen, D. D., and Laurendeau, N. M. (2001). Laser-induced fluorescence measurements and modeling of nitric oxide in premixed flames of CO+H2+CH4 and air at high pressures. I. Nitrogen fixation. Combust. Flame 125:729. Chen, S. L., McCarthy, J. M., Clark, W. D., Heap, M. P., Seeker, W. R., and Pershing, D. W. (1986). Bench and pilot scale process evaluation of reburning for in-furnace NOX reduction. Proc. Combust. Instit. 21:1159. Chen, S. L., Kramlich, J. C., Seeker, W. R., and Pershing, D. W. (1989). Optimization of reburning for advanced NOX control on coal-fired boilers. J. Air Waste Mgmt Assoc. 39:1159. Chomiak, J., Longwell, J. P., and Sarofim, A. F. (1989). Combustion of low calorific value gases: Problems and prospects. Prog. Energy Combust. Sci. 15:109. Denton, D. L. (2004). IGCC’s environmental and operational capabilities today. Paper presented at the Workshop on Gasification Technologies, Indianapolis, June. Feese, J. J., and Turns, S. R. (1998). Nitric oxide emissions from laminar diffusion flames: Effects of air-side versus fuel-side diluent addition. Combust. Flame 113:66. Frandsen, F., Dam-Johansen, K., and Rasmassen, P. (1994). Trace elements from combustion and gasification of coal—An equilibrium approach. Prog. Energy Combust. Sci. 20:115. GE Energy. (2005). Premixer design for high hydrogen fuels. Final report, U.S. DOE/NETL, DE-FC26-03NT41893. Giles, D. E., Som, S., and Aggarwal, S. K. (2006). NOX emission characteristics of counterflow syngas diffusion flames with airstream dilution. Fuel 85:1729. Granite, E. J., King, W. P., and Pennline, H. W. (2002). Techniques for mercury control and measurement in gasification systems. Paper presented at the 5th International Symposium on Gas Cleaning at High Temperature, Morgantown, WV, September. Gruber, F., and Herdin, G. R. (1997). The use of H2-content process gas in gas engines. Proc. 1997 ICE Fall Tech. Conf. (ASME) 29:17. Harrington, J. E., Smith, G. P., Berg, P. A., Noble, A. R., Jeffries, J. B., and Crosley, D. R. (1996). Evidence for a new NO production mechanism in flames. Proc. Comb. Inst. 26:2133. Hasegawa, T., Sato, M., and Nakata, T. (2001). A study of combustion characteristics of gasified coal fuel. J. Eng. Gas Turb. Power 123:22. Hasegawa, T., and Tamaru, T. (2007). Gas turbine combustion technology reducing both fuelNOX and thermal-NOX emissions for oxygen-blown IGCC with hot/dry synthetic gas cleanup. J. Eng. Gas Turb. Power 129:358. Helble, J. J., Mojtahedi, W., Lyyränen, J., Jokiniemi, J., and Kauppinen, E. (1996). Trace element partitioning during coal gasification. Fuel 75:931.

190

Synthesis Gas Combustion: Fundamentals and Applications

Hornick, M. (2002). Polk power station IGCC. Paper presented at the 2nd DOE/UN Interna­ tional Conference and Workshop on Hybrid Power Systems, Charlotte, NC, April. Hornick, M. (2006). Polk power station. Paper presented at the Gasification Technologies Workshop, Tampa, FL, March 2–3. Houghton, J. T., Ding, Y., Griggs, D. J., Noguer, M., van der Linden, P. J., and Xiaosu, D., Eds. (2001). Climate change 2001: The scientific basis. Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change (IPCC). Cambridge, UK: Cambridge University Press. Intergovernmental Panel on Climate Change (IPCC). (2007). Fourth assessment report: The physical science basis. Kathmann, R., and Huder, K. (2001). Emission free efficiency. Hydrocarbon Eng. 6:75–77. Khan, S. R. (2006). Environmental footprints of IGCC and PC plants—An update. Paper presented at the Workshop on Gasification Technologies, Tampa, FL, March 2–3. Kim, H.-T., and Choi, B.-C. (2001). Coal-derived syngas combustion for the application in IGCC system. ACS Symp. Ser. 46:532. Konnov, A. A., Colson, G., and De Ruycka, J. (2000). The new route forming NO via NNH. Combust. Flame 121:548. Konnov, A. A., Colson, G., and De Ruycka, J. (2001). NO formation rates for hydrogen combustion in stirred reactors. Fuel 80:49. Li, H., and Karim, G. A. (2005). Exhaust emissions from an SI engine operating on gaseous fuel mixtures containing hydrogen. Int. J. Hydrogen Energy 30:1491. Lissianski, V., Zamansky, V., and Rizeq, G. (2002). Integration of direct combustion with gasification for reduction of NOX emissions. Proc. Combust. Instit. 29:2251. Lukachko, S. P., Waitz, I. A., Miake-Lye, R. C., Brown, R. C., and Anderson, M. R. (1998). Production of sulfate aerosol precursors in the turbine and exhaust nozzle of an aircraft engine. J. Geophysical Res. 103:16159. Marinov, N. M., Pitz, W. J., Westbrook, C. K., Castaldi, M. J., and Senkan, S. M. (1996). Modeling of aromatic and polycyclic aromatic hydrocarbon formation in premixed methane and ethane flames. Combust. Sci. Tech. 116/117: 211. Milani, A., and Nelli, D. (1992). Low-NOX combustion techniques applied to steelworks plants firing gas and oil. J. Inst. Energy 65:35. Milne, T. A., Abatzoglou, N., and Evans, R. J. (1998). Biomass gasification “tars”: Their nature, formation and conversion. Technical Report NERL/TP-570-25357, U.S. Department of Energy, National Renewable Energy Laboratory, November. Mustafi, N. N., Miraglia, Y. C., Raine, R. R., Bansal, P. K., and Elder, S. T. (2006). Sparkignition engine performance with “powergas” fuel (mixture of CO/H2): A comparison with gasoline and natural gas. Fuel 85:1605–12. Oakey, J., Simms, N., and Kilgallon, P. (2004). Gas turbines: Gas cleaning requirements for biomass-fired systems. Mater. Res. 7:17. Park, J., Choi, J. G., Keel, S.-I., and Kim, T.-K. (2003). Flame structure and NO emissions in gas combustion of low calorific heating value. Int. J. Energy Res. 27:13391. Radian Corp. (1995). A study of toxic emissions from a coal-fired gasification plant. U.S. DOE Report DE-AC22-93PC93253, Pittsburgh Energy Technology Center, Pittsburgh, PA. Ratafia-Brown, J. A., Manfredo, L. M., Hoffman, J. W., Ramezan, M., and Stiegel, G. J. (2002). An environmental assessment of IGCC power systems. Paper presented at the 19th Annual Pittsburgh Coal Conference, Pittsburgh, PA, September 23–27. Rutar, T., Kramlich, J. C., Malte, P. C., and Glarborg, P. (1996). Nitrous oxide emissions control by reburning. Combust. Flame 107:453. Sarofim, A. F., and Flagan, R. C. (1976). NOX control for stationary combustion sources. Prog. Energy Combust. Sci. 2:1–25. Seinfeld, J. H., and Pandis, S. N. (1998). Atmospheric chemistry and physics. New York: John Wiley & Sons.

Pollutant Formation and Control

191

Shoffstall, D. R. (1977). Burner design criteria for NOX control from low-Btu gas combustion: Ambient fuel temperature. Vol. I. Report EPA-600/7-77-094a, prepared for the U.S. Environmental Protection Agency, Institute of Gas Technology. Shoffstall, D. R., and Waibel, R. T. (1977). Burner design criteria for NOX control from lowBtu gas combustion: Elevated fuel temperature. Vol. II. Report EPA-600/7-77-094b, prepared for the U.S. Environmental Protection Agency, Institute of Gas Technology. Smith, G. P., Golden, D. M., Frenklach, M., Moriarty, N. W., Eiteneer, B., Goldenberg, M., Bowman, C. T., Hanson, R., Song, S., Gardiner, W. C., Jr., Lissianski, V., and Qin, Z. (1999). GRI-Mechanism 3.0. http://www.me.berkeley.edu/gri_mech. Smith, J. A., and Bartley, G. J. J. (2000). Stoichiometric operation of a gas engine utilizing synthesis gas and EGR for NOX control. J. Eng. Gas Turb. Power 122:617. Smoot, L. D., Hill, S. C., and Xu, H. (1998). NOX control through reburning. Prog. Energy Combust. Sci. 24:385. Sorokin, A., Katragkou, E., Arnold, F., Busen, R., and Schumann, U. (2004). Gaseous SO3 and H2SO4 in the exhaust of an aircraft gas turbine engine: Measurements by CIMS and implications for fuel sulfur conversion to sulfur (VI) and conversion of SO3 to H2SO4. Atmos. Environ. 38:449. Stapf, D., Ehrhardt, K., and Leuckel, W. (1998). Modeling of NOX reduction by reburning. Chem. Eng. Tech. 21:412. Streets, W. L. (1963). Gas turbine and jet engine fuels. Progress Report 4 for U.S. Navy, Contract N600 (19)-58219. U.S. Department of Energy. (2002). Major environmental aspects of gasification-based power generation technologies. Final report, National Energy Technology Laboratory, Pittsburgh, PA. U.S. Environmental Protection Agency. (1995). SITE technology capsule: Texaco gasification process. Publication EPA 540/R-94/514a, Cincinnati, OH. Waibel, R. T., and Fleming, E. S. (1979). Development of combustion data to utilize low-Btu gases as industrial fuels. Report EPA-600/7-78-191, prepared for the U.S. Environmental Protection Agency, Institute of Gas Technology. Waibel, R. T., Fleming, E. S., and Larson, D. H. (1978). Pollutant emissions from “dirty” low- and medium-Btu gases. Report FE-2489-48, prepared for the U.S. Environmental Protection Agency, Institute of Gas Technology. Weiland, N., and Strakey, P. (2007). Global NOX measurements in turbulent nitrogen-diluted hydrogen jet flames. DOE/NETL-IR-2007-100, U.S. Department of Energy. Wendt, J. O. L., Sternling, C. V., and Matovich, M. A. (1973). Reduction of sulfur trioxide and nitrogen oxides by secondary fuel injection. Proc. Combust. Instit. 14:897. Wu, K.-T., Lee, H. T., Juch, C. I., Wan, H. P., Shim, H. S., Adams, B. R., and Chen, S. L. (2004). Study of syngas co-firing and reburning in a coal fired boiler. Fuel 83:1991. Zhang, H. R., Eddings, E. G., and Sarofim, A. F. (2008). A journey from heptane to liquid transportation fuels. 1. The role of the allylic radical and its related species in aromatic precursor chemistry. Energy and Fuels 22:945–953. Zhang, H. R., Eddings, E. G., and Sarofim, A. F. (2007). Criteria for selection of components for surrogate of natural gas and transportation fuels. Proc. Combust. Instit. 31:401.

7 Syngas Utilization

Geo A. Richards, Kent H. Casleton, and Nathan T. Weiland

Contents 7.1 Introduction................................................................................................... 193 7.2 Syngas Utilization in Gas Turbines............................................................... 193 7.2.1 Fuel Composition............................................................................... 194 7.2.2 Fuel Purity......................................................................................... 197 7.2.3 Fuel Pressure Drop and Mixing........................................................ 197 7.2.4 Combustor Pressure Drop.................................................................. 198 7.2.5 Emissions Requirements.................................................................... 199 7.2.6 Combustor Configuration..................................................................200 7.2.6.1 Diffusion Flame Combustors..............................................200 7.2.6.2 Premixed Syngas Combustion............................................202 7.2.7 Diluent Options..................................................................................203 7.2.8 Exhaust Aftertreatment.....................................................................204 7.3 Reciprocating Engines...................................................................................204 7.4 Oxy-Combustion............................................................................................206 7.5 Chemical Looping Systems...........................................................................208 7.6 Fuel Cells and Synthesis Gas......................................................................... 210 7.7 Fuel and Chemical Production from Synthesis Gas...................................... 214 7.8 Conclusions.................................................................................................... 216 References............................................................................................................... 217

7.1 Introduction Chapter 1 describes the process of making syngas from solid feedstocks. Syngas can be used in various applications to produce power, or it can be used as a chemical feedstock. This chapter provides an outline of the various system trade-offs that exist when deciding how to use the syngas in specific applications. Both current technology, like gas turbines and reciprocating engines, and emerging technology, like fuel cells, oxy-fuel systems, and chemical looping systems, are reviewed. A brief discussion of syngas use in the chemical and fuel production industries is also given. Subsequent chapters provide more technical details on these various applications.

7.2 Syngas Utilization in Gas Turbines There are many considerations in developing a combustor for syngas turbine applications. Depending on the gasification process and input solid fuel, the syngas 193

194

Synthesis Gas Combustion: Fundamentals and Applications

Table 7.1 Considerations in Syngas Combustor Design and Development Fuel composition

Fuel purity Fuel pressure drop and mixing Combustor pressure drop Emissions requirements Combustor configuration • Premixed • Diffusion flame Diluent option

Exhaust aftertreatment

Affects peak flame temperature and fuel–air ratio for a given firing temperature. Low heating value fuels require more volume flow, changing the fuel manifold volume, injector, and combustor. Trace contaminants should be controlled by gas purification (Chapter 1), but use of exhaust aftertreatment catalysts may require stringent sulfur removal. Maximum fuel pressure is set by the gasifier conditions. Fuel pressure is used to produce jet penetration in premixers and diffusion flame fuel injectors. Higher combustor pressure drop can be used to enhance mixing and improve liner cooling, but reduces cycle efficiency (typical range ~ 3 to 5%). Set by permit requirements; typical ranges, 2 to 25 ppm NOX @ 15% oxygen, with similar CO requirements. Exhaust aftertreatment is an alternative to combustor development. Premixed combustors have not yet been commercialized for syngas but are being developed (see text). Diffusion flame combustors are used with syngas and can operate with different diluents. Current NOX performance with diluents is 15 ppm NOX @ 15% oxygen. Steam generated from low-grade process heat, or nitrogen from the air separation unit. The quantity of available diluent and the effect on engine cycle efficiency should be evaluated. The approaches to mixing either nitrogen or steam in the combustor or fuel manifold are considerations. Both NOX and CO emissions can be controlled by adding catalytic exhaust aftertreatment. The added complexity and effect on efficiency should be compared against combustor emissions improvements.

composition can cover a wide range. Both steam and nitrogen may be available to dilute the fuel to achieve desired mass flow (and power output) and control NOX emissions. The engine may need to meet different emissions goals at different installations. These and other considerations are summarized in Table  7.1, with subsequent sections providing more discussion on each topic.

7.2.1 Fuel Composition Todd (2000) compiled syngas compositions for multiple plants using syngas fuel. The range of syngas properties is presented on a volume percent basis in Table  7.2. Note that the wide variability in the lower heating value (LHV) of undiluted syngas is primarily due to the difference between oxygen-blown gasifier systems and air-blown systems in which the high nitrogen

Table 7.2 Syngas Fuel Variability H2 CO CH4 CO2 N2 + Ar H2O H2/CO LHV   (Btu/ft3)   (MJ/m3) Dilute equiv. LHV   (Btu/ft3)   (MJ/m3)

Min

Max

   8.6   22.3    0.0    1.6    0.2 —    0.33

  61.9   55.4    8.2   30.0   49.3   39.8    2.36

128    5.02

319   12.57

110    4.33

200    7.88

Source: Adapted from Todd (2000).

195

Syngas Utilization

content reduces the volumetric heating value of the syngas. In an oxygen-blown gasifier system, the by-product nitrogen from the air separation unit (ASU) can be used to dilute the syngas after purification, and with both gasifier systems, steam or water can also be injected to control NOX emissions, as discussed further below. Factoring in the effect of the diluent, the equivalent LHV values shown in Table 7.2 still vary by about a factor of 2. In addition, syngas can contain widely varying H2:CO ratios, and in some situations, it may be desirable to allow fuel blends to change at a specific site. For example, in refinery operations, it may be desirable to use excess plant hydrogen in the turbine during specific times, but exclude it at other times. Since the combustor is typically designed for a specific fuel blend, this would complicate the combustor’s operation, and emphasizes the need to develop combustors that can operate on widely variable fuels. Lower heating values of syngas are 1/3 to 1/10 those of typical natural gas heating values in the United States, as shown in Table 7.3. Thus, for the same heat input, the volume flow of fuel is much larger for syngas than for natural gas, and changes appreciably among various types of syngas and syngas dilution scenarios. For a combustor designer, this means that added volume flow must pass through the syngas combustor, requiring larger fuel manifolding, control valves, and injectors. As discussed in Chapter 1, it is usual practice to purify the syngas after cooling to nearly ambient temperature. After purification, the cooled syngas could be used directly in the turbine, but it is often advantageous to reheat the syngas using low-grade heat rejected elsewhere in the power plant. This can be accomplished by heat exchange, or by injecting low-grade steam into the syngas. The steam addition increases the turbine mass flow, producing more turbine power, and acts as a diluent to reduce peak flame temperatures. Thus, it is important to consider the actual syngas conditions (temperature and diluents) when planning the combustor configuration and control.

Table 7.3 U.S. Natural Gas Composition Variability (Species Concentrations Expressed in Mole Percent) Methane Ethane Propane C4 and higher N2 + CO2 HHV   (Btu/ft3)   (MJ/m3)

Minimum

Mean

74.5 0.5 0.0 0.0 0.0

93.9 3.2 0.7 0.4 2.6

Maximum 98.1 13.3 2.6 2.1 10.0

970 36.14

1033 38.46

1127 41.97

Source: Adapted from Liss et al. (1992), as reported by Klassen (2005).

196

Synthesis Gas Combustion: Fundamentals and Applications

Adiabatic Flame Temperature, K

2800

Tfuel = 200 C, no diluent

2600 2400 2200 2000 1800 1600

0.2

2800 Adiabatic Flame Temperature, K

H2 CO CH4

0.4

0.6

0.8 1.0 1.2 1.4 1.6 Equivalence Ratio, φ (a)

1.8

Tfuel = 200 C; 50% N2 dilution

2.0

2.2

H2 CO CH4

2600 2400 2200 2000 1800 1600

0.2

0.4

0.6

0.8 1.0 1.2 1.4 1.6 Equivalence Ratio, φ (b)

1.8

2.0

2.2

Figure 7.1  Comparison of flame temperatures for methane, hydrogen, and carbon monoxide for (a) pure fuels and (b) fuels with 50% nitrogen dilution. Inlet conditions are 200°C fuel, air at 15 atm via isentropic compression (646 K). (Adapted from Casleton et al., 2008.)

The fuel composition also affects the peak flame temperature, which in turn controls NOX formation. Figure 7.1 compares the peak flame temperatures for methane, hydrogen, and carbon monoxide. These temperatures were calculated using Cantera with species thermodynamic information from GRI-Mech 3.0 (Goodwin, 2003; Smith et al., n.d.). Results are shown for 200°C fuel, with air conditions representative of isentropic compression to 15 atm. Notice in Figure 7.1a that the peak flame temperatures for hydrogen and carbon monoxide are approximately 200 K greater than that for methane (the principal component of natural gas). These high temperatures explain why it is difficult to achieve low-NOX performance with undiluted

Syngas Utilization

197

syngas. Figure 7.1b shows the effect of adding 50% nitrogen diluent to the fuel (by volume). This reduces the peak flame temperature significantly, so that diffusion flame combustion temperatures are at least comparable to that of natural gas.

7.2.2 Fuel Purity Chapter 1 describes the process of syngas purification. By the time the syngas arrives at the gas turbine, it is required that gas cleanup has removed particulate, alkali, and metal compounds to levels that are acceptable to the turbomachinery. In addition to these species, coal also contains nitrogen (typically 0.5 to 2 weight percent— dry, ash-free basis) in various molecular forms. This nitrogen can be converted into ammonia, HCN, or molecular nitrogen during the gasification process. The distribution among the different product species depends on the forms of nitrogen in the source coal and the gasification process (Higman and van der Burgt, 2003). Ammonia or HCN not removed during syngas cleanup can be converted to NOX in the turbine combustor. The conversion of these fuel-bound nitrogen species to NOX is nonlinear, depending on factors including input level of these species, combustor pressure, and methane content of the syngas (Giles et al., 2006; Battista and Dudley, 1995; Kelsall et al., 1994). The primary sulfur species H2S and COS in syngas can be readily purified to levels that will meet emissions standards for SO2. However, the use of selective catalytic reduction (SCR) systems to reduce engine exhaust NOX can impose stringent sulfur removal requirements. In SCR systems (see Section 7.2.8), injected ammonia reduces NOX to nitrogen and water over a catalyst bed, and exhaust stream SO2 levels must be controlled to avoid ammonium sulfates that can foul the catalyst surfaces or other downstream equipment. The fuel sulfur levels compatible with SCR aftertreatment have been the subject of debate, but recent experience suggests acceptable SCR performance with sulfur less than 10 ppmv in the fuel gas (de Biasi, 2005).

7.2.3 Fuel Pressure Drop and Mixing In syngas turbines, fuel pressure is established by gasification conditions and control valves ahead of the turbine. Compared to natural gas, the greater volume flow of syngas requires much larger ductwork for the fuel to avoid unacceptable pressure loss, which is why diluents are often not added to the fuel manifold—adding diluent flows can require even larger manifolds. For engines where wide fuel variability is expected, it should be recognized that the injector relies on fuel jet penetration to achieve planned mixing profiles in the combustor (or premixer). Differences in the volumetric fuel heating value will lead to differences in jet penetration. For example, at a fixed heat release, a fuel with a higher volumetric heating value will have a lower volumetric flow rate. For fuels with similar density, the lower volume flow rate means that the fuel jet will have less momentum to mix with surrounding air, and this may change fuel–air mixing profiles. At a fixed heat input, it can be shown (Casleton et al., 2008) that the jet penetration Y for two fuels A and B is inversely proportional to the Wobbe index (WI), defined below:

198





Synthesis Gas Combustion: Fundamentals and Applications

YA WI B = YB WI A WI =

HHV SG

(7.1)



(7.2)

In these equations, YA and YB are the vertical heights of fuel jets injected at a right angle to cross-flow air, where each fuel jet produces the same heat release for fuels A and B. HHV is the fuel higher heating value (on a gaseous standard volume basis, e.g., MJ/m3) and SG is the specific gravity. The Wobbe index can also be defined based on the lower heating value (LHV), leading to confusion when comparing fuels. The gas heating industry usually uses HHV, but the turbine industry often uses LHV. Baukal and Schwartz (2001) discuss how the Wobbe index is used as an interchangeability parameter in naturally aspirated burners. In those applications, constant Wobbe index between fuels ensures that both fuels will have the same heat input for comparable fuel pressure. In gas turbines, the heat input is controlled by the engine throttle, but two fuels with the same Wobbe index will have the same mixing profiles, as above.

7.2.4 Combustor Pressure Drop No matter what fuel is used, it is important to minimize combustor pressure losses. Any pressure lost in the combustor represents energy that cannot be recovered by turbine expansion. Typical engine cycles allow 3 to 5% pressure drop for the combustor. For a given engine pressure ratio, this pressure drop is a function of the mass flow through the combustor (Lefebvre, 1999), which is in turn related to the degree of integration between the gas turbine and the gasifier in an integrated gasification combined cycle (IGCC) plant. In cases where there is full integration, air extracted from the gas turbine compressor for the gasifier’s operation essentially returns to the combustor in the form of increased syngas and nitrogen diluent mass flows. This yields a combustor mass flow similar to that in natural gas-fueled engines (Becker and Schetter, 1993), and hence a similar pressure drop. However, full integration of gas turbine and gasifier operations complicates plant operation; thus, IGCC plant designs often use a separate air compressor for gasifier operations, with 0 to 50% air extraction from the gas turbine (Rosenberg et al., 2005). With reduced or no air extraction from the compressor, the combined air, syngas, and diluent mass flow in the combustor increases by up to 14% over natural gas–fired engines (Todd, 2000; Chisea et al., 2005). Larger-diameter syngas combustors have evolved to reduce the pressure drop resulting from the increased mass flow rates with limited gasifier integration (Jones, 2005). As discussed later, premixed combustors are in a state of development for syngas. The high flame speed of hydrogen in syngas makes it difficult to avoid flashback or flame anchoring within the premixer. This can be minimized by using a large premixer air velocity. Because of the limits on pressure drop, high-velocity premixer flows should be carefully slowed in a diffuser if possible.

Syngas Utilization

199

7.2.5 Emissions Requirements Regulations on permissible emissions of a number of species, including NOX, CO, and SO2, are key considerations in developing the combustor. Actual emission requirements are site specific. They can differ from state to state, and sometimes vary for different regions within individual states. In some situations, operators may be required to apply a best-available control technology (BACT) for emissions control. The U.S. Environmental Protection Agency (EPA) maintains a database with information from state and local permitting agencies on best-available technologies for air pollution control (U.S. EPA, n.d.). For natural gas engines, NOX emissions lower than 10 ppmv @ 15% oxygen can be achieved by lean premixed combustor designs. Stringent NOX emission requirements (i.e., ~2 to 5 ppm) and BACT considerations usually require postcombustion cleanup such as selective catalytic reduction (SCR). For syngas fueled engines, many installations using dilute diffusion flames achieve NOX levels of ~15 ppmv @ 15% oxygen (Jones, 2005). As of 2006, this 15 ppm limit was considered state-of-the-art for syngas-fired engines (Nexant, 2006). While premixed combustion may have an advantage in NOX emissions, it has not yet been developed for commercial syngas applications for reasons explained below. SCR offers the possibility of further reductions of NOX emissions below this 15 ppm value, although as of 2006, SCR application to IGCC systems had not been significantly deployed (Nexant, 2006). More recent air permit applications for construction of IGCC turbine systems call for SCR systems in order to achieve NOX emissions of 5 ppmv or less at 15% O2. As described in Chapter 6, NOX emissions are typically formed via several chemical pathways (Warnatz et al., 1996), with the thermal NOX mechanism often being the dominant pathway in gas turbine combustion. This mechanism is a strong function of the peak flame temperature, and most NOX control strategies involve reduction of the peak flame temperature in one form or another. In diffusion flame combustion, this is typically achieved with nitrogen or steam injection into the combustion zone, while in premixed systems, mixing the fuel with significant amounts of excess air reduces flame temperatures and NOX to acceptable levels (Lefebvre, 1999). Also important in natural gas combustion, the prompt NOX mechanism is a result of chemical reactions involving combustion by-products of hydrocarbons, which are not present in large quantities in syngas (see Table 7.2). This mechanism is therefore typically not a concern in syngas-fired turbines. However, coal-bound nitrogen can produce ammonia or other radicals that can initiate this NOX mechanism unless they are removed from the raw syngas, as discussed in Chapter 6. Reducing NOX emissions in syngas combustion is thus primarily accomplished by suppressing the thermal NOX mechanism; however, other NOX generation routes may then become more important as a result. Konnov et al. (2002) found that the NNH route can be important at fuel-rich conditions in H2/CO flames, as well as in lower-temperature, lean hydrogen flames (Konnov, 2003). The N2O route has also been shown to be a significant NOX generation mechanism at high pressures and lower-temperature conditions (Warnatz et al., 1996; Steele et al., 1995). Low carbon monoxide emissions are a result of high combustion efficiency, which is attained by careful combustor design. In contrast to hydrogen, CO oxidation to

200

Synthesis Gas Combustion: Fundamentals and Applications

CO2 is relatively slow, and requires substantial time at high temperatures to achieve complete combustion. These requirements are in direct conflict with NOX reduction strategies that rely on low combustion temperatures or low flame residence times to reduce NOX emissions. Successful combustor designs must therefore strike a compromise that attains acceptable levels of both pollutants. In natural gas diffusion flame combustors, CO emissions are minimized with about 20% excess air in the primary combustion zone. More than 20% excess air results in lower combustion temperatures and quenching of CO combustion reactions may occur, while less excess air may yield poor fuel–air mixing and regions of incomplete combustion (Lefebvre, 1999). In combustors burning particularly low heating value syngas, it may be necessary to operate closer to stoichiometric conditions than in conventional engines to attain the desired turbine inlet temperature, emphasizing the need to ensure adequate mixing to achieve good CO oxidation (Vogt, 1980).

7.2.6 Combustor Configuration There are two primary choices for syngas combustor design: diffusion flame combustion or premixed combustion. The latter option is not yet commercially available for syngas, but there are several advantages to premixed designs that motivate research and development. Both styles of combustor are discussed below. 7.2.6.1 Diffusion Flame Combustors Diffusion flame combustors were used in most stationary gas turbines prior to 1990. They have an advantage over premixed combustors in that the flame is usually very stable, and there is no possibility of flame flashback. However, without dilution, the flame temperature is high enough to produce significant NOX emissions. Significant experience exists in developing diffusion flame combustors, and they continue to be used nearly exclusively in aeroengine applications. The reader may consult standard texts for details on how to configure a diffusion flame combustor (e.g., Lefebvre, 1999). Specific differences for syngas combustors are reviewed here. In a traditional diffusion flame combustor, fuel is injected into the primary zone, where the stoichiometry is near unity. High primary zone temperatures help to stabilize the flame and complete CO oxidation, but also can produce appreciable NOX. Additional air is added to the primary zone products via secondary and (sometimes) tertiary air jets entering along the axis of the combustor. The additional air completes combustion and achieves the desired turbine inlet temperature—but must be designed to achieve specified temperature profiles at the turbine inlet (defined by the pattern and profile factors for the temperature profiles along the blade, and around the turbine annulus, respectively). Compared to natural gas or liquid fuel diffusion flame combustors, the key distinctions for syngas applications are the added volume flow associated with low heating value fuels and the potential to use diluents to control the primary zone flame temperature. Development of syngas turbine combustors dates back into the 1960s, as reviewed by Becker and Schetter (1993). Vogt (1980) and Beebe and Blanton (1985) presented many of the issues, which can be summarized as follows.

Syngas Utilization

201

Compared to conventional fuels, the greater volume and mass flow of the syngas fuel means that the temperature of the fuel becomes a significant consideration in the overall firing temperature. Also, early tests of low-Btu syngas combustors showed that although the firing temperature was lower than natural gas applications, liner temperatures were higher due to the increased mass flow and velocity in these early combustors (Beebe and Blanton, 1985). In addition, air extracted from the compressor discharge for gasifier operations may come at the expense of the cooling air, also raising liner temperatures. Battista and Dudley (1995) and Cook et al. (1995) describe how a conventional combustor could be modified to use syngas with diluents. Studies were conducted with syngas compositions from air-blown and oxygen-blown gasifiers. Three diluents (steam, nitrogen, and carbon dioxide) were compared. The diluents were added in two different ways: either with the fuel or separately in the primary zone of the combustor. The results showed that NOX levels were correlated with peak flame temperatures, and did not change with diluent at a given flame temperature. The results also showed that there was not a significant difference when the diluents were added to the fuel versus the combustor air. This is a very useful result because it is simpler to add the diluent to the airstream. Jones (2005) notes that in practical designs, the fuel manifold volume is already very large for syngas fuels, and would need to be even larger to accommodate fuel dilution. Fundamental studies suggest that there may yet be an (unexploited) emissions advantage to diluting the fuel versus the air. Feese and Turns (1998) studied the location of fuel dilution in boiler applications, and note that multiple factors must be considered when assessing the effect of fuel- versus air-side diluents, with special care taken to assess the effect of changing the flame residence time on NOX. Chen and Driscoll (1990) and Gabriel et al. (2000) showed that NOX emissions from laboratory diffusion flames correlate with the flame volume divided by the fuel volume flow rate (a characteristic flame residence time), and may be affected by the Lewis number of the fuel. Marek et al. (2005) showed that multipoint diffusion flame injectors using hydrogen can produce NOX levels that are comparable to state-of-the-art lean direct injection (LDI) schemes for liquid fuels, using low overall equivalence ratios without diluents to limit combustor exit temperatures. These studies are the motivation for ongoing research into diffusion flame designs for syngas. Preliminary research (Weiland et al., 2007) has shown that NOX emissions for diluted hydrogen flames have a distinct benefit when the fuel (not the air) is diluted with nitrogen. These benefits would generally be realized in cases where excess air is supplied to the primary combustion zone, which is contrary to common practice for traditional combustors, but standard for LDI combustors (Marek et al., 2005). In these cases, NOX reduction with dilution of the fuel is attributed to reductions in both the peak flame temperature and flame residence time. Dilution of the fuel stream ensures that all of the diluent passes through the flame front to serve as a heat sink and thus reduce peak flame temperatures, whereas dilution of the air will allow some of the diluent to bypass the flame if air is present in excess. In addition, diluting the fuel increases fuel–air mixing in the presence of excess air and results in smaller flame volumes, while also increasing the volume flow rate of the fuel.

202

Synthesis Gas Combustion: Fundamentals and Applications

Both of these effects decrease the flame residence time and hence its NOX production (Weiland et al., 2007). Further, the high diffusivity of hydrogen out of a diluted hydrogen fuel jet can be used to create regions of higher hydrogen content in the immediate vicinity of the fuel injection point than can be attained with dilution of the airstream. This improves the stability of the flame anchoring point, which then allows for faster mixing and reduction of flame residence times for further reducing NOX emissions. These results may only be beneficial in combustion of high hydrogen fuel, however, as the low residence times and rapid mixing with excess air are more likely to result in unacceptable levels of unburned CO using a syngas fuel. 7.2.6.2 Premixed Syngas Combustion Since the late 1980s and early 1990s, requirements to reduce NOX emissions have brought about the implementation of premixed combustion approaches in stationary gas turbines. As the name implies, fuel and air are premixed prior to entering the combustion chamber. Introducing adequate excess air reduces the peak flame temperature sufficiently to avoid the production of excessive levels of thermally generated NOX. Potential drawbacks to this approach, however, include flame flashback, autoignition, and combustion dynamics. In large part, these issues have been addressed for natural gas-fired engines. However, the high hydrogen content in typical syngas has complicated the resolution of these issues for syngas applications. During flashback, the flame, which is normally anchored in the lower-velocity region of the combustion chamber, propagates upstream into the higher-velocity region of the premixer. This can be a problem for systems that are hydrogen fueled because hydrogen’s combustion kinetics are much faster than the kinetic rates for other hydrocarbon fuels, such as natural gas. In addition, the mass and thermal diffusivities of hydrogen are high compared to methane. As a result, the laminar flame speed for hydrogen in atmospheric pressure air is nearly an order of magnitude larger than that for methane. Flashback into the premixer can lead to catastrophic failure because those components are more susceptible to overheating and damage. Tests of flame stability in swirl-stabilized flames show that the operating range of the hydrogen fuel–air ratio between the limits of flame blow-off and flashback is much narrower than that for natural gas (Noble et al., 2006; Straub et al., 2006). This suggests that premixed systems fueled with hydrogen may require careful control of the fuel– air ratio for successful operation in the region bounded by blow-off and flashback. Autoignition refers to the spontaneous ignition of premixed fuel and oxidant without the introduction of an external ignition source. When this phenomenon takes place in the fuel–air premixer, overheating can occur with resulting damage to components. Detailed knowledge of ignition delay characteristics is required for the fuels of interest to minimize autoignition problems as well as maintain stable combustion conditions (Lieuwen et al., 2008). Efforts to resolve these issues for syngas fuels include measurements of ignition delay times (Kalitan et al., 2007; Mittal et al., 2006) and development of detailed chemical kinetic mechanisms that are validated with a variety of experimental measurements for syngas fuels (Sung and Law, 2008; Chaos and Dryer, 2008). Modern premixed gas turbine combustion systems usually operate fuel-lean, close to the combustor’s blowout margin in order to control NOX emissions. As a result, they

203

Syngas Utilization

can be very susceptible to flow fluctuations, leading to unstable combustor behavior and combustion dynamics. Combustion dynamics are produced by the coupling of unsteady heat release and pressure waves in combustion systems. The amplitude of these pressure oscillations can be high enough to affect operational stability or lead to failure of hardware components. A variety of different physical mechanisms have been identified that can drive combustion instabilities in gas turbine systems. Lieuwen and Yang (2005) provide more details on the fundamental mechanisms as well as modeling, diagnostics, and possible control schemes for combustion instabilities. Susceptibility to combustion dynamics can be influenced by fuel composition or heating value, which could affect the combustion kinetics and subsequently impact the flame shape or location (Lieuwen et al., 2008). Properties and compositions of candidate fuels for low-emissions gas turbine are tightly specified, and deviation from these fuel specifications can have effects ranging from poor operational performance to possible hardware damage. Table 7.2 suggests that considerable variability is possible for syngas fuel composition and heating value; thus, it is possible that premixed syngas combustors will be limited to specific fuel compositions or require changes in design or operation to accommodate different syngas compositions.

7.2.7 Diluent Options The source and quantity of diluent available to the combustor (either nitrogen or steam) depends on the power plant configuration. As described in Chapter 1, the gasification process is often oxygen blown, meaning that an air separation unit (ASU) is used to provide oxygen. The remaining nitrogen can then be used in the combustor as a diluent.* As noted in Section 7.2.4, there are various approaches to integrating the oxygen and nitrogen production in the gas turbine. These are described by Smith et al. (1997) and Smith and Klosek (2001), and range from complete integration (where the air from the gas turbine compressor is used in the ASU) all the way to independent operation (where a separate compressor for air, nitrogen, and oxygen is used). The main point for the combustor design is to note that nitrogen is available for diluting the flame. The quantity of available nitrogen can be estimated as follows. Assume that the coal has an approximate C:H ratio of 1:1. Then, the stoichiometries for the ASU, gasifier, and fuel are as follows:

ASU:

Air → O2 + 3.76N2



Gasifier:

CH + ½ (O2)from ASU → CO + ½ H2



Fuel dilution: CO + ½ H2 + ½ (3.76 N2)from ASU → diluted fuel

In the last equation, the nitrogen mole fraction on the left side is 56%. A similar calculation using water gas shift and CO2 separation (to create hydrogen fuel) will produce comparable results. Thus, as a rule of thumb, there is a practical limit of about 50% dilution of the fuel with nitrogen from the ASU. *

Alternatively, air-blown gasifiers already have the fuel significantly diluted with nitrogen.

204

Synthesis Gas Combustion: Fundamentals and Applications

Where even greater dilution is needed, steam can be used, either as the primary diluent or in combination with the nitrogen diluent. As explained by Smith et al. (1997), up to 20% moisture can be added to the dry nitrogen, an approach that has been used at the Tampa Electric Polk Power IGCC plant (Geosits and Schmoe, 2005). Alternatively, waste heat from the gasification process can also be used to raise lowgrade steam, which can be directly added to the combustor as a diluent. Adding steam to the combustor contributes to the power output from turbine expansion, but is usually not highly efficient because the turbine pressure ratio is much smaller than the steam cycle pressure ratio. Nevertheless, the combination of reducing NOX emissions and producing power makes steam addition attractive. As a NOX control diluent, steam is superior to nitrogen due to both its higher specific heat, which reduces thermal NOX, and its increased chemical reactivity, which can reduce NOX from both the prompt and N2O mechanisms (Giles et al., 2006; Steele et al., 1995). One drawback to steam or moisture addition is that the water ends up in the exhaust flow and is consumed by the power plant. Water use has recently become a greater concern in power plant operation (e.g., Couch, 2005). In addition, increasing the steam content in the combustion products increases the heat transfer to downstream components, which can result in reduced turbine blade and thermal barrier coating lifetimes. To maintain turbine component reliability when steam dilution is employed, turbine inlet temperatures are typically reduced to yield metal temperatures that are consistent with natural gas operation (Chiesa et al., 2005; Jones, 2005).

7.2.8 Exhaust Aftertreatment If acceptable levels of pollutant emissions cannot be attained within the combustor, then postcombustion controls can be added to reduce NOX and CO emissions. Aftertreatment of CO emissions typically involves oxidation over a platinum or palladium catalyst, which can reduce emissions to 2 to 10 ppm levels. As mentioned above, NOX emissions are commonly reduced using SCR in natural gas combinedcycle plants, where ammonia is injected and reactions occur over a catalyst to reduce NOX to N2 and water. These reactions occur over a narrow temperature range from 560 to 670 K (550 to 750°F); thus, the SCR process is often implemented in heat recovery steam generators attached to the gas turbine exhaust (Lefebvre, 1999; Nexant, 2006). Important considerations with the SCR process include size, cost, fouling of the catalyst with sulfur compounds, potential emissions of ammonia (also an EPA-controlled substance), and disposal of the spent catalyst, which can also be a hazardous material (Babcock and Wilcox, 2005).

7.3 Reciprocating Engines Reciprocating internal combustion (IC) engines represent a well-established technology for power generation. Benefits of the technology include: • Relatively low capital cost • Proven reliability • Good part-load performance

Syngas Utilization

205

• High operating efficiency • Modularity, resulting in shorter delivery and setup time than other competing technologies As a result, reciprocating internal combustion engines play a key role in standby and distributed energy power generation. Conventional fuels for these engines include gasoline, diesel fuel, or natural gas. However, there is also interest in the application to a variety of unconventional fuels, including biogas (Chellini, 2007) as well as waste gases from coke ovens and steel mills (“Waste Gas-Burning Engines,” 2006). Biogas (e.g., gas from landfills, anaerobic digesters, etc.) tends to be predominantly methane and CO2, while waste gases from coke ovens and steel mills can have high concentrations of H2 or CO. In contrast, the application of IC engine technology to the utilization of syngas fuels has received much less attention. Boehman and Le Corre (2008) reviewed work done with the use of syngas in internal combustion engines, with particular emphasis on dual-fuel diesel applications. For coal syngas, a key consideration for gas engine application is the size of the engine compared to the size of gasification plant that is required for overall system economics. Gas engine sizes are typically 10 to 20 MW or less, while coal gasification systems are generally 200 MW or more. In general, biomass gasification systems are smaller than coal-based systems, and so are better fits for reciprocating engines. In one assessment (Antares Group, 2003) that evaluated the feasibility of eight different technologies with biomass gasification, results indicated that internal combustion engines demonstrated proven technical feasibility and could be cost competitive with high-cost natural gas–distributed generation plants. Some commercial experience with these systems exists in Europe, although none existed in the United States at the time of this report (2003). A number of engine manufacturers were contacted to obtain feedback on potential technical concerns regarding utilization of low heating value, bio-derived syngas in these engines (Antares Group, 2003). Some of the concerns that were expressed include the low heating value of the fuel, the changing heating value, and the problem of contaminants (tars and liquids, high moisture levels, particulates, and so forth). In the late 1980s, the technical feasibility of a high-pressure, coal-gas-fed diesel engine system was assessed (Greenhalgh, 1992a, 1992b). This system was based on air-blown, fixed-bed, high-pressure processing of run-of-mine coal to produce low heating value syngas for use in an ignition-assisted, high-compression diesel engine. The long-range application was to line-haul railroad locomotives. Proof-of-principle rig tests were performed with a synthetic low heating value gas with a glow-plug-assisted diesel engine. These tests demonstrated the feasibility of the concept and identified some areas requiring additional development. A number of the shortcomings were due to the batch nature of the coal processor rather than the utilization of syngas in the engine (e.g., achieving acceptable engine performance with a fuel supply that varies in temperature, composition and energy content, and operation on a nonsteady-state duty cycle characteristic of locomotive engine applications). Additional studies with the batch-operated fuel processor showed promise, but demonstrated the need for sulfur and particulate capture in this application.

206

Synthesis Gas Combustion: Fundamentals and Applications

7.4 Oxy-Combustion The desire to advance the technology of carbon capture and storage is fostering interest in oxy-combustion approaches, including those that involve syngas. In these processes, combustion is accomplished using oxygen rather than air. Products of combustion for a hydrocarbon fuel are then simply water and CO2. Separation of water from the postcombustion products generates a stream of highly concentrated CO2. This CO2 product stream can be sequestered or used in other applications, such as enhanced oil recovery or other chemical processes. This technology has been demonstrated for natural gas power cycle applications (Anderson and Pronske, 2006) and examined for pulverized coal power generation systems (Buhre et al., 2005). However, an oxy-fuel cycle could also operate in syngas applications. Figure  7.2 shows two possible implementations of these concepts in power cycles. For both the CO2-diluted and steam-diluted oxy-fuel examples shown in Figure 7.2a and b, respectively, the nature of the working fluid in the turbine is considerably different from that in conventional applications. The high CO2 and steam contents will potentially result in substantial changes in important properties of the working fluid, such as specific heat and radiative and convective heat transfer properties. Chiesa et al. (2005) discusses some of the impacts of these changes on turbine operation. As noted above, an essential characteristic of oxy-combustion schemes is the replacement of combustion air with oxygen. While this produces an exhaust stream that is almost exclusively CO2 and H2O, the production of highly enriched O2 for consumption in the combustion process can be a substantial cost penalty, in terms of both capital investment and operating expenses for the plant. Successful application of this technology will require combustor operation at near stoichiometric conditions to reduce the oxygen demand and careful combustor design to maximize efficiency of mixing of fuel and oxygen. Minimizing excess oxygen in the exhaust stream is also beneficial from another standpoint; excess oxygen represents a potential corrosion concern for sequestration applications that transport CO2 via pipeline. There is considerable variation in this recommended O2 limit, ranging from 100 ppm to less than 2 ppm (Moreira, 2006). Consequently, it is desirable to minimize excess oxygen in the combustor. Regardless of the specific choice of cycle configuration, the desire to operate the combustor as close as possible to stoichiometric conditions is a challenge for designers of oxy-fuel combustion systems. In addition, a lack of fundamental data in the literature for dilute oxy-fuel systems further complicates the task of designers to minimize excess oxygen in these systems. The scarcity of relevant data is due in part to the lack of prior technological development in this area that would have otherwise driven the generation of applicable information. Furthermore, it is difficult to study flames where the reactants are mostly steam, since water easily condenses where oxygen or fuel is added without proper preheating. For these reasons, simple properties like flame speeds have not been documented for steam-diluted oxy-fuel flames. CO2-diluted systems, in contrast, have been reported, at least for simple laminar flames (Lewis and von Elbe, 1987). Kinetic mechanisms used to estimate reaction rates have likewise not been validated with high levels of H2O or CO2 in nearly stoichiometric flames.

207

Syngas Utilization O2 Syngas

Combustor

CO2

Condenser

CO2 Liquid H2O

(a) O2 Syngas

Combustor

Liquid H2O Condenser

CO2

(b)

Figure  7.2  Syngas use in (a) CO2-diluted and (b) H2O-diluted oxy-fuel power cycles. (Adapted from Casleton et al., 2008.)

Accurate kinetic data are especially important for these systems because operation near stoichiometric necessarily involves kinetically limited reactions as both oxygen and fuel ideally approach very small concentrations. Equally important, adequate mixing between the fuel and oxidizer is required in practical combustor designs because even a small mismatch in the fuel or oxygen distribution will lead to pockets where either fuel or oxygen cannot be consumed. From a combustion standpoint, available kinetic models predict that CO oxidation will be much slower when CO2 is used as the primary diluent (Richards et al., 2005), so the combustor design would need to have a longer residence time to complete CO oxidation.

208

Synthesis Gas Combustion: Fundamentals and Applications

It should be noted that the oxy-fuel combustion approach solves one important problem inherent to most combustion systems: emissions controls. Even with modest oxygen purity, the nitrogen concentration in the oxy-fuel system will be much lower than in a comparable syngas-air engine, thereby producing very low NOX levels. Even if this were not the case, sequestration of the (mostly) CO2 exhaust would dispose of any pollutants, making oxy-fuel systems very attractive from the standpoint of both CO2 and traditional emissions controls.

7.5 Chemical Looping Systems Chemical looping is a relatively new approach to power generation that can use any hydrocarbon fuel and produce a separate stream of CO2. The process is similar to oxy-fuel combustion because the fuel is not mixed with atmospheric air, but with oxygen supplied by metal oxides, avoiding the need for oxygen production by an air separation unit. This is shown schematically in Figure  7.3. The metal (Me) is sent to a reactor vessel where air reacts with the metal to create the metal oxide (MeO), which is then sent to the fuel reactor. In the fuel reactor, the metal oxide is reduced back to a metal, and the oxygen reacts with the fuel to create CO2 and water. Depending on the metal, both the fuel and air reactions can be exothermic, so that steam can be generated from heat released in either reactor. Chemical looping systems can be used for both gasification and combustion. In a simple description of the gasification process, the fuel reactor in Figure 7.3 operates without enough oxygen, so that syngas is produced rather than combustion products. This process is sometimes called chemical looping reforming, and has been demonstrated as a technique to reform natural gas by Ryden et al. (2006). Variations on this concept have been studied that allow separation of hydrogen and CO2, using coal CO2 H2O

Air (Hot)

Metal oxide (MeO)

Metal (Me)

Air (Ambient)

Fuel

Figure 7.3  Chemical looping power system. (Adapted from Casleton et al., 2008.)

Syngas Utilization

209

as the feedstock. Rizeq et al. (2003) described a system using three reactors where coal is gasified using steam, with CO2 being absorbed by circulating solids that are regenerated in a second reactor, producing pure streams of CO2 and hydrogen from coal from the first and second reactors. A third reactor provides an oxygen carrier needed to supply oxygen for char combustion in the second reactor. Andrus et al. (2006) describe a hybrid combustion-gasification system that uses chemical looping either as a combustor or as a gasifier with inherent hydrogen separation. Again, the process can use coal as a feedstock, and may include direct separation of the CO2 and hydrogen. If continued development is successful, these research projects may lead to alternative paths to create syngas or hydrogen, without requiring an oxygen separation plant. Although the subject of this book is syngas utilization, chemical looping combustion of coal is mentioned as an alternative to creating and burning syngas. Chemical looping combustion of gaseous fuel has been successfully demonstrated in laboratory studies (Abad et al., 2006; Johansson et al., 2006a, 2006b). Laboratory studies and process concepts for solid fuel chemical looping combustion are also being investigated, and show promise as a simple method for using solid fuel to raise steam, with inherent CO2 separation (Lyon and Cole, 2000; Cao and Pan, 2006; Tian et al., 2008; Tobias et al., 2009). Chemical looping combustion (CLC) of synthesis gas, generated by conventional gasification processes, is also an approach to generate power with inherent CO2 separation. In this approach, the fuel reactor in Figure 7.3 is supplied with syngas, and power could be generated by a steam cycle. The chemical looping process could alternatively be conducted at pressure, in a gas turbine cycle. There have been a number of recent studies comparing the system configurations for both natural gas and synthesis gas systems. Jin and Ishida (2000) presented a chemical looping concept that combines coal gasification and chemical looping combustion in a turbine cycle. Without accounting for CO2 compression, the authors suggested an LHV efficiency of 51% for the CLC approach, which compares favorably to a more conventional IGCC system having 46% efficiency. For natural gas fuel, Naqvi et al. (2004) evaluated chemical looping applied to both steam and combined-cycle power plants. The combined-cycle plant could achieve 50% LHV efficiency, including the energy penalty to compress separated CO2 to 100 bar for sequestration. As pointed out by Anheden and Svedberg (1998), chemical looping systems can be arranged to actually reduce the exergy loss that is associated with traditional combustion approaches. If a heat engine can efficiently utilize the available energy (exergy) of the fuel, this would be an advantage. In practice, the peak temperature of the chemical looping process is limited by the properties of the oxygen-carrier metal. For most existing heat engines, like combined-cycle turbines, it is more beneficial to have a peak cycle temperature well above the temperature limits of proposed oxygen-carrier metals and oxides. For this reason, Consonni et al. (2004) analyzed a chemical looping combined cycle with an optional combustor, fired after the chemical looping process. The combustor serves to raise the peak temperature to increase the cycle efficiency, but also adds some CO2 to the exhaust stream. With the fired combustor, the cycle efficiency was as high as 52% LHV. The additional combustor released CO2 equal to about half that of a comparable natural gas combined cycle.

210

Synthesis Gas Combustion: Fundamentals and Applications

Lozza et al. (2006) also evaluated a fired combustor option in a combined cycle, but used hydrogen for the supplemental fuel, avoiding CO2 emissions. The hydrogen comes from a three-reactor chemical looping system where iron is oxidized with steam to produce hydrogen, and the iron oxide is then reduced with natural gas. This approach produced a combined-cycle efficiency of 51% LHV, accounting for CO2 compression to 150 bar. A comparable baseline natural gas combined cycle operating at the same conditions was 57% LHV. A key aspect to developing chemical looping systems is defining the performance and durability of the metal oxide (usually referred to as the oxygen carrier). Numerous carriers have been studied (Jerndal et al., 2006), including some analysis of the effects of coal gas impurities like sulfur, and also conversion under pressurized conditions (Siriwardane et al., 2006; Garcia-Labiano et al., 2006). The combination of a high-performance carrier and an efficient system configuration is the subject of current research that may lead to an efficient chemical looping system for syngas use. As with oxy-combustion systems, these chemical looping systems also avoid many emissions control requirements of standard combustion systems by virtue of producing a sequestration-ready waste stream containing CO2 and other trace pollutants.

7.6 Fuel Cells and Synthesis Gas As described in the Fuel Cell Handbook (2004), there are five major types of fuel cells that operate on gaseous fuels, and a sixth that operates on methanol. The fuel cells are distinguished by the type of electrolyte that separates the fuel and air, and can be further divided into two subcategories: low temperature and high temperature. The low-temperature fuel cells are listed below, along with their commonly used acronyms and operating temperature range. Except for the methanol cell, these systems are designed to use hydrogen as a fuel: • • • •

Polymer electrolyte membrane fuel cell (PEM, 40 to 80°C) Alkaline fuel cell (AFC, 65 to 220°C) Phosphoric acid fuel cell (PAFC, 205°C) Direct methanol fuel cell (DMFC, 60 to 100°C)

Two other fuel cell types can be used directly with synthesis gas. Synthesis gas can be created by reforming any hydrocarbon fuel (natural gas, transportation fuels, etc.), but the emphasis in this book is syngas from solid fuel gasification. The fuel cells capable of using syngas are as follows: • Molten carbonate fuel cell (MCFC, 650°C) • Solid oxide fuel cell (SOFC, 600 to 1000°C) Details on each type of fuel cell can be found in Vielstich et al. (2003). The discussion in this book centers on integration of the solid oxide system with coal synthesis gas. As explained below, the higher SOFC operating temperature has a potential for higher efficiency when combined with a heat engine, and is currently being developed for integration with coal gasification systems (Surdoval, 2007).

211

Syngas Utilization H2O, CO2 + Residual fuel

Syngas (H2, CO) Anode side O2–

Electrolyte

e– –

e

Electric load

Cathode side Air (O2, N2)

Figure 7.4  Solid oxide fuel cell schematic. (Adapted from Casleton et al., 2008.)

A schematic of an SOFC fuel cell is shown in Figure 7.4. Air flows across the cathode, where oxygen reacts electrochemically with electrons returning from the electrical load to create oxygen anions (O2 + 4e – → 2O2–). These anions are essentially forced through the electrolyte by the concentration gradient that exists in the electrolyte because anions react with fuel at the anode to create H2O. The electrons are supplied at a voltage E that is ideally given by the Nernst equation, shown below for the simplest case of pure hydrogen fuel:

E = E0 +

Ru ⋅ T P ⋅ P1/ 2 ⋅ ln H 2 O 2 2F PH 2O

(7.3)

The term E0 is the standard potential for the reaction, meaning the voltage that exists for standard conditions of temperature and pressure for all reactants and products. The term on the right accounts for the actual reactant and product partial pressures and their temperature T. The constants Ru and F are the universal gas constant and the Faraday constant. If the water exists as a gaseous product, the standard potential for H2/O2 fuel cell reaction is 1.18 volts. This is the ideal voltage and can be considered an open-circuit voltage. The actual operating voltage will be lower because of various losses connected with current flows and fuel conversion; more details are found in Chapter 11 in this book. The main points for this discussion are as follows:



1. A single cell can typically produce a potential of 1 V or less. Thus, cells must be arranged in “stacks” to create voltage levels that are technically useful. As a result, the architecture for the stack is a significant consideration in routing fuel, air, and electrical connections to the collection of stacks needed to generate many megawatts of power. 2. Not all of the fuel can actually be used in the cell. As the fuel is consumed, the partial pressure of the hydrogen in the Nernst equation approaches zero, and the logarithmic term becomes negative, meaning the voltage will drop below desired levels. Thus, for practical reasons, only about 80% of the fuel is used, and the remainder must be oxidized downstream by direct mixing

212



Synthesis Gas Combustion: Fundamentals and Applications

with air or oxygen. For high efficiency, the heat released by fuel combustion should be used in some bottoming heat engine, discussed later. 3. Because of irreversibility, significant heat is generated during electricity production by the fuel cell. Resistive losses in the cathode, anode, and electrolyte are combined with so-called overpotentials at the cell interfaces to reduce the cell efficiency from ideal conditions. The heat release may represent 50% of the fuel heating value, and must be managed by providing enough cathode airflow to cool the stack, or (possibly) by endothermic fuel reforming reactions on the anode. Although the heat represents irreversible losses in the fuel cell, for SOFCs, this heat is produced at high temperatures, and can be used as high-grade heat to produce power in a heat engine.

The three points discussed above are true for SOFC systems using any syngas. An issue that is particular to coal synthesis gas is that the amount of carbon monoxide is greater than in syngas generated, for example, by reforming natural gas. This means that care must be taken to avoid forming carbon on the anode, a condition that can typically be met by adding steam to the fuel to avoid the carbon-forming region. As discussed in Chapter 1, most syngas cleanup systems operate at “cold” temperatures and will produce moisture-free syngas, so the addition of steam needs to be considered in the fuel preparation. There are multiple approaches to gas purification, and each approach has distinct pros and cons that must be evaluated, based on the requirements for gas cleaning (Chapter 1). Defining specific gas cleanup requirements for SOFCs is the subject of current research (Trembly et al., 2007a, 2007b, 2007c) and must consider all the impurities that might exist in coal syngas, including As, Pb, Hg, S, Cl, and compounds of these substances. Also noted above is the need to manage thermal energy in the SOFC stack, to keep the temperature within the operating limits of the cell and stack. This can be accomplished by flowing enough air through the stack to remove excess heat. For SOFC systems that require a small temperature change across the stack, the needed cooling flow usually requires about five times the amount of air as is required for stoichiometric reaction of the fuel and air. Moving this much air can require a significant energy input for coal-based systems, and it may be advantageous to evaluate other methods to cool the stack. In some situations, it has been shown that there is a benefit in using endothermic fuel-reforming reactions to cool the stack of a natural gas–fueled system (TAIX LLC, 2003), so less air must be pumped through the cathode to cool the stack. In natural gas–fueled systems, this approach is known as internal reforming and has been practiced in various fuel cells (Blomen and Mugerwa, 1993). For most syngas applications, the fuel is predominantly a mixture of hydrogen and carbon monoxide, and there is no possibility for significant internal reforming. This eliminates the potential to cool the stack via reforming the fuel. From this standpoint, a gasifier with high methane content could be advantageous for fuel cell systems because it could reduce the airflow requirements of the fuel cell. Chapter 1 notes that methane content is significant only for low-temperature gasifiers, but fuel cell systems must then utilize or remove any tars produced by lower-temperature gasification.

213

Syngas Utilization 4 Turbine Condenser Air Water

Rejected heat

1 3

Pump

Fuel cell

2

Cathode 700°C

Recuperator

0°C

Fuel cell 700°C 2

Compressor 0

0°C Air In

(a)

5

1

Exhaust

700°C

850°C

Cathode

3

850°C

Turbine 4 (b)

Figure  7.5  Idealized hybrid fuel cell power cycles using (a) Rankine and (b) Brayton cycles. (Adapted from Casleton et al., 2008.)

The high-temperature heat produced by the fuel cell can be used to operate a heat engine, producing a combined cycle that has a greater efficiency than the fuel cell alone. There are multiple ways to configure such a system, including using a steam (Rankine) cycle or a turbine (Brayton) cycle; a comparison of the options is found in the Fuel Cell Handbook (2004). Figure 7.5 shows a schematic comparing a steam cycle configuration to a turbine cycle configuration. In the Rankine cycle of Figure 7.5a, air enters the system and is warmed by a heat exchanger to a temperature sufficient for the cathode entrance (~700°C). The fuel cell rejects heat to the air, raising the temperature in this example to 850°C, and the heat is then used to raise steam for the Rankine cycle. Not shown, the anode products contain unused fuel, so that combustion with air may raise the temperature even higher than the 850°C used in this example. State-of-the-art steam cycles operate around 600°C, so the finite temperature difference between the fuel cell exhaust (850°C) and the steam cycle peak temperature (600°C) introduces a loss in exergy that cannot be recovered.

214

Synthesis Gas Combustion: Fundamentals and Applications

In contrast, the recuperative gas turbine cycle shown in Figure  7.5b accepts the fuel cell exhaust at 850°C and expands it in the turbine to the temperature of the heat exchanger. Thus, the hot gas from the fuel cell exhaust is used at the highest temperature, providing a potential efficiency advantage compared to the steam cycle. The disadvantage of the turbine cycle is that the fuel cell is pressurized, and losses in the turbomachinery must be accounted for when comparing efficiency. The trade-offs between these various cycle arrangements are the subject of multiple studies (Fuel Cell Handbook, 2004; Veyo et al., 2002; Rao et al., 2004; Campanari and Macchi, 1998), but no study to date has shown specific advantages for a system configuration that accounts for integration with the coal gasifier. One interesting advantage of the SOFC fuel cell is that the anode flow can be used to separate CO2. By keeping the anode flow separate from the cathode, the anode exhaust contains just CO2 and H2O, along with any unused fuel. If the fuel can be oxidized with oxygen from an air separation unit, it is easy to separate the exhaust CO2 by simply condensing the water. Kvamsdal et al. (2007) compared the efficiency of natural gas systems that separate CO2, comparing chemical looping, oxy-fuel combustion, a hydrogen turbine (reforming the fuel to H2), and a hybrid turbine–fuel cell system. Among these types of systems, the hybrid turbine–fuel cell was shown to have the highest efficiency, with inherent separation of CO2. A similar study for coal-fueled systems has not been carried out to date, and would need to account for thermal integration of the gasifier with the rest of the power system.

7.7 Fuel and Chemical Production from Synthesis Gas Although the emphasis of this book is on the use of syngas for power production, it should be recognized that syngas plays an even larger role in chemical and fuel production. The most recent industry survey (Gasification World Database, 2007) shows that 45% of syngas is used for chemical production, 28% is used to produce liquid fuels (via Fischer-Tropsch route), and 19% is used for power production. The remaining 8% is used to produce gaseous fuels. Because of rising prices for petroleum and natural gas, solid feedstocks and gasification technology are receiving even greater interest from the chemical industry (Tullo and Tremblay, 2008). Where chemicals or fuel are produced from syngas, it is often useful to combine power production with the chemical production. Thus, even though power production is a smaller part of syngas use, it is expected to play an important role in future chemical or fuel plants. As noted above, one of the largest uses of syngas today is chemical and liquid fuel production by the Fisher-Tropsch (FT) method. The FT process was developed in Germany during the 1920s by Franz Fischer and Hans Tropsch, and can be represented by two global reactions (Mako and Samuel, 1984):

2n H2 + n CO ←→ (--CH2--)n + n H2O + heat



n H2 + 2n CO ←→ (--CH2--)n + n CO2 + heat

The expression in parenthesis (--CH2--) represents hydrocarbons with an approximate C:H ratio averaging 1:2, but may include species ranging from C1 up to C60,

215

Syngas Utilization

including both paraffins and olefins, as well as alcohols. Thus, the global reactions listed above represent a complex process that is promoted over iron or cobalt catalysts at temperatures of 225 to 365°C and pressures of 5 to 40 atm (Probstein and Hicks, 1990). The combination of operating conditions and the type of catalyst determine the selectivity toward desired products, and the process usually requires upgrading in a separate step to produce particular hydrocarbon blends. The upgraded FT hydrocarbons can be used for a range of products, from shampoo to jet fuel. Steynberg and Dry (2004) reviewed many aspects of the FT process as currently practiced, while the website www.fischer-tropsch.org provides many historical papers on this subject. In addition to the FT chemistry, syngas can also be used to produce other fuels or products. For example, syngas can also be converted to methanol by reaction over a catalyst. The basic chemistry is

CO + 2 H2 → CH3OH

(∆H298 = –25.34 kJ/mol)



CO2 + 3 H2 → CH3OH + H2O

(∆H298 = –49.47 kJ/mol)

Depending on the choice of catalyst, this synthesis reaction can be carried out at different pressures and temperatures. Early ZnO/Cr2O3 catalysts operate at high pressure and temperature (350 bar, 400 °C), but modern Cu/ZnO/Cr2O3 catalysts can operate at 50 to 100 bar and <250°C (Twigg, 1996). Methanol is a precursor to multiple chemical products, including ethylene and propylene, which are basic ingredients for much of the chemical industry. Because of drawbacks like toxicity, water solubility, and low energy density, methanol is not likely to be used as a transportation fuel, but it can be converted to gasoline using a catalytic technique (Kam et al., 1984). The details of how methanol can be produced from synthesis gas starting with coal are presented by Supp (1990). Substitute natural gas (SNG) can also be produced from syngas. The methanation reaction converts the syngas carbon monoxide and hydrogen to methane and steam:

CO + 3 H2 → CH4 + H2O

(∆H298 = –206 kJ/mol)

This reaction is promoted by a nickel catalyst, operating at temperatures from 300 to 400°C, and at elevated pressure (Probstein and Hicks, 1990). At present, there is only one plant in the United States making substitute natural gas from coal (Dittus and Johnson, 2001), but rising natural gas prices have renewed interest in SNG (Amick, 2007). Some concepts using different gasification approaches have been proposed using hydrogen gasification or catalysts to promote direct production of methane from the coal (Hobbs, 2007; Perleman, 2007). These alternate approaches can reduce some of the thermal energy released during syngas production, and may offer improved conversion efficiency. Perhaps the simplest chemical produced from syngas is hydrogen. As explained in Chapter 1, hydrogen production from coal has gained considerable interest because the hydrogen can be directly used for power production in combined-cycle power plants, avoiding CO2 emissions if geological sequestration is practiced (see Chapter 1). Aside from power production, hydrogen has numerous industrial uses,

216

Synthesis Gas Combustion: Fundamentals and Applications

such as ammonia production or refinery processing. Ammonia is produced by hydrogen reaction with nitrogen:

1/2 N2 + 3/2 H2 → NH3

(∆H700K = –52.6 kJ/mol)

This synthesis is practiced at temperatures around 400°C and pressures above 70 atm using iron-based catalysts; more details are found in Twigg (1996). The current gasification database lists thirty-five gasification plants producing ammonia as a product stream (Gasification World Database, 2007). When syngas is converted to a different fuel, such as a liquid fuel or SNG, the efficiency of converting the original solid fuel heating value to the product heating value is an important consideration. It is fortunate that the syngas conversion reactions listed above are all spontaneous at process conditions, but their exothermic character also means that a portion of the original feedstock fuel energy is released as heat, and not incorporated in the product fuel. And, because a particular reaction stoichiometry is needed to conduct the synthesis reactions, it is often beneficial to mix the product yield to include several chemicals and on-site power generation to make the best use of the input fuel properties. Mako and Samuel (1984) reported the overall efficiency of a coal-to-liquid plant that was designed for liquid fuel production alone. This is compared to a mixed-output plant that produced a blend of liquid fuel and SNG. Expressing efficiency as the ratio of the input to output fuel heating value, this comparison showed that the efficiency of the all-liquid fuel plant was just 44%, compared to 57% efficiency for the mixed-output plant (liquid fuel + SNG). This comparison shows the remarkable benefit from optimizing the product slate to make optimal use of the input fuel energy. More recently, Wang et al. (2008) compared various combinations of FT-liquid and power production, with a particular emphasis on how the gas turbine is configured to use partially converted products of the FT reactors. A range of effective efficiencies (coal energy to electrical + FT fuel energy) from 52.2 to 56% were reported, depending on how the plant was configured.

7.8 Conclusions This chapter provides a survey of methods for utilizing syngas for energy production, along with presenting various challenges or benefits that can be derived from such systems. The use of syngas in gas turbines is the most mature of these technologies, though there are still opportunities for improvement, most notably in designing low-NOX combustors that do not require diluents, and can operate on a wide range of syngas compositions and other fuels. Reciprocating engines are also a very mature technology, but their relatively small size makes them better suited to combustion of biofuels or gasified biomass than to coal syngas. Oxy-fuel combustion and chemical looping systems represent emerging technologies that hold promise for efficient power production from syngas with integrated carbon dioxide capture and sequestration, though significant research into these systems will be required before they are ready for commercial deployment. Of the various types of fuel cells available, solid oxide fuel cells show the most promise for efficient operation on syngas, particularly

Syngas Utilization

217

if a bottoming cycle is used to generate additional power from its high-temperature exhaust. In addition, recent increases in oil and natural gas prices have renewed an interest in producing liquid FT fuels from coal feedstocks, while the use of syngas in the chemical production industry presents many opportunities for combined ­chemical and power systems. All told, there are several promising approaches to efficiently using syngas for power production, many of which also enable integration of carbon capture and storage activities, providing a needed mechanism for meeting the energy needs of the future in an environmentally responsible manner.

References Abad, A., Mattisson, T., Lyngfelt, A., and Ryden, M. (2006). Chemical looping combustion in a 300W continuously operating reactor system using manganese-based oxygen carrier. Fuel 85:1174. Amick, P. (2007). Gasification industry roundtable, major developments and emerging trends. Paper presented at the 2007 Gasification Technology Conference, San Francisco. Avail­ able at http://www.gasification.org/Docs/2007_Papers/01AMIC.pdf. Anderson, R., and Pronske, K. (2006). CES gas generator plant: A new kind of aeroderivative. Modern Power Systems May 20–27. Andrus, H. E., Thibeault, P. R., and Jain, S. (2006). Alstom’s hybird combustion-­gasification chemical looping technology development—Phase II. Paper presented at the 23rd Inter­ national Pittsburgh Coal Conference. Anheden, M., and Svedberg, G. (1998). Exergy analysis of chemical looping combustion systems. Energy Convers. Mgmt. 39:1967. Antares Group. (2003). Assessment of power production at rural utilities using forest thinnings and commercially available biomass power technologies. Final report, Task Order TOA KDC-9-29462-19, prepared for USDA, DOE, and NREL. Available at http://www. antaresgroupinc.com/DOERUSreport.htm. Babcock and Wilcox. (2005). Steam—Its generation and use, ed. J. B. Kitto and S. C. Stultz. 41st ed. Barberton, OH: Babcock and Wilcox Company. Battista, R. A., and Dudley, J. C. (1995). Full-scale F technology combustor testing of simulated coal gas. Final report, DOE-MC/27221-1. Baukal, C. E., and Schwartz, R. E. (2001). The John Zink combustion handbook. Boca Raton, FL: CRC Press. Becker, B., and Schetter, B. (1993). Use of LHV in a gas turbine. Bioresources Technol. 46:55. Beebe, K. W., and Blanton, J. C. (1985). Design and development of a heavy-duty industrial gas turbine combustion system for low-Btu coal gas fuel. ASME Paper 85-GT-45. Blomen, L. J., and Mugerwa, M. N. (1993). Fuel cell systems. New York: Plenum Press. Boehman, A. L., and Le Corre, O. (2008). Combustion of syngas in internal combustion engines. Combust. Sci. Tech. 108:1193. Buhre, B. J. P., Elliott, L. K., Sheng, C. D., Gupta, R. P., and Wall, T. F. (2005). Oxy-fuel combustion technology for coal-fired power generation. Prog. Energy Combust. Sci. 31:283. Campanari, S., and Macchi, E. (1998). Thermodynamic analysis of advanced power cycles based upon solid oxide fuel cells, gas turbine and Rankine bottoming cycles. ASME Paper 98-GT-585. Cao, Y., and Pan, W.-P. (2006). Investigation of chemical looping combustion by solid fuels. 1. Process analysis. Energy Fuels 20:1836. Casleton, K. H., Breault, R. W., and Richards, G. A. (2008). System issues associated with syngas production and combustion. Combust. Sci. Tech. 180:1013.

218

Synthesis Gas Combustion: Fundamentals and Applications

Chaos, M., and Dryer, F. L. (2008). Syngas combustion kinetics and applications. Combust. Sci. Tech. 180:1053. Chellini, R. (2007). Jenbacher engines power bioenergy park. Diesel & Gas Turbine Worldwide, January/February, pp. 14–15. Chen, R., and Driscoll, J. F. (1990). Nitric oxide levels of jet diffusion flames: Effects of coaxial air and other mixing parameters. Proc. Combust. Instit. 23:281. Chiesa, P., Lozza, G., and Mazzocchi, L. (2005). Using hydrogen as gas turbine fuel. ASME J. Eng. Gas Turb. Power 127:73. Consonni, S., Lozza, G., Pelliccia, G., Rossini, S., and Saviano, F. (2004). Chemical looping combustion for combined cycles with CO2 capture. ASME Paper GT2004-53503. Cook, C. S., Corman, J. C., and Todd, D. M. (1995). System evaluation and LBTU fuel combustion studies for IGCC power generation. ASME J. Eng. Gas Turb. Power 117:673. Couch, G. (2005). Use of coal in areas of water shortage. London: IEA Clean Coal Centre. de Biasi, V. (2005). SCR poses problems but can be used with IGCC to achieve ultra-low NOX. Gas Turbine World, July/August, pp. 28–30. Dittus, M., and Johnson, D. (2001). The hidden value of lignite coal. Paper presented at the Gasification Technology Conference, San Francisco. Available at http://www.­ gasification.org/Docs/2001_Papers/GTC01009.pdf. Feese, J. J., and Turns, S. R. (1998). Nitric oxide emissions from laminar diffusion flames: Effect of air-side versus fuel-side diluent addition. Combust. Flame 113:66. Fuel Cell Handbook. 7th ed. (2004). Department of Energy Contract DE-AM26-99FT40575, National Technical Information Services, Springfield, VA. Available at www.netl.doe.gov. Gabriel, R., Navedo, J., and Chen, R. (2000). Effects of fuel Lewis number on nitric oxide emissions of diluted H2 turbulent jet diffusion flames. Combust. Flame 121:525. Garcia-Labiano, F., Adanez, J., deDiego, L. F., Gayan, P., and Abad, A. (2006). Effect of pressure on the behavior of nickel-based oxygen carriers for chemical looping combustion. Energy Fuels 20:26. Gasification World Database. (2007). Available at www.netl.doe.gov. Geosits, R. F., and Schmoe, L. F. (2005). IGCC—The challenges of integration. Paper presented in panel session, ASME TurboExpo 2005, Reno, NV, June 8. Giles, D. E., Som, S., and Aggarwal, S. K. (2006). NOX emission characteristics of counterflow syngas diffusion flames with airstream dilution. Fuel 85:1729. Goodwin, D. G. (2003). An open-source, extensible software suite for CVD process simulation. In Chemical Vapor Deposition XVI and EUROCVD 14, Pennington, NJ, ed. M. Allendorf, F. Maury, and F. Teyssandier, Electrochemical Society Proceedings Series, PV 2003–08, p. 155. See also http://www.cantera.org. Greenhalgh, M. L. (1992a). High-pressure coal fuel processor development, task 1. Proof of principle testing. Topical report, DOE/MC/25141-3339, November. Greenhalgh, M. L. (1992b). High-pressure coal fuel processor development. Final report, DOE/MC/25141-3340, December. Higman, C., and van der Burgt, M. (2003). Gasification. Boston: Gulf Professional Publishing (Elsevier). Hobbs, R. (2007). Development of a hydrogasification process for co-production of substitute natural gas (SNG) and electric power from western coals—Phase 1. Topical report, DOE Cooperative Agreement DE-FC26-06NT42759. Jerndal, E., Mattisson, T., and Lyngfelt, A. (2006). Thermal analysis of chemical looping combustion. Trans. IChemE A Chem. Eng. Res. Des. 84:795. Jin, H., and Ishida, M. (2000). Investigation of a novel gas turbine cycle with coal gas fueled chemical looping combustion. In Proceedings of the ASME, Advanced Energy Systems Division, AES-Vol. 40, 547. Johansson, E., Mattisson, T., Lyngfelt, A., and Thunman, H. (2006a). Combustion of syngas and natural gas in a 300W chemical looping combustor. Chem. Eng. Res. Des. 84:819.

Syngas Utilization

219

Johansson, E., Mattisson, T., Lyngfelt, A., and Thunman, H. (2006b). A 300W laboratory reactor system for chemical-looping combustion with particle circulation. Fuel 85:1428. Jones, R. M. (2005). Gas turbine requirements for integrated gasification combined cycle (IGCC) applications. Paper presented at POWER-GEN International, Las Vegas, NV. Kalitan, D. M., Mertens, J. D., Crofton, M. W., and Petersen, E. L. (2007). Ignition and oxidation of lean CO/H2 fuel blends in air. J. Propul. Power 23:1291. Kam, A. Y., Schreiner, M., and Yurchak, S. (1984). Mobil methanol to gasoline (MTG) process. In Handbook of synfuel technology, ed. R. A. Myer, 2–75, 111. New York: McGraw-Hill Publishing. Kelsall, G. J., Smith, M. A., and Cannon, M. F. (1994). Low emission combustor development for an industrial gas turbine to utilize LCV fuel gas. ASME J. Eng. Gas Turb. Power 116:559. Klassen, M. (2005). Power generation. In NGC+ Interchangeability Work Group. Appendix C.3. White paper on natural gas interchangeability and noncombustion end use. Konnov, A. A. (2003). On the relative importance of different routes forming NO in hydrogen flames. Combust. Flame 134:421. Konnov, A. A., Dyakov, I. V., and De Ruyck, J. (2002). Nitric oxide formation in premixed flames of H2 + CO + CO2 and air. Proc. Combust. Instit. 29:2171. Kvamsdal, H. M., Jordal, K., and Bolland, O. (2007). A quantitative comparison of gas turbine cycles with CO2 capture. Energy 32:10. Lefebvre, A. H. (1999). Gas turbine combustion. 2nd ed. Ann Arbor, MI: Taylor and Francis. Lewis, B., and von Elbe, G. (1987). Combustion, flames, and explosions of gases. 3rd ed. Orlando, FL: Academic Press. Lieuwen, T., McDonell, V., Santavicca, D., and Sattelmayer, T. (2008). Burner development and operability issues associated with steady flowing syngas fired combustors. Combust. Sci. Tech. 180:1169. Lieuwen, T. C., and Yang, V., Eds. (2005). Combustion instabilities in gas turbine engines: Operational experience, fundamental mechanisms, and modeling. Progress in Astronautics and Aeronautics Series, Vol. 210. Reston, VA: AIAA Press. Liss, W., Thrasher, W., Steinmetz, G., Chowdiah, P., and Attari, A. (1992). Variability of natural gas composition in select major metropolitan areas of the United States. Final report to GRI, Report GRI-92/0123, March. Lozza, G., Chiesa, P., Romano, M., and Savoldelli, P. (2006). Three reactor chemical looping combustion for high-efficiency electricity generation with CO2 capture from natural gas. ASME Paper GT2006-90345. Lyon, R. K., and Cole, J. A. (2000). Unmixed combustion: An alternative to fire. Combust. Flame 121:249. Mako, P. F., and Samuel, W. A. (1984). The Sasol approach to liquid fuels from coal via the Fischer-Tropsch reaction. In Handbook of synfuel technology, ed. R. A. Myer, 2–5, 43. New York: McGraw-Hill Publishing. Marek, C. J., Smith, T. D., and Kundu, K. (2005). Low-emission hydrogen combustion for gas turbines using lean direct injection. AIAA-2005-3776. Mittal, G., Sung, C.-J., and Yetter, R. A. (2006). Autoignition of H2/CO at elevated pressures on a rapid compression machine. Int. J. Chem. Kinetics 38:516–29. Moreira, P. (2006). Gasification as viewed by a hydrogen producer. Paper presented at the Gasification Technologies Conference, Washington, DC, October 1–4. Naqvi, R., Bolland, O., Brandvoll, O., and Helle, K. (2004). Chemical looping combustion—Analysis of natural gas fired power cycles with inherent CO2 capture. ASME Paper GT2004-53359. Nexant. (2006). Environmental footprints and costs of coal-based integrated gasification combined cycle and pulverized coal technologies. EPA Report EPA/430/R-06/006. Available at www.epa.gov/air/caaac/coaltech/2007_01_epaigcc.pdf.

220

Synthesis Gas Combustion: Fundamentals and Applications

Noble, D. R., Zhang, Q., Shareef, A., Tootle, J., Myers, A., and Lieuwen, T. (2006). Syngas mixture composition effects upon flashback and blowout. ASME Paper GT2006-90470. Perleman, A. (2007). U.S. Senate, Committee on Energy and Natural Resources, Hearing on Advances in Clean Coal Technology, testimony of Andrew Perelman, August 1. Avail­ able at http://energy.senate.gov/public/index.cfm?IsPrint=true&FuseAction=Hearings. Hearing&Hearing_ID=75b64865-1f21-47c2-b38a-21f409e2696d. Probstein, R. F., and Hicks, R. E. (1990). Synthetic fuels. Cambridge, MA: pH Press. Rao, A. D, Yi, Y., Brouwer, J., and Samuelsen, G. S. (2004). Analysis and optimization of a solid oxide fuel cell and intercooled gas turbine (SOFC-ICGT) hybrid cycle. J. Power Sour. 132:77. Richards, G. A., Casleton, K. H., and Chorpening, B. T. (2005). CO2 and H2O diluted oxy-fuel combustion for zero-emission power. J. Power Energy 218:121. Rizeq, G., Subia, R., Frydman, A., West, J., and Zamansky, V. (2003). Development of unmixed fuel processor for production of H2, electricity, and sequestration-ready CO2. Paper presented at the 20th Annual International Pittsburgh Coal Conference, September 15–19. Rosenberg, W. G., Alpern, D. C., and Walker, M. R. (2005). Deploying IGCC technology in this decade with 3 party covenant financing. Vol. I, May 2005 revision. ENRP Discussion Paper 2004-07. Cambridge, MA. Available at http://belfercenter.ksg.harvard.edu/files/ igcc_vol1.pdf. Ryden, M., Lyngfelt, A., and Mattisson, T. (2006). Synthesis gas generation by chemical looping reforming in a continuously operating laboratory reactor. Fuel 85:1631. Siriwardane, R., Chaudhari, K., Zinn, A., Simonyi, T., and Robinson, C. (2006). Oxygen-carrier development for chemical looping combustion of coal-derived synthesis gas. Paper presented at the 23rd Annual International Pittsburgh Coal Conference, September 25–28. Smith, A. R., and Klosek, J. (2001). A review of air separation technologies and their integration with energy conversion processes. Fuel Process. Tech. 70:115. Smith, A. R., Klosek, J., and Woodward, D. W. (1997). Next generation integration concepts for air separation units and gas turbines. J. Eng. Gas Turb. Power 119:298. Smith, G. P., Golden, D. M., Frenklach, M., Moriarty, N. W., Eiteneer, B., Goldenberg, M., Bowman, C. T., Hanson, R. K., Song, S., Gardiner, W. C., Jr., Lissianski, V. V., and Qin, Z. (n.d.). http://www.me.berkeley.edu/gri_mech/. Steele, R. C., Malte, P. C., Nicol, D. G., and Kramlich, J. C. (1995). NOX and N2O in leanpremixed jet-stirred flames. Combust. Flame 100:440. Steynberg, A., and Dry, M., Eds. (2004). Studies in surface science and catalysis: FischerTropsch technology. Vol. 152. Amsterdam: Elsevier. Straub, D. L., Bedick, C., Sidwell, T., Strakey, P., and Casleton, K. (2006). Flashback studies in a lean premixed combustor operating on hydrogen–natural gas fuel blends. Paper 06S-35 presented at the Western States Section of the Combustion Institute, 2006 spring meeting, Boise, ID. Sung, C.-J., and Law, C. K. (2008). Fundamental combustion properties of H2/CO mixtures: Ignition and flame propagation at elevated pressures. Combust. Sci. Tech. 180:1097. Supp, E. (1990). How to produce methanol from coal. Berlin: Springer. Surdoval, W. (2007). U.S. Department of Energy Fossil Energy Fuel Cell Program. Paper presented at the 8th Annual Solid State Energy Conversion Alliance (SECA) Workshop, August 7–9. Availabe at http://www.netl.doe.gov/publications/proceedings/07/ SECA_Workshop/. TAIX LLC. (2003). Scale-up of planar SOFC stack technology for MW-level combined cycle systems. Report prepared for U.S. Department of Energy, National Energy Technology Laboratory, October. Available at http://www.netl.doe.gov/technologies/coalpower/ fuelcells/publications/SOFC_GTHybrid_Scaleup_FinalReport.pdf.

Syngas Utilization

221

Tian, H., Simonyi, T., and Siriwardane, R. (2008). Investigation of chemical looping combustion of coal utilizing various oxygen carriers. Paper presented at the 7th Annual Con­ ference on Carbon Capture and Sequestration, Pittsburgh, PA, May 5–8. Tobias, M., Lyngfelt, A., and Leion, H. (2009). Chemical looping with oxygen uncoupling for combustion of solid fuels. Int. J. Greenhouse Gas Control 3:11. Todd, D. M. (2000). Gas turbine improvements enhance IGCC viability. Paper presented at Gasification Technologies Conference, San Francisco, October 8–11. Trembly, J., Gemmen, R. S., and Bayless, D. J. (2007a); The effect of coal syngas containing AsH3 on the performance of SOFCs: Investigations into the effect of operational temperature, current density and AsH3 concentration. J. Power Sources 171:818. Trembly, J., Gemmen, R. S., and Bayless, D. J. (2007b). The effect of coal syngas containing HCl on the performance of solid oxide fuel cells: Investigations into the effect of operational temperature and HCl concentration. J. Power Sources 169:347. Trembly, J., Gemmen, R. S., and Bayless, D. J. (2007c). The effect of IGFC warm gas cleanup system conditions on the gas-solid partitioning and form of trace species in coal syngas and their interaction with SOFC anodes. J. Power Sources 163:986. Tullo, A. H., and Tremblay, J.-F. (2008). Coal: The new black. Chemical and Engineering News, March 17, pp. 15–22. Twigg, M. V. (1996). Catalyst handbook. 2nd ed. London: Manson Publishing. U.S. EPA. (n.d.). RACT/BACT/LAER clearinghouse (RBLC). Available at http://cfpub1.epa. gov/rblc/htm/bl02.cfm. Veyo, S. E., Shockling, L. A., Dederer, J. T., Gillett, J. E., and Lundberg, W. L. (2002). Tubular solid oxide fuel cell/gas turbine hybrid cycle power systems: Status. J. Eng. Gas Turb. Power 124:845. Vielstich, W., Lamm, A., and Gasteiger, W., Eds. (2003). Handbook of fuel cells: Fundamentals and survey of systems. Vol. 1. Hoboken, NJ: John Wiley & Sons. Vogt, R. L. (1980). Low Btu coal gas combustion in high temperature turbines. ASME Paper 80-GT-170. Wang, X., Xiao, Y., Xu, S., and Guo, Z. (2008). Predicting the performance of system for the co-production of Fischer-Tropsch synthetic liquid and power from coal. J. Eng. Gas Turb. Power 130:011401-1. Warnatz, J., Maas, U., and Dibble, R. W. (1996). Combustion: Physical and chemical fundamentals, modeling and simulation, experiments, pollutant formation. Berlin: Springer-Verlag. Waste gas-burning engines reach milestone. (2006). Power, November/December, pp. 8–9. Weiland, N., Chen, R.-H., and Strakey, P. (2007). Effects of coaxial air nitrogen-diluted hydrogen jet diffusion flame length and NOX emission. Paper presented at the Fall Technical Meeting of the Eastern States, Combustion Institute, Charlottesville, VA, October 21–24.

Combustion 8 Catalytic of Syngas John Mantzaras Contents 8.1 8.2 8.3 8.4 8.5 8.6

Introduction to Catalytic Combustion........................................................... 223 Catalytic Combustion Methodologies in Power Generation......................... 226 Emissions in Catalytic Combustion............................................................... 229 Reactor Thermal Management...................................................................... 230 Catalysts for High-Temperature Applications............................................... 234 Catalytic Combustion of Syngas................................................................... 235 8.6.1 Numerical Results.............................................................................. 236 8.6.2 Catalytic Combustion of Hydrogen/Air Mixtures............................. 239 8.6.2.1 Impact of Gas‑Phase Chemistry......................................... 239 8.6.2.2 Effect of Pressure................................................................ 242 8.6.2.2 Light-Off Temperatures......................................................244 8.6.3 Catalytic Combustion of CO/Air Mixtures.......................................246 8.6.4 Catalytic Combustion of H2/CO Mixtures........................................ 247 8.6.4.1 Hetero-/Homogeneous Chemistry Coupling...................... 247 8.6.4.2 Surface Temperatures......................................................... 250 8.6.4.3 Light-Off Temperatures...................................................... 252 8.7 Conclusions.................................................................................................... 255 Acknowledgments................................................................................................... 256 References............................................................................................................... 256

8.1 Introduction to Catalytic Combustion Heavy-duty natural gas–fired turbines based on the lean‑premixed combustion technology currently attain NOX emissions of 25 ppm (15% O2) or less. The forthcoming emission legislation is likely to become more stringent, at least for power plants located in or near urban areas where the lowest achievable emissions rate (LAER) standards may be enforced. During the past two decades, heterogeneous (catalytic) combustion has been intensively investigated for gas turbines of power generation systems (Figure  8.1) as a means to reduce NOX emissions and provide improved combustion stability. Heterogeneous fuel conversion is accomplished in ceramic or metallic honeycomb reactors (Figure 8.1b) that are coated with an active catalyst and have suitably large surface‑to‑volume ratios. The underlying physicochemical processes within 223

224

Synthesis Gas Combustion: Fundamentals and Applications

(a) Gas turbine

(b) Honeycomb reactor

50

–1

UIN TIN YIN

Uniform inlet properties

(c) Catalytic channel

Diffusion Surface reactions Catalytic ignition

m

Gaseous combustion

Development length

Convection Radiation

m

U T

Heat conduction in solid wall

OH H 2O

U

0.5–2.0 mm

Radiation

50

Homogeneous ignition

Figure  8.1  Catalytic combustion in power generation: (a) gas turbine using catalytic combustion technology, (b) honeycomb catalytic reactor, and (c) physicochemical processes within each catalytic channel.

each catalytic channel of the reactor, which ultimately determine the overall combustion performance, are depicted in Figure 8.1c. Preheated fuel/air premixtures are admitted in each catalytic channel at inflow velocities that guarantee, in most cases, laminar flows (Appel et al., 2005b). Fuel and oxidizer diffuse transversely to the catalytically active surface of the channel. Catalytic ignition (light‑off) is attained at a certain distance from the entry, which depends not only on the operating conditions (inflow velocity and temperature, pressure, fuel‑to‑air stoichiometry, fuel type) and the catalytic reactivity, but also on key in‑channel heat transfer mechanisms (heat conduction in the solid walls and surface radiation heat transfer). Therein, the surface temperature has reached a sufficiently high level such that the heterogeneous conversion shifts from the kinetically controlled to the transport‑controlled regime (Pfefferle and Pfefferle, 1986). Following catalytic ignition, the fuel and oxidizer react vigorously at the catalyst surface forming a degenerate diffusion reaction sheet (Williams, 1985). This terminology reflects the fact that both reactants diffuse from the same side of the sheet and that the position of the reaction sheet is fixed in space. Heat and reaction products diffuse back to the main flow and a homogeneous (gas‑phase) ignition is then established inside the channel (Figure 8.1c), if conditions are appropriate. Given the lower activation energy of the catalytic reaction pathway compared to that of the gaseous pathway (for example, in methane the catalyst reduces the apparent activation energy by a factor of about 2), homogenous ignition can be initiated downstream of the light‑off position.

225

Catalytic Combustion of Syngas 500 ppm OH Homogeneous ignition

250

0.2 mm

Solid wall

Wall Temperature (K)

UIN TIN

OH 0

10

R = 1 mm 20 30 x (mm) (a)

40

50

1.6

1400 1200

1.2

Catalytic conversion

1000 800 600

0 Light-off

10

20

x (mm)

0.8

Gaseous conversion 30

40

0.4 50

0.0

CH4 Conversion (gr/m2s)

0

(b)

Figure 8.2  Predicted catalytic combustion of a fuel‑lean (φ = 0.3) CH4/air mixture over Pt. Inlet velocity and temperature UIN = 5 m/s and TIN = 700 K, respectively, pressure of 1 bar, channel radius of 1 mm, solid wall thickness of 0.2 mm, and solid thermal conductivity of 1 W/mK. (a) OH radical distribution in ppm‑mass (half the channel domain is shown), and (b) axial profiles of wall temperature and methane conversions: the gaseous conversion has been integrated over the channel radius.

Fundamentally, there are four main coupling routes between the two reaction pathways affecting gaseous combustion. The catalytically induced near‑wall fuel depletion (see Figure 8.1c, the limiting reactant profiles are similar to the shown velocity profiles) inhibits homogeneous ignition, whereas the heat transfer from the hot catalytic walls to the reacting gas promotes ignition. In addition, gas‑phase ignition is mildly inhibited by the recombination of radicals on the catalyst (Appel et al., 2002; Reinke et al., 2005). Moreover, it can either be inhibited or promoted by heterogeneously produced major species (notably H2O), depending on the fuel type. These issues will be discussed for particular syngas fuel compositions in Section 8.6. The catalytic burner design requirements can be summarized in the computational results shown in Figure  8.2, obtained from a two-dimensional model (see Section 8.6.1) that accounts for all the relevant processes given in Figure 8.1c. A tubular catalytic channel coated with platinum is considered. The channel has a radius of 1 mm, leading to a honeycomb structure in Figure 8.1b with a low cell density of ca. 100 cpsi (cells per square inch). The axial profiles in Figure 8.2b exemplify the basic design needs: minimizing the light‑off length, maximizing the catalytic fuel conversion, maintaining the wall temperature at a tolerable level (<1300 K), and avoiding the onset of gas‑phase combustion within the catalytic module. Furthermore, these conditions should be achieved with a minimum pressure drop. Homogeneous

226

Synthesis Gas Combustion: Fundamentals and Applications

ignition within the catalytic module (as seen from the OH radical map in Figure 8.2a) is deemed detrimental to the catalyst integrity and may lead to reactor meltdown. Although this statement generally holds for natural gas fuels, it may not be the case for diffusionally imbalanced fuels with Lewis numbers less than unity (such as hydrogen or hydrogen‑rich syngas), as will be shown in Section 8.6.2.1. The catalysts in power generation systems are usually supported noble metals. The support is a porous metal oxide layer (e.g., Al2O3 or ZrO2) that furnishes high surface area, a uniform dispersion of the precious metal, and features a good adhesion to the ceramic or metallic reactor substrate structure.

8.2 Catalytic Combustion Methodologies in Power Generation The catalytically stabilized thermal (CST) combustion concept was initially studied (Pfefferle, 1974). Therein, part of the fuel is converted heterogeneously in one or more catalytic reactor modules, with the remaining combusted in a follow‑up homogeneous burnout zone (Figure 8.3) (Beebe et al., 2000; Carroni et al., 2002). CST is an ultralow NOX combustion technology (Krill and Kesselring, 1978; Schlegel et al., 1996) with demonstrated NOX emissions of 3 ppm or less in a variety of test configurations (Beebe et al., 2000; Karim et al., 2002; Schmitz et al., 2005). It is more cost‑efficient than conventional tail‑end cleanup NOX methods (ONSE, 1999), and it provides improved flame stability and lower combustion‑induced pressure pulsations than those typically encountered in lean‑premixed burners (Richards et al., 2001; Yee et al., 2001). Furthermore, the catalytic technology can improve the clean utilization of renewable and “dirty” fuels (Burch and Southward, 2000; Witton et al., 2003). Biogas, for example, contains significant amounts of NH3 (fuel‑bound nitrogen), which typically lead to enhanced NOX emissions during gas‑phase combustion. Catalytic treatment can be applied to oxidize NH3 and subsequently reduce NOX back to N2. Coal‑derived syngas is another candidate fuel suitable for catalytic combustion.

Fuel injector

CST Combustion Mixer and preburner

Air

TCD (compressor discharge) < 450°C

Fuel-lean catalyst

TIN, CAT > 500°C < 0.50

Gas-phase combustion zone

850 < TOUT, CAT < 1000°C

TFinal ~ 1300°C

Figure 8.3  Catalytically stabilized thermal combustion (CST) approach.

Catalytic Combustion of Syngas

227

The operational principle of a conventional CST‑based natural gas turbine burner is shown in Figure 8.3, along with typical design temperatures. Fuel is premixed with air at an overall fuel lean stoichiometry upstream of a catalytic module. The catalytic design (including the number of catalytic modules, cell density, length, crosssectional area, and catalytically active material) depends on the specific operating conditions. In fuel‑lean natural gas combustion, Pd‑based catalysts offer a realistic solution given their satisfactory activity and low volatilization rate. A preburner (Figure  8.3), however, may still be needed to raise the gas inlet temperature to a level required for catalytic ignition. Due to material stability limitations and the steady increase in the gas turbine combustion temperature (up to 1400°C in modern machines), original design concepts with complete combustion in the catalytic module (Kajita et al., 1990) were soon abandoned in favor of the CST alternative of Figure 8.3. CST is a hybrid combustion approach, wherein the catalyst only preheats the gas to a temperature at which homogeneous combustion can be initiated and aerodynamically stabilized downstream of the catalytic section (Furuya et al., 1987; Dalla Betta et al., 1993). The heterogeneous fuel conversion—and hence the temperature at the exit of the catalytic reactor—is limited by the catalytic reaction rate and the fraction of the fuel directed through the catalyst. Part of the fuel and air may bypass the catalytic section such that the output of the catalytic reactor forms a pilot flame that in turn stabilizes the homogeneous combustion of the bypassed reactants. For diffusionally neutral fuels with Lewis numbers close to unity (e.g., methane), a fractional catalytic fuel conversion does not by itself warrant surface temperatures less than the adiabatic equilibrium temperature of the incoming reacting mixture. Nonadiabatic operation of the catalytic reactor or finite‑rate heterogeneous chemistry must be employed in order to moderate the surface temperatures to levels tolerable by the catalyst and the reactor structure (up to ~1300 K). Details regarding the reactor thermal management and the accompanying design strategies will be discussed in Section 8.4. Finally, the fuel‑lean catalytic combustion is of interest to applications other than gas turbines. These include radiant burners, industrial boilers, portable heaters, and microreactors, to mention a few. Catalytically stabilized combustion can be a viable combustion technology not only for natural gas but also for syngas or other hydrogen‑rich fuels such as biogas. In particular, catalytic combustion approaches appear well suited for low calorific value syngas‑based fuels with large inert dilution (biogases, etc.) due to the associated lower combustion temperatures. Catalytic combustion of syngas‑based fuels, with emphasis on gas turbine operation, has been investigated in Groppi et al. (1996) for CO/H2 mixtures and Johansson et al. (2002) for gasified biomass. General catalytic combustion studies for biogas fuels, focusing mainly on catalyst development and characterization, have been reported in Johansson et al. (1999), Berg et al. (2000), Lietti et al. (2000), Thevenin et al. (2001), Kusar et al. (2003), and Chao et al. (2004). A full CST concept (Figure 8.3) with a postcatalyst homogeneous combustion zone, however, has not been elaborated in the case of syngas‑based fuels. A recent alternative to CST for turbines fueled with natural gas, referred to as catalytic-rich combustion (Figure  8.4), entails catalytic partial oxidation (CPO) of the fuel (Karim et al., 2002; Griffin et al., 2004) to synthesis gas. In the

228

Synthesis Gas Combustion: Fundamentals and Applications Fuel injector

Fuel-Rich Combustion

Fuel-lean mixture Mixer

Air

Fuel-rich catalyst

Fuel-lean mixture

TCD < 450°C

TIN, CAT ~ TCD 2< <4

Main flame Catalytic pilot flame Main flame

TFinal ~1300°C

Figure 8.4  Fuel‑rich catalytic combustion methodology.

approach shown in Figure  8.4, part of the air and fuel flows are mixed with an overall fuel‑rich stoichiometry and then directed into the CPO reactor. The products (mainly hydrogen‑rich synthesis gas and unconverted reactants) create a pilot flame that stabilizes the main fuel‑lean homogeneous combustion of the bypassed reactants. This approach can be applied in existing lean‑premixed combustors by injecting hydrogen‑rich CPO products at specific locations so as to increase the overall burner stability. In other designs, all of the fuel is driven in the CPO reactor and the bypass only consists of the airflow (Smith et al., 2006). The fuel‑rich methodology has a number of advantages compared to the conventional CST. The most significant ones are a lower catalyst light‑off temperature than that from fuel‑lean mixtures (Veser et al., 1999), thus eliminating the preburner shown in Figure 8.3, which is a source of NOX emissions; an extended extinction limit compared to the lean catalytic combustion (Schneider et al., 2006, 2008; Smith et al., 2006); the control of catalytic combustion by the availability of oxygen, which precludes the complete consumption of fuel inside the CPO module, even in the event of accidental gas‑phase ignition; and the enhanced stability of the follow‑up flame due to the hydrogen content (Griffin et al., 2004; Smith et al., 2005). In particular, it has been demonstrated that the produced syngas can reduce the blowout limit temperature by ca. 100 K (Griffin et al., 2004). The fuel‑rich approach provides single‑digit NOX emissions, and moreover, the catalyst surface temperature is a moderate function of the specific fuel‑rich stoichiometry (Smith et al., 2006). This particular behavior is related to the extended CPO extinction limits and will be clarified in Section 8.4. The catalytic‑rich combustion methodology is also applicable to syngas fuels. In this case the catalyst does not have a prime CPO function (at least for syngas fuels with low methane content), but acts as a preheater and stabilizer for the ­follow‑up homogeneous combustion zone. This approach is suitable for a wide range of ­syngas‑based fuels that include low calorific value fuels, whereby flame stability is an issue, and also for hydrogen‑rich coal‑derived syngas, for which lean‑premixed

229

Catalytic Combustion of Syngas

combustion entails the risk of flame flashback. These advantages, along with the control of the catalytic conversion by the air and not by the fuel supply, have led to the development of integrated hetero‑/homogeneous combustors for coal‑derived syngas and high‑hydrogen-containing fuels (Etemad et al., 2004). The surface temperature can also be better controlled compared to lean combustion, as will be clarified in Section 8.4. Variants of the fuel‑rich combustion methodology for syngas fuels include the rich‑quick‑lean (RQL) combustion (Tham and Chen, 2005), where the catalytic module preferentially oxidizes the hydrogen of the syngas mixture such that the second‑stage homogeneous combustion can be accomplished without the danger of flame flashback.

8.3 Emissions in Catalytic Combustion Catalytically stabilized combustion of lean fuel/air mixtures offers the potential of significant reduction in NOX emissions (Beebe et al., 2000), while retaining acceptable CO emissions of 20 ppm or less. The lower NOX emissions, which are the driving force for the advancement of catalytic combustion, are due to the following reasons. Catalysts reduce the concentration of hydrocarbon radicals, which are needed for the formation of prompt (or Fenimore) NOX at temperatures below the final flame temperature. Increasing the fraction of catalytic fuel conversion decreases the hydrocarbon radical pool and thus reduces the formation of prompt NOX, which is the main NOX chemical production route at temperatures up to 1400°C (Schlegel et al., 1994, 1996). This is clearly illustrated in Figure 8.5 for CST of a methane/air mixture with a typical, for gas turbines, adiabatic flame temperature of 1300°C and a postflame residence time of 20 ms.

Relative Decrease in NOX

0 Tad = 1300°C

–20 –40 –60 –80

–100

0

20

40

60

80

100

% Catalytic Fuel Conversion

Figure 8.5  Measured reduction of NOX emissions in catalytically stabilized combustion versus fractional catalytic conversion (0% corresponds to pure gas‑phase combustion). Lean methane/air mixture with adiabatic flame temperature of 1300°C. (Adapted from Schlegel et al., 1996. With permission.)

230

Synthesis Gas Combustion: Fundamentals and Applications

Despite the fact that CST can accommodate the combustion of ultralean mixtures, the operational window of fuel/air stoichiometry is rather narrow (dictated by both catalyst overheating and extinction). This results in less thermal NOX production due to the better reactant mixing quality required by the catalysts (deviation in the fuel‑to‑air ratio that is equivalent to about ±30 K variation in the adiabatic flame temperature). Under perfect premixedness in current gas turbines, flame temperatures are moderate (<1400°C) such that the Zeldovich NOX mechanism is not significant. Imperfect mixing, however, results in local flame temperature variations of up to 150 K, which generate particularly high levels of thermal NOX. Thermal NOX formation becomes more important as the firing temperatures of modern large turbines are elevated for efficiency improvement (up to 1500°C, corresponding to adiabatic flame temperatures of about 1600°C). Numerical simulations indicate that the H2O and CO2 produced by the upstream catalytic section have a strong mitigating impact on thermal NOX production in the homogeneous combustion zone. Experiments (Krill and Kesselring, 1978) have confirmed this trend for pressures and temperatures up to 10 bar and 1700°C, respectively. In addition, experiments and predictions indicate that NOX emissions from catalytically stabilized flames are significantly less sensitive to temporal fuel/air unmixedness (within the ranges tolerated by CST) than pure homogeneous combustion (Boehman and Dibble, 2000). The fuel‑rich combustion approach also offers NOX emissions of 3 ppm or less with CO and unburned hydrocarbon emissions of less than 10 and 2 ppm, respectively, at pressures up to 16 bar (Etemad et al., 2004; Smith et al., 2006). Although fundamental studies of NOX formation in fuel‑rich catalytic combustion have not yet been reported, the chemical origin of emission reductions is much the same as in fuel‑lean combustion. Finally, biogas fuels may contain NH3, which is also a source of NOX emissions, irrespective of the specific catalytic combustion methodology. The development of appropriate catalysts for the oxidation of NH3 and further reduction of NOX has alleviated this problem (Burch and Southward, 2000; Kusar et al., 2005).

8.4 Reactor Thermal Management The attained surface temperature is an important design parameter since it directly impacts the catalyst stability and reactor integrity (and in addition, for the fuel‑rich concept of Figure 8.4, the syngas selectivity). Even in the absence of external heat losses, the coupling of transport and chemistry can give rise to surface temperatures noticeably different from the corresponding adiabatic equilibrium temperatures. There are three mechanisms responsible for such deviations: diffusional imbalance of the limiting reactant (Lewis number, Le, ≠ 1), finite‑rate surface chemistry, and nonequilibration of the combustion products due to short reactor residence times. The former two are controlling in fuel‑lean combustion, whereas all three reasons can play a role in fuel‑rich combustion. The constraint on the location of the catalytic reaction zone (channel wall) when coupled to nonequal heat and mass diffusivities gives rise to super‑ or underadiabatic

231

Catalytic Combustion of Syngas

surface temperatures. It can be shown (Mantzaras, 2006) that the surface temperature, TW, in channel flow catalytic combustion is TW = TG + LeFβ−1 (YF,G − YF,W )q cp



(8.1)

where TG is the bulk gas temperature, YF,G and YF,W the fuel mass fractions in the bulk of the gas and at the wall, respectively, q the heat release per unit mass of fuel, cp the mixture heat capacity, LeF the Lewis number of the deficient reactant (fuel in CST), and β = 0 or 1/3 for fully developed and developing channel flows, respectively. Fully developed flows usually are not realizable in CST due to the short channel aspect ratio, high inlet velocity, and the combustion‑induced flow acceleration. For infinitely fast (transport‑limited) catalytic reactions, YF,W = 0 and the maximum attainable surface temperature becomes TW = TG + LeFβ−1∆T



(8.2)

where ΔT = Tad – TG = YF,G q/cp is the adiabatic combustion temperature rise. Fuels with Le < 1 (Le > 1) lead to superadiabatic (underadiabatic) surface temperatures under conditions of infinitely fast catalytic chemistry. For a given Lewis number of the fuel, finite‑rate surface kinetics (YF,W ≠ 0) always reduce the surface temperature (see Equation 8.1) due to incomplete combustion. An example is presented in Figure 8.6 (Appel et al., 2002) for fuel‑lean (φ = 0.28) H2/air catalytic combustion in 2100

T

W

Temperature (K)

1800

,2

TW,1

1500 TW,3

Tad

1200 900

Tgas,2

600 300

0

50

100

150

200

250

300

x (mm)

Figure  8.6  Catalytic combustion of fuel‑lean (φ = 0.28) H2/air in a 300 mm long Pt‑coated channel. Computed surface temperatures for finite‑rate chemistry (TW,1) and for transport‑­limited operation (TW,2). The mean gas temperature for the mass‑transport‑limited case (Tgas,2) is also shown. The symbols refer to experiments under nonadiabatic operation (TW,3), and the horizontal line marked Tad denotes the adiabatic equilibrium temperature. (Adapted from Appel et al., 2002. With permission.)

232

Synthesis Gas Combustion: Fundamentals and Applications

a 300 mm long rectangular channel coated with Pt. Two predicted surface temperatures are provided, one (TW,1) obtained with finite‑rate catalytic chemistry (based on the reaction scheme of Deutschmann et al., 2000) and the other (TW,2) with an artificially imposed infinitely fast chemistry. Superadiabatic temperatures are attained in both cases (Tad = 1148 K) that are more pronounced in the far upstream regions due to the higher amounts of available hydrogen. In the transport‑limited case, the temperature at x = 0 approaches the theoretical value of Equation 8.2, T = TIN + Le–2/3ΔT ≈ 2100 K (ΔT = Tad – TIN and Le ≈ 0.32 for hydrogen under fuel‑lean conditions). The attained temperatures in both computed cases are unacceptably high and may cause reactor meltdown. The measured surface temperatures of Figure 8.6 (TW,3) have been obtained with a special water‑cooling arrangement at the channel entry (Appel et al., 2002). Superadiabatic surface temperatures of the type presented in Figure 8.6 can be of great concern for syngas fuels with high hydrogen contents and also for natural gas combustion due to the slight diffusional imbalance of methane (Le ≈ 0.95). A practical solution to protect the reactor from overheating is to apply passive cooling (Carroni et al., 2003): therein, only every second channel of the honeycomb reactor is coated (alternately coated structure), such that the flow in the uncoated channels cools the walls of the active channels. Another option is to induce finite‑rate surface chemistry with the use of diffusion barrier layers (thin inert porous layers coated on top of the catalytically active surface) or Pd‑based catalysts that self‑regulate their temperature by deactivating above ~750°C (at 1 bar). The foregoing analysis did not consider the impact of gas‑phase reactions. Homogeneous combustion in hydrogen‑rich fuels can actually have a moderating impact on the surface temperatures, as further discussed in Section 8.6.2.1. Another type of surface superadiabaticity occurs in the case of multiple or competing catalytic reaction pathways. In the fuel‑rich CPO approach in Figure 8.4, for example, the mildly exothermic partial oxidation or the endothermic catalytic steam reforming and dry reforming of methane are slow reactions compared to the fast and strongly exothermic total oxidation. The former reactions are not equilibrated in the typically short contact times (~10 ms) of CPO reactors, leading to superadiabatic surface temperatures (Veser and Frauhammer, 2000; Schneider et al., 2006), which have to be controlled by appropriate selection of the fuel‑to‑air stoichiometry or nonadiabatic operation. Interestingly, the measured catalyst surface temperatures are rather insensitive to considerable variations of the fuel‑to‑air ratio. In natural gas combustion, Smith et al. (2006) measured a modest 50°C difference in catalyst temperatures for a variation in the fuel‑rich mixture stoichiometry corresponding to a 300°C difference in adiabatic flame temperatures. Arguments have been put forth to explain this behavior under the control of combustion by a limiting airflow. However, the origins of this insensitivity are clearly kinetic: by decreasing the fuel‑rich stoichiometry, the excess combustion heat is used for the endothermic reforming reactions, resulting in higher syngas selectivities and thus surface temperature moderation. The same reason is also responsible for the extended extinction limits (CPO can be sustained at room inlet temperatures): Schneider et al. (2008) have shown that a decrease of the inlet temperature promotes the exothermic total oxidation (the syngas yields

233

Catalytic Combustion of Syngas 1500

Tad

1400

1300 1250

0.4

x=

0.2 0.0 0.0

1200 0

20

x (mm)

60

1m

m

40

mm 10

r (mm)

1350

1150

80

0.6

0.1 YO2 40

20 0.2

% Oxygen Conversion

100

x=

Surface Temperature (K)

1450

0 60

Figure  8.7  Computed axial profiles of surface temperature and oxygen conversion for catalytic combustion of a fuel‑rich (φ = 5.0) H2/air mixture in a 0.6 mm radius tubular channel, under transport‑limited operation. The inset figure provides radial profiles of O2 mass fraction at two selected axial locations; r = 0 and 0.6 mm are the centerline and wall, respectively. The line marked Tad denotes the adiabatic equilibrium temperature.

drop) such that the reactor temperature changes only modestly. Despite the moderate variations in average surface temperatures for appreciable changes in the inlet temperature or fuel‑to‑air stoichiometry, localized hot spots may pose a concern since total oxidation cannot be avoided at the very beginning of the catalytic section (Veser and Frauhammer, 2000; Schneider et al., 2006). In fuel‑rich combustion of a high-hydrogen‑content syngas fuel, the mechanism that controls the surface temperature is the diffusional imbalance of the limiting oxygen reactant. Figure 8.7 provides computed axial profiles of surface temperatures and oxygen fractional conversion for a fuel‑rich (φ = 5.0) H2/air mixture in a catalytic cylindrical channel with a diameter of 1.2 mm under transport‑limited operation (the latter is manifested by the zero oxygen concentrations at the catalytic surface; see inset of Figure 8.7). The inlet temperature and velocity are TIN = 380 K and UIN = 10 m/s, respectively. The pressure is 1 bar and the thermal conductivity of the solid is 16 W/mK, simulating a metallic FeCr‑alloy honeycomb structure (Figure  8.1b). The computations are carried out with a two-dimensional numerical model (see Section 8.6.1). Underadiabatic surface temperatures are attained over the channel length where oxygen is still available. Equation 8.2 also holds when the Lewis number of the limiting oxygen is used, which under fuel‑rich conditions is LeO ≈ 2.0. In the present case, Tad = 1468 K, and Equation 8.2 yields TW ≈ 1100 K. The lowest predicted temperatures appear at x ≈ 0, where the oxygen availability is the highest, and are somewhat higher than the predicted lowest values due to the upstream conduction of heat in the solid. For a catalytic reactor design with a fractional oxygen conversion, underadiabatic temperatures can be attained over the entire channel length, thus greatly aiding the reactor thermal management.

234

Synthesis Gas Combustion: Fundamentals and Applications

8.5 Catalysts for High-Temperature Applications In fuel‑lean combustion of methane or natural gas, the optimal catalyst should be active at the low compressor discharge temperatures (350 to 400°C), yet not lead to the full adiabatic combustion temperature to prevent catalyst deactivation or cause melting. Palladium offers the only practical solution, since it is the most active low‑temperature methane oxidation catalyst. Moreover, the catalytically active PdO component is decomposed at a pressure‑dependent temperature (~750°C at 1 bar and ~950°C at 15 bar) to relatively inactive metallic Pd (Farrauto et al., 1992). This phenomenon leads to a self-regulation of the palladium temperature and can be used to prevent catalyst overheating. Despite its high activity, Pd has light‑off and extinction temperatures well above 350°C, at least for the short residence times of gas turbines (catalytic reactor gas hourly spatial velocities (GHSVs) in excess of 106 h–1), thus requiring the use of a preburner. Apart from Pd, a number of high‑temperature stable complex metal oxide compositions have been investigated (see the review in Zwinkels et al., 1993), the most promising of all being substituted hexaaluminate materials (Eguchi and Arai, 1996). However, the enhanced thermal stability of oxide materials is not accompanied by a noble‑metal‑like activity. It has been clearly demonstrated (McCarty and Wise, 1990) that high catalytic activity is connected to low thermal stability. This inherent trade-off between stability and activity presents great challenges in the design of practical fuel‑lean catalytic combustion systems. Attempts to pursue full catalytic conversion using multicatalyst approaches (a palladium catalyst for light‑off and a more stable and less active metal oxide in the downstream region) have not progressed due to mechanical integrity issues and ever‑increasing final firing temperatures of gas turbines (>1400°C). The fuel‑rich combustion of methane circumvents, as discussed in Section 8.2, many of the activity problems and totally overcomes the issue of catalyst extinction, allowing for a wider selection of catalysts. Experiments on fuel‑rich methane combustion have shown the suitability of Pd, Pt, and Rh noble metals (Lyubovsky et al., 2003). Nonetheless, Rh is the preferred choice due to its higher hydrogen selectivity in CPO of methane (Hickman and Schmidt, 1993) and its better resistance to carbon deposition. Thus, a noble metal catalyst is also required in fuel‑rich combustion of natural gas for activity and selectivity reasons. In the catalytic combustion of syngas‑based fuels, both noble metal and metal oxide catalysts have been investigated. For biogas fuels, Pt‑ and Pd‑based catalysts have been tested by Zwinkels et al. (1993), Johansson et al. (1999), and Pocoroba et al. (2000), while metal oxide catalysts (hexaaluminates and perovskites) have been studied in Johansson et al. (2002) and Ersson et al. (2006). Hexaaluminates have also been tested in a model combustion with a CO/H2 syngas fuel (Groppi et al., 1996). Comparisons between noble‑metal-containing and metal oxide catalysts for biomas fuels (Pocoroba et al., 2000; Kusar et al., 2003) have shown the superiority of the former in terms of light‑off. For the high GHSV of power generation systems, noble metal catalysts are hence the best choice. Even if metal oxide catalysts prove appropriate for certain syngas fuels at turbine‑relevant conditions, issues of fuel flexibility (operation with either syngas or natural gas or co‑firing of both fuels) may again dictate the use of noble metal catalysts.

Catalytic Combustion of Syngas

235

Impurities pose a concern in catalytic combustion systems. Sulfur, in particular, at levels of a few ppm, has a negative impact on the performance of noble metal catalysts for methane oxidation (Thevenin et al., 2001). Sulfur impurities usually appear in the form of H2S, which is oxidized to SO2 and SO3 that reacts on the surface to form metal sulfates. This leads to catalyst poisoning, which is more pronounced at low temperatures where sulfate adsorption is favored. Noble metals are overall more resistant to sulfur poisoning than transition metal oxide catalysts, as the latter form more stable sulfates. The catalyst support can also impact the sulfur poisoning. In noble‑metal-supported catalysts, the use of alumina leads to preferential adsorption of the sulfates on the support rather than on the active sites (Lampert et al., 1997). For biomass‑derived syngas fuels, ash, water, sulfur, alkali, and chlorine compounds are present. The deactivation of supported Pd and Pt catalysts has been studied with respect to those components in Johansson et al. (1999). It has been shown that supported Pt catalysts undergo a strong deactivation during the ignition of carbon monoxide and hydrogen, while their activity in methane oxidation remains largely unchanged. Pd‑based catalysts, however, exhibit an opposite behavior; that is, they lose their activity more effectively in the oxidation of methane than in carbon monoxide or hydrogen. Finally, the catalyst substrate in power generation systems is usually metallic (e.g., FeCr‑alloy, as in Figure 8.1b), which is more resistant to thermal shock than ceramic materials. In addition, the honeycomb structure is created by stacking or winding corrugated FeCr‑alloy sheets (~25 to 50 µm thick), which gives added flexibility in applying different catalyst coatings at various reactor locations and also facilitates passive cooling measures (Carroni et al., 2003). The operational temperature of metal substrates, however, is limited to below 1000°C.

8.6 Catalytic Combustion of Syngas The main chemical difference between syngas‑based and natural gas fuels is the sizable H2 and CO content of the former. This section presents fundamental catalytic combustion aspects of fuel‑lean syngas fuels consisting of H2 and CO and the underlying hetero‑/homogeneous chemistry interactions between these two fuel components. Simulation results are summarized for different H2/CO compositions at turbine‑relevant conditions (for details, see Mantzaras (2008)), and pertinent experiments are brought to attention. Platinum is the chosen catalyst due to its well‑studied kinetics for the oxidation of fuel‑lean H2/air and CO/air mixtures and its good reactivity. Even though metal oxides have also been used as catalysts for H2/CO combustion (Groppi et al., 1996), their activity at the high spatial velocities encountered in power generation systems is not warranted. The combustion characteristics of the two syngas components (H2 and CO) are first identified. The heterogeneous chemical coupling of CO and H2 is then addressed, and the impact of gas‑phase chemistry in realistic reactor geometries and pressures is clarified. Finally, reactor thermal management issues are presented, in light of the diffusional imbalance of hydrogen that gives rise to superadiabatic surface temperatures.

236

Synthesis Gas Combustion: Fundamentals and Applications

8.6.1 Numerical Results The underlying processes in a single catalytic channel of a honeycomb reactor (Figure  8.1c) are simulated with an elliptic two-dimensional numerical code. The dynamic light‑off behavior is addressed with a transient model, whereas stationary performance is tackled with a steady model (for model details see Schneider et al., 2008). The fuel/air is preheated in the range of 300 to 700 K and the pressure varies from 1 to 15 bar. For catalyst thermal stability reasons, the syngas has such a composition that the adiabatic equilibrium temperature does not exceed 1300 K. A catalytic channel is modeled as an equivalent cylindrical tube with a length of L = 75 mm and an internal radius of r = 0.6 mm (Figure 8.8), leading to a confinement (surface‑to‑volume ratio) typical of those encountered in commercial catalytic reactors (Groppi et al., 1996; Eriksson et al., 2006; Schneider et al., 2006). The solid substrate has a thickness of δ = 50 µm. Half of it is included in the numerical domain of Figure 8.8 due to consideration of adjacent channels. The solid thermal conductivity is λs = 16 W/mK, corresponding to a FeCr‑alloy metallic honeycomb structure (as the one in Figure 8.1b). The radiation emissivity of the inner channel surfaces is ε = 0.6, while the inlet and the outlet sections are treated as black bodies (ε = 1.0). For the oxidation of H2/CO mixtures over Pt, the detailed heterogeneous reaction scheme of Deutschmann et al. (2000) is used (see Table  8.1). This scheme reproduces catalytic ignition and steady combustion characteristics of pure H2, CO, and CH4 fuels as well as their mixtures (Deutschmann et al., 1996, 2000; Appel et al., 2002; Reinke et al., 2004). The inclusion of homogeneous chemistry in catalytic combustion systems deserves some attention. Prior to homogeneous ignition, there is appreciable heterogeneous fuel depletion (see, e.g., Figure 8.2b), which reduces the already considerably fuel‑lean inlet stoichiometries to ultralean levels. Moreover, the temperatures in catalytic combustion are moderate (up to 1400 K) and the heterogeneously formed major species (i.e., H2O) can be very efficient collision partners in gas‑phase chain-terminating reactions (Bui et al., 1996; Appel et al., 2002; Reinke et al., 2005). On the other hand, the hetero‑/homogeneous radical coupling via adsorption‑desorption reactions is generally weak (Appel et al., 2002; Reinke et al., 2005). All aforementioned factors greatly impact the aptness of gas‑phase chemical reaction schemes developed in the absence of catalytic reactions. Comparative studies of various H/O and C1/H/O mechanisms during catalytic combustion of H2 L = 75 mm Adiabatic surface

qrad

Heat conduction in solid

r

UIN, TIN

qrad

x

Platinum catalyst

δ = 25 µm r = 0.6 mm

Symmetry axis

Figure 8.8  Channel geometry used in the numerical simulations.

qrad

237

Catalytic Combustion of Syngas

Table 8.1 Catalytic Reaction Scheme for H2/CO Oxidation on Pt A(γ)

b

0.023 1.8 × 1021 4.5 × 1010 1.0 1.0 0.75 1.0 1.6 × 1020

0.0 –0.5 0.5 0.0 0.0 0.0 0.0 0.5

   0.0    0.0    0.0    0.0    0.0    0.0    0.0    0.0

Surface Reactions S9. H(s) + O(s) = OH(s) + Pt(s) S10. H(s) + OH(s) = H2O(s) + Pt(s) S11. OH(s) + OH(s) = H2O(s) + O(s) S12. C(s) + O(s) → CO(s) + Pt(s) S13. CO(s) + Pt(s) → C(s) + O(s) S14. CO(s) + O(s) → CO2(s) + Pt(s)

3.7 × 1021 3.7 × 1021 3.7 × 1021 3.7 × 1021 1.0 × 1018 3.7 × 1021

0.0 0.0 0.0 0.0 0.0 0.0

  11.5   17.4   48.2   62.8 184.0 105.0

Desorption Reactions S15. 2O(s) → O2 + 2Pt(s) S16. 2H(s) → H2 + 2Pt(s) S17. H2O(s) → H2O + Pt(s) S18. OH(s) → OH + Pt(s) S19. CO2(s) → CO2 + Pt(s) S20. CO(s) → CO + Pt(s)

3.7 × 1021 3.7 × 1021 1.0 × 1013 1.0 × 1013 1.0 × 1013 1.0 × 1013

0.0 0.0 0.0 0.0 0.0 0.0

213.2 – 60θO   67.4 – 6θH   40.3 192.8   20.5 125.5

Adsorption Reactions S1. S2. S3. S4. S5. S6. S7. S8.

O2 + 2Pt(s) → 2O(s) O2 + 2Pt(s) → 2O(s) H2 + 2Pt(s) → 2H(s) H + Pt(s) → H(s) O + Pt(s) → O(s) H2O + Pt(s) → H2O(s) OH + Pt(s) → OH(s) CO + Pt(s) → CO(s)

E

Note: In the surface and desorption reactions, the reaction rate coefficient is k = ATbexp(–E/RT), A [mole-cm-Kelvin-s] and E [kJ/mol]. In the adsorption reactions, except S2, S3, and S8, A denotes a sticking coefficient (γ). Reactions S1 and S2 are duplicate. Reactions S3 and S8 have a Pt order of 1 and 2, respectively. The suffix (s) denotes a surface species and θi the coverage of surface species i. The surface site density is Γ = 2.7 × 10–9 mol/cm2. Source: From Deutschmann et al., 2000. (With permission.)

or CH4 over Pt have revealed significant discrepancies in their capacity to reproduce measured homogeneous ignition characteristics. Appel et al. (2002) have validated the gas‑phase scheme of Warnatz et al. (1996) in hetero‑/homogeneous combustion of hydrogen over Pt at atmospheric pressure. Recent studies at pressures up to 10 bar (Mantzaras et al., 2008) have examined the gas‑phase mechanisms of Li et al. (2003) and Warnatz (2005). For the present simulations, the hydrogen gas-phase scheme of Warnatz (2005) is augmented by the C1 mechanism of Warnatz et al. (1996), which has recently been adapted to reproduce high‑pressure CH4/air homogeneous ignition experiments over Pt (Reinke et al., 2005). The H2/CO relevant part of this mechanism is shown in Table 8.2.

238

Synthesis Gas Combustion: Fundamentals and Applications

Table 8.2 Homogeneous Chemical Reaction Mechanism for H2/CO b

E

–0.46 2.90 1.52 2.40

70.09 26.44 14.47 –8.84

H2/O2 Dissociation-Recombination 1.01 × 1017   5. H + H + M = H2 + M   6. O + O + M = O2 + M 5.40 × 1013 2.19 × 1022   7. H + OH + M = H2O + M

–0.6 0.00 –2.00

0.00 –7.40 0.00

HO2 Formation-Consumption   8. H + O2 + M = HO2 + M H + O2 + M = HO2 + M   9. HO2 + H = H2 + O2 10. HO2 + H = OH + OH 11. HO2 + H = H2O + O 12. HO2 + O = OH + O2 13. HO2 + OH = H2O + O2

1.47 × 1012 2.11 × 1018 1.05 × 1014 4.46 × 1014 1.45 × 1012 1.63 × 1013 3.91 × 1016

0.60 –0.80 0.00 0.00 0.00 0.00 0.00

0.00 0.00 8.56 5.82 0.00 –1.86 88.79

H2O2 Formation-Consumption 14. 2HO2 = H2O2 + O2 15. 2HO2 = H2O2 + O2 16. OH + OH + M = H2O2 + M OH + OH + M = H2O2 + M 17. H2O2 + H = H2 + HO2 18. H2O2 +H = H2O + OH 19. H2O2 + O = OH + HO2 20. H2O2 +O = H2O + O2 21. H2O2 + OH = H2O + HO2 22. H2O2 + OH = H2O + HO2

4.22 × 1014 1.32 × 1011 1.57 × 1013 5.98 × 1019 1.68 × 1012 1.02 × 1013 4.21 × 1011 4.21 × 1011 1.64 × 1018 1.90 × 1012

0.00 0.00 0.00 –0.80 0.00 0.00 0.00 0.00 0.00 0.00

50.14 –6.82 0.00 0.00 15.71 14.97 16.63 16.63 123.05 1.79

CO Reactions 23. CO + OH = CO2 + H 24. CO + HO2 = CO2 + OH 25. CO + O + M = CO2 + M 26. CO + O2 = CO2 + O

4.76 × 107 1.50 × 1014 7.10 × 1013 2.50 × 1012

1.23 0.00 0.00 0.00

0.29 98.70 –19.00 200.00

HCO Reactions 27. HCO + M = CO + H + M 28. HCO + H = CO + H2 29. HCO + O = CO + OH 30. HCO + O = CO2 + H 31. HCO + OH = CO + H2O

3.95 × 1014 9.00 × 1013 3.00 × 1013 3.00 × 1013 1.00 × 1014

0.00 0.00 0.00 0.00 0.00

70.30 0.00 0.00 0.00 0.00

H2/O2 Reactions   1. H + O2 = O + OH   2. O + H2 = H + OH   3. H2 + OH = H2O + H   4. OH + OH = O + H2O

A 5.20 × 10 8.97 × 103 2.17 × 108 3.57 × 104

15

239

Catalytic Combustion of Syngas

Table 8.2 (continued) Homogeneous Chemical Reaction Mechanism for H2/CO HCO Reactions 32. HCO + O2 = CO + HO2 33. HCO + HCO = CH2O + CO

A

b

3.00 × 10 3.00 × 1013 12

0.00 0.00

E 0.00 0.00

Note: Reaction rate k = ATbexp(–E/RT), A [mole‑cm-Kelvin-s] and E [kJ/mol]. Third-body efficiencies: ω·(H2O) = 6.5, ω· (O2) = ω(N2) = 0.4, ω·(H2) = 1.0, ω·(CO·) = 0.75, ω·(CO2) = 1.5· . The reaction pairs (14, 15) and (21, 22) are duplicate. Reactions 8 and 16 are Troe reactions centered at 0.5 (second entries are the low-pressure limits). Source: From Warnatz et al., 1996, 2005.

Before discussing the hetero‑/homogeneous combustion of H2/CO mixtures, it is instructive to first address the combustion of pure H2 and CO fuels. This facilitates the identification of the particular characteristics of each fuel component and aids the subsequent discussion on the hetero‑/homogeneous chemical coupling during combustion of syngas fuels with varying H2/CO compositions.

8.6.2 Catalytic Combustion of Hydrogen/Air Mixtures The catalytic combustion of fuel‑lean hydrogen/air mixtures is first presented in order to identify key issues regarding the reactor thermal management and the impact of gaseous chemistry. 8.6.2.1 Impact of Gas‑Phase Chemistry Predicted axial profiles of the catalytic (C) and gaseous (G) hydrogen conversion rates as well as of the wall temperature (TW) are provided in Figure 8.9 for H2/air mixtures at different equivalence ratios and p = 1 bar. The G profiles of Figure 8.9 have been computed by integrating the local volumetric gaseous reaction rates across the channel radius. The upstream surface temperatures exceed by up to 300 K the adiabatic equilibrium temperatures (Tad in Figure 8.9) due to the diffusional imbalance of hydrogen, as explained in Section 8.4. Complete hydrogen consumption is achieved in all cases at the channel exit, resulting in wall temperatures in the far downstream regions that are only up to 30 K lower than Tad due to the imposed radiation heat losses. The superadiabatic temperatures at the entry compound the catalytic combustion of hydrogen or hydrogen‑rich fuels, and require careful strategies for reactor thermal management. The contribution of the gaseous pathway is practically zero at the lower equivalence ratios (Figure 8.9a and b). However, the importance of homogeneous combustion is increasing at higher φ (Figure 8.9c and d). A usual premise in reactor design

240

Synthesis Gas Combustion: Fundamentals and Applications 6

1350 (a)

= 0.10

TW

Hydrogen Conversion Rate (g/m2s)

4 TW

3 2

C

1 0

6

1190

5

950 870 5

15 30 45 60 75 0

= 0.20

4 C

C

(d) = 0.24

Tad G

Tad

1

15 30 45 60 75

Tw

TW

3

5

790 1650 1570 1490 1410 1330 1250 1170

G 0

1110 1030

G

(c)

0

1270

Tad

0

2

= 0.15 T ad

C

G

5

(b)

Wall Temperature (K)

5

15 30 45 60 75 0

5

15 30 45 60 75

1090

Axial Distance x (mm)

Figure 8.9  Computed axial profiles of catalytic (C, solid lines) and gas‑phase (G, dotted lines) conversion rates of hydrogen, and wall temperature (TW, dashed lines) in the channel of Figure 8.8. Four H2/air equivalence ratios are shown in (a–d). In all cases p = 1 bar, UIN = 20 m/s, and TIN = 600 K. The horizontal lines marked Tad denote the adiabatic equilibrium temperature. For clarity, the first 10 mm is shown in an expanded scale.

considers gaseous combustion detrimental to the catalyst and substrate integrity, and hence homogeneous ignition inside the catalyst module is undesirable. In the case of hydrogen fuel, however, gaseous combustion is beneficial, as it moderates the surface temperatures. This is clarified in Figure 8.10a, where the case of φ = 0.24 in Figure 8.9d is compared against the same case computed without the inclusion of gaseous chemistry. It is seen that homogeneous chemistry decreases the surface temperatures by as much as 90 K and the peak temperature by ~30 K. This rather surprising behavior has been observed experimentally and further clarified theoretically by Appel et al. (2002, 2005b). In fuels with Le < 1 the flame is confined near the wall, as also shown by the OH radical map in Figure 8.11a (pertaining to the case in Figure 8.9d), thus shielding the catalyst from the hydrogen‑rich channel core and reducing the heterogeneous conversion that is responsible for the superadiabatic temperatures. Furthermore, the near‑wall flame confinement leads always to combined heterogeneous and homogeneous conversions due to reduced residence times inside the narrow gaseous combustion zone and the subsequent leakage of the hydrogen fuel through this zone to the catalytic surface (see, for example, curves C and G in Figure 8.9d at 1 mm < x < 15 mm). This is in contrast to methane combustion, whereby upon homogeneous ignition the dominant fuel conversion pathway is the homogeneous one (Dogwiler et al., 1999; Reinke et al., 2005).

241

Catalytic Combustion of Syngas 6 5

TW

TW

4

1650

(a) r = 0.6 mm

1570 1490 1410

2

C

1

G

Tad

1330

C

1250 1170

0 0

5

15

6

30

45

60

75

1650

(b) r = 1.2 mm

5 TW

4

1

C

1410 1330 1250

C

0 0

1490

Tad

G

2

1570

TW

3

5

15

1090

Wall Temperature (K)

Hydrogen Conversion Rate (g/m2s)

3

1170 30

45

60

75

1090

Axial Distance x (mm)

Figure 8.10  Predicted axial profiles of catalytic (C, solid lines) and gas‑phase (G, dotted lines) conversion rates of hydrogen, and wall temperature (TW, dashed lines). Black lines: Catalytic and gaseous chemistry included. Gray lines: Only catalytic chemistry included. Simulations for channel radii: (a) 0.6 mm and (b) 1.2 mm. In all cases p = 1 bar, UIN = 20 m/s, φ = 0.24, and TIN = 600 K. For clarity, the first 10 mm is shown in an expanded scale.

The heterogeneous reactivity of hydrogen on Pt is high, leading to practically transport‑limited catalytic conversion at realistically high reactor velocities. This is illustrated by the very low near‑wall levels of hydrogen in Figure 8.11b (already at the beginning of the channel) and also by the high catalytic conversion rates at the channel entry (see curves C in Figure 8.9 at x ≈ 0). The light‑off length (using the rather strict definition as the axial position where the hydrogen wall levels drop to 10% of the corresponding centerline values) is, for all cases, less than 2.5 mm. The computed results in Figure 8.9 indicate that homogeneous combustion can be important at realistically large channel confinements (i.e., small radii, such as r = 0.6 mm). An increase of the channel radius to r = 1.2 mm enhances the contribution of the

242

Synthesis Gas Combustion: Fundamentals and Applications

r (mm)

0.6

(a) OH

–0.6 0.6

–0.6

(b) H2

10

0 min

x (mm)

20

30 max

Figure  8.11  Computed two-dimensional species mass fraction distributions for the case of Figure 8.9d: (a) OH and (b) H 2. For clarity, only the first 30 mm of the channel is shown. The OH and H 2 mass fractions range from 0.0 to 9.95 × 10 –4 and from 0.0 to 6.99 × 10 –3, respectively.

gaseous pathway, as shown in Figure 8.10b. In the wider channel in Figure 8.10b, the moderating impact of gaseous chemistry on the surface temperature is stronger, resulting in peak surface temperatures ca. 100 K lower than those of the narrower channel (Figure 8.10a). The aforementioned interplay of the hetero‑/homogeneous reaction pathways and its coupling to fluid transport is hence quite rich, suggesting certain reactor design procedures. Larger hydraulic diameter channels are preferable in moderating the surface temperatures during hydrogen and hydrogen‑rich catalytic combustion. The reduction of the surface‑to‑volume ratio when using wider channels is not that critical, because the catalytic conversion is high (given the very large molecular diffusivity and catalytic reactivity of hydrogen). Even so, increasing the channel hydraulic diameter alone may not be sufficient to control the temperatures to acceptable levels. Given the fact that syngas compositions with corresponding adiabatic equilibrium temperatures of at least 1400 K are required for gas turbines, Figures  8.9 and 8.10 suggest that additional passive cooling measures may be needed. Such measures include honeycomb reactors with alternately coated channels (sequence of catalytically active and inactive channels), as in Carroni et al. (2003) and Appel et al. (2005a). The passive cooling approach necessitates a CST combustion methodology (Figure 8.3), that is, the inclusion of a postcatalyst flame zone to complete the conversion of the hydrogen flowing through the noncatalytic channels. 8.6.2.2 Effect of Pressure Before addressing the hetero‑/homogeneous chemistry coupling at high pressures, a brief discussion on the pure homogeneous ignition characteristics of hydrogen is necessary (Mantzaras, 2008). Ignition delay times are provided in Figure 8.12 for a φ = 0.28 H2/air mixture at different pressures. At moderate temperatures of 1000 K or less, the gaseous reactivity decreases (the ignition delay increases) with rising

243

Catalytic Combustion of Syngas

T=

1e–2

0K

100

Ignition Delay (sec)

T= 1e–3

0K

105

T=

0K

110

T=

0K

115

1e–4

T

=

00

12

T

1e–5 1

2

3 4 5 Pressure (bar)

=

10

K

50

12

K

15

Figure 8.12  Predicted gas‑phase ignition delays for a φ = 0.28 hydrogen/air mixture as a function of pressure at different temperatures.

pressure, and for p > 8 bar it practically remains constant. At higher temperatures the reactivity initially increases with rising pressure and then drops, with the turning point shifted to higher pressures for higher temperatures. The implication for practical systems is that at moderately high channel wall temperatures (which for the inhomogeneous two-dimensional configuration in Figure 8.8 also depend on the inlet temperature), the onset of gas‑phase ignition is suppressed with rising pressure. This has also been verified in recent experiments (Mantzaras et al., 2008) at pressures up to 10 bar in an optically accessible, water‑cooled rectangular catalytic reactor (Pt‑coated) with a transverse separation of 7 mm. As seen in both measurements and predictions in Figure 8.13, the suppression of gaseous combustion for surface temperatures of up to 1200 K and inlet temperature of 310 K is already substantial at 4 bar. Figure 8.14 provides predicted catalytic and gas‑phase hydrogen conversion rates at 5 and 15 bar. The other conditions are the same as in the p = 1 bar case of Figure 8.9c with the exception of the inlet velocity, which is reduced with increasing pressure so as to maintain the same mass throughput. Comparison of Figures 8.9c and 8.14 reveals that gaseous combustion is now promoted at high pressures. For the modest imposed heat losses and for the significant preheat TIN = 600 K, the surface temperatures are high enough (Figure 8.14) such that a rise in pressure promotes gas‑phase ignition, in accordance with Figure  8.12. As also explained in Section 8.6.2.1, an enhanced homogeneous conversion moderates the surface temperatures. Therefore, the peak temperatures for p = 5 and 15 bar in Figure 8.14 are lower than the peak temperature for p = 1 bar in Figure 8.9c.

244

Synthesis Gas Combustion: Fundamentals and Applications 1a

7 mm

p = 1 bar

1b 0 2a

408 816 p = 4 bar

2b 0

5

0

10

40 1250

80

x (mm)

120

160

200

1c

Wall Temperature (K)

1200 1150 1100 1250

2c

1200 1150 1100 0

40

80

x (mm)

120

160

200

Figure 8.13  Catalytic combustion of fuel‑lean (φ = 0.28) H2/air mixture in a rectangular channel: (a) LIF‑measured OH distribution, (b) numerically predicted OH distribution (ppmv), and (c) measured wall temperatures (upper wall, circles; lower wall, triangles). Pressure of 1 bar (1) and 4 bar (2). The inlet Reynolds number is 2000 ant TIN = 310 K in both cases. The arrows in case 1 define the onset of homogeneous ignition. (Adapted from Mantzaras et al., 2008. With permission.)

In practical catalytic reactors with narrow channels of ~1 mm in hydraulic diameter, the aforementioned reduction of the gaseous reactivity with increasing pressure at moderate temperatures may become an irrelevant issue: gas‑phase chemistry can altogether be minimal even at p = 1 bar due to the increased surface‑to‑­volume ratios that in turn allow for complete hydrogen catalytic consumption during the elongated gas‑phase induction zones. It is emphasized that the catalytic conversion is aided by the large diffusivity of hydrogen and its high reactivity on platinum, even at very modest surface temperatures. 8.6.2.2 Light-Off Temperatures The previous steady‑state computations cannot unequivocally determine whether the obtained stable burning solutions are feasible for specific initial conditions (e.g., it

245

6

(a) p = 5 bar

5

(b) p = 15 bar

UIN = 4.0 m/s

4

UIN = 1.33 m/s

Tw

3 2 1

C

0 0

G

5

Tad

15 30 45 60 75 0

C

1410 1330 Tad

5

1570 1490

Tw G

1650

1250 1170

Wall Temperature (K)

H2 Conversion Rate (g/m2s)

Catalytic Combustion of Syngas

1090 15 30 45 60 75

Axial Distance x (mm)

Figure 8.14  Computed axial profiles of catalytic (C, solid lines) and gaseous (G, dotted lines) conversion rates of H2, and wall temperature (TW, dashed lines): (a) p = 5 bar and (b) p = 15 bar. In both cases TIN = 600 K and φ = 0.20. The inlet velocity is reduced with rising pressure so as to maintain the same mass throughput. For clarity, the first 10 mm is shown in an expanded scale.

is realistic to consider an initial solid temperature equal to the gas inlet temperature). Transient catalytic ignition (light‑off) computations are hence performed for the conditions of Figure 8.9 and inlet temperatures lower than 600 K. Equivalence ratios of 0.20 and 0.24 are considered, that is, mixtures with sufficient exothermicity for power generation. The light‑off time is defined as the time required for the solid to reach within 5 K of its corresponding steady‑state temperature, provided that the steady solution corresponds to a vigorous burning state (wall temperatures close to Tad or greater) and not to a weakly reacting state. For p = 10 bar, UIN = 2 m/s, TIN = 600 K, and φ = 0.20, the steady solution is reached after 0.2 s; by reducing the inlet temperature to 380 K, the corresponding time increases to 3.2 s. It should be pointed out that the computed light‑off times are only indicative of the easiness of catalytic ignition since they also depend on specific reactor and catalyst parameters such as linear velocity, geometry, heat loss mechanisms, catalyst dispersion, and so forth. Nonetheless, for the particular reactor parameters in this study, which resemble those encountered in gas turbines, ignition could always be achieved for inlet temperatures in the range of 360 to 380 K. Therefore, the high catalytic reactivity of hydrogen appears, at a first instance, attractive for the catalytic combustion of syngas fuels. Detailed transient light‑off simulations will be presented in Section 8.6.4.3. In conclusion, hydrogen catalytic combustion can be initiated at industrially relevant velocities/pressures and at inlet temperatures as low as 360 K. Of major concern in hydrogen catalytic combustion is the reactor thermal management due to the attained superadiabatic surface temperatures. Gas‑phase combustion cannot always be ignored, particularly at high surface temperatures and modest geometrical confinements. Moreover, gaseous combustion is inhibited with increasing pressure, due to the intrinsic hydrogen kinetics, for wall temperatures up to ca. 1200 K (and TIN ≈ 300 K) and is promoted for higher wall and inlet temperatures. Finally, gaseous combustion moderates the reactor temperatures by suppressing the heterogeneous conversion that drives the superadiabaticity.

246

Synthesis Gas Combustion: Fundamentals and Applications

8.6.3 Catalytic Combustion of CO/Air Mixtures

80

(b) = 0.24

(a) = 0.20

60

TW

40

C

TW

20

Tad

Tad

1500 1420 1340

C

1260

0

1180 0

5

15 30 45 60 75 0 5 Axial Distance x (mm)

15 30 45 60 75

Wall Temperature (K)

CO Conversion Rate (g/m2s)

The combustion of carbon monoxide is simpler than that of hydrogen due to the absence of homogeneous chemistry (gas‑phase combustion of CO cannot be initiated in dry air, at temperatures of interest, without the presence of moisture or hydrogen) (Glassman, 1996). Another factor that leads to better control of the surface temperatures is the nearly diffusionally neutral transport of CO. On the other hand, CO can be oxidized catalytically in dry air over noble metal or metal oxide catalysts (Arnby et al., 2004; Mhadeshwar and Vlachos, 2005; Shiau et al., 2006). Figure 8.15 provides axial profiles of catalytic CO conversion rates and surface temperatures for two stoichiometries, with UIN = 20 m/s, TIN = 600 K, and p = 1 bar. The surface temperatures do not exceed the adiabatic equilibrium values, thus greatly simplifying the reactor design. For the conditions in Figure 8.15, catalytic ignition is achieved at x ≈ 2.1 mm for φ = 0.20, and at x ≈ 1.6 mm for φ = 0.24. The shorter light‑off distance at richer stoichiometries is an outcome of combined chemical and thermal effects (reaction exothermicity) during steady combustion and should not be confused with the observed behavior in dynamic catalytic ignition. In transient catalytic ignition (Deutschmann et al., 1996), chemical effects dominate since the temperature is practically constant over the catalytic induction zone. In this case CO self‑inhibits its ignition through an excessive surface coverage of CO(s) blocking the adsorption of oxygen. The same type of self‑inhibition also controls hydrogen catalytic ignition, whereby H(s) blocks the adsorption of oxygen (Deutschmann et al., 1996). When the pressure is increased from 1 to 15 bar and the mass throughput is kept constant, the fuel conversion and surface temperatures virtually collapse onto each other (Mantzaras, 2008). This is because under transport‑limited catalytic operation, and in the absence of gaseous chemistry, the only controlling parameter in channel flow combustion is the Reynolds number (Mantzaras and Benz, 1999), which is fixed for a given mass throughput.

1100

Figure 8.15  Computed axial profiles of catalytic (C, solid lines) conversion rates of CO, and wall temperature (TW, dashed lines) for two CO/air equivalence ratios. In both cases p = 1 bar, UIN = 20 m/s, and TIN = 600 K. The horizontal lines marked Tad denote the adiabatic equilibrium temperature. For clarity, the first 10 mm is shown in an expanded scale.

Catalytic Combustion of Syngas

247

Transient light‑off simulations indicate that for the turbine‑relevant stoichiometries of φ = 0.20 and 0.24 and the mass throughputs in Figure 8.15, the minimum inlet temperatures required for catalytic ignition range between 650 and 700 K. For example, when p = 10 bar, UIN = 2 m/s, φ = 0.20, and TIN = 700 K, the corresponding light‑off time is 4.5 s (transient catalytic ignition of CO and its comparison with CO/H2 mixtures will be presented in Section 8.6.4.3). Carbon monoxide is thus less reactive than hydrogen on Pt, suggesting—at a first instance—that the addition of the latter may aid the ignition of the former. It is finally noted that for certain metal oxide catalysts, the reactivity of CO can actually be higher than that of H2 (Groppi et al., 1996). Although this property may appear attractive for specific syngas fuels with large CO contents, care must be exercised in using such catalysts at the high linear velocities of gas turbines.

8.6.4 Catalytic Combustion of H2/CO Mixtures The hetero‑/homogeneous chemical coupling of the two fuel components is discussed first, followed by reactor thermal management issues. Finally, transient light‑off studies are addressed. 8.6.4.1 Hetero-/Homogeneous Chemistry Coupling Computations are initially performed for the geometry in Figure 8.8 under fixed wall temperatures in order to isolate thermal from chemical effects. Various H2/CO/air mixtures are investigated, having a combined H2 and CO content of 7.75% vol. To identify the chemical impact of the added CO, additional computations are performed by replacing CO with a fictitious species CO* that has the same thermodynamic and transport properties as CO, but does not participate in any heterogeneous or homogeneous chemical reaction. Figure 8.16 shows axial profiles of the catalytic (C) and gaseous (G) hydrogen and carbon monoxide conversion rates (black lines) at four different wall temperatures for a H2/CO/air mixture with 7.25% vol. H2 and 0.5% vol. CO, p = 10 bar, UIN = 2 m/s, and TIN = 600 K. In the same figure, profiles are also given for a corresponding H2 /CO*/air mixture in terms of the relevant catalytic and gaseous hydrogen conversion rates (gray lines). For wall temperatures of 1300 and 800 K (Figure 8.16a and b), the hydrogen C and G curves of the H2/CO/air and H2/CO*/air mixtures virtually coincide. Homogeneous combustion is present at TW = 1300 K, as shown by the H2 and CO gaseous conversion curves in Figure 8.16a. The gaseous conversion of hydrogen is practically unaffected by the presence of CO and its accompanying gasphase or catalytic chemistry. This is because the homogeneous combustion of CO is initiated by OH radical attack on CO, and as such, it does not deplete the hydrogen fuel to any noticeable extent (the initiation step OH + CO = CO2 + H is followed by the attack of H on O2 producing OH; therefore, OH serves as a homogeneous catalyst that is not overdepleted). Furthermore, the catalytic CO chemistry does not affect the homogeneous combustion of hydrogen since the hetero‑/homogeneous radical coupling (notably via O in the case of CO fuel) is weak, and also there are no major products in CO combustion that can couple as effectively as H2O with the gaseous

248

Synthesis Gas Combustion: Fundamentals and Applications 1.0 CH2, CH2

H2 and CO Conversion Rates (g/m2s)

0.6

CH2, CH2

GH2, GH2

0.4

CCO

CCO

0.2 GCO

0.0 1.0

(b) TW = 800 K

(a) TW = 1300 K

0.8

0

5

10

15

30 45 60 75 0

5

10

15

(d) TW = 550 K

(c) TW = 700 K

0.8

CH2

0.6 0.4

CCO

30 45 60 75

CH2

CH2

0.2

CH2, CCO

0.0 0

5

10

15

30 45 60 75 0 5 10 Axial Distance x (mm)

15

30 45 60 75

Figure 8.16  Computed axial profiles of catalytic (C, solid lines) and gaseous (G, dotted lines) conversion rates of CO and H2, for four constant wall temperatures. The fuel consists of 7.25% H2 and 0.5% CO vol., with the balance air, p = 10 bar, UIN = 2 m/s, and TIN = 600 K (black lines). The C and G conversions of H2 are also provided when CO is replaced by inert CO* (gray lines). In (a) and (b) the black and gray CH2 lines coincide. For clarity, the first 20 mm is shown in an expanded scale.

chemistry of hydrogen (Appel et al., 2002). In contrast, the gaseous combustion of CO is crucially dependent on hydrogen. The OH radicals that initiate the gaseous combustion of CO are provided by the hydrogen homogeneous pathway; the hydrogen catalytic pathway itself is a poor producer of radicals so as to appreciably affect the gaseous combustion of CO. Finally, the heterogeneous and homogeneous pathways convert CO in parallel over most of the channel length (Figure 8.16a), since at the moderate temperatures of catalytic combustion systems the gaseous oxidation of CO is slow. The catalytic conversion rate of hydrogen is unaffected by the presence of CO for surface temperatures at least as low as 800 K (see Figure  8.16a and b). At sufficiently high temperatures, the surface is primarily covered by O(s) and free platinum sites (see Figure  8.17a). The predictions with H2/CO*/air mixtures (not shown in Figure 8.17) reveal practically the same coverage for O(s) and Pt(s). The free site coverage is sufficient to accommodate the heterogeneous oxidation of both fuel components, which proceeds without any appreciable chemical interaction between H2 and CO. Nevertheless, as the wall temperature is reduced to 700 K or less, there is a marked deviation in the hydrogen conversion rates (Figure 8.16c).

249

Catalytic Combustion of Syngas (a) TW = 1300 K

O(s)

100 10–1

Pt(s)

10–2

OH(s)

10–3

H2O(s)

10–4 10–5

H(s)

10–6

CO(s)

Surface Coverage

10–7 10–8 10–9 10

CO ( 2 s)

–10

100 10–1

H(s)

10–2

Pt(s)

10–3 10–4

Pt(s)

CO(s )

H2O(s)

10–5

H2 O

(s)

OH(s)

10–6 10–7

O(s)

10–8

C(s)

10–9 10–10

(b) TW = 700 K O(s) OH(s)

CO(s)

H(s)

CO ( 2 s) 0

15

30 45 Axial Distance x (mm)

60

75

Figure 8.17  Surface coverage for the conditions of Figure 8.16 and wall temperatures of (a) 1300 K and (b) 700 K.

For a substantial length (down to x ≈ 37 mm), CO inhibits the catalytic conversion of hydrogen as seen by comparing the black and gray CH2 curves of Figure 8.16c; over this reactor extent, the main surface coverage is CO(s) (Figure 8.17b), greatly reducing the O(s) and free sites. At x ≈ 37 mm, there is an abrupt catalytic ignition of CO that depletes rapidly this species, thus reducing CO(s) and increasing the O(s) and OH(s) coverage. Complete conversion of hydrogen is attained at the reactor exit, such that the areas under the black and gray CH2 curves in Figure 8.16c are equal. At even lower surface temperatures (Figure 8.16d, TW = 550 K), CO(s) blocking dominates due to the high sticking coefficient of CO (Table  8.1), thus inhibiting the catalytic conversion not only of hydrogen (which would otherwise occur even at this low surface temperature in the absence of CO) but also of CO itself. For surface temperatures below about 700 K, carbon monoxide suppresses the hydrogen catalytic conversion even for a small 0.5% vol. content of the former species (Figure  8.16c). It turns out that even a considerably higher CO content (3.75% vol. H2 and 4.00% vol. CO) does not alter appreciably this limit temperature

250

Synthesis Gas Combustion: Fundamentals and Applications

(Mantzaras, 2008). Computations at pressures of 1 and 15 bar (maintaining the same mass throughput as in Figure 8.16) have also shown that the transition temperature of ~700 K is nearly independent of pressure. In conclusion, at temperatures above ca. 700 K, the chemical coupling between the catalytic pathways of hydrogen and carbon monoxide is minimal. At sufficiently high temperatures, the homogeneous chemistry of hydrogen is practically unaffected by the presence of CO, while the CO gaseous pathway is crucially dependent on the gas‑phase hydrogen chemistry. At surface temperatures below 700 K there is a strong catalytic chemistry coupling between H2 and CO, with the latter species inhibiting the conversion of the former. 8.6.4.2 Surface Temperatures Steady computations are used in order to determine the maximum surface temperatures attained during catalytic combustion of H 2/CO/air mixtures. Axial profiles of the computed catalytic and gaseous conversion rates as well as of the surface temperatures are presented in Figure  8.18 for various H 2/CO compositions. For all cases in Figure 8.18, p = 1 bar, TIN = 600 K, and UIN = 20 m/s, while the sum of the H 2 and CO compositions is fixed to 7.75% vol. Additional plots are

3 2

H2 and CO Conversion Rates (g/m2s)

TW

CH2

(a) 7.25% H2 0.50% CO

Tad

GH2

0 0

5

3 2

TW

1

Tad

0

5

1410

1250 1170

15 30 45 60 75 0

5 (d) 0.50% H2 7.25% CO

CCO Tad

TW

CH2

0

1490

1330

CH2

GCO

(c) 1.00% H2 6.75% CO

4

(b) 3.75% H2 4.00% CO

TW

CCO

1

CCO

15 30 45 60 75 1490

CCO

1410 Tad

1330 1250

CH2 15 30 45 60 75 0 5 Axial Distance x (mm)

Wall Temperature (K)

4

1170 15 30 45 60 75

Figure 8.18  Computed axial profiles of catalytic (C, solid lines) and gas‑phase (G, dotted lines) conversion rates of H2 and CO, and wall temperature (TW, dashed lines). H2/CO/air mixtures with four different H2/CO compositions. In all cases p = 1 bar, UIN = 20 m/s, and TIN = 600 K. The horizontal lines marked Tad denote the adiabatic equilibrium temperature. For clarity, the first 10 mm is shown in an expanded scale.

H2 and CO Conversion Rates (g/m2s)

4

CH2

3

CCO

TW

(a) 3.75% H2 4.00% CO

2 GCO

1 GH2

0 0

Tad

(b) 1.00% H2 6.75% CO

1410 Tad

TW C H2

1490

CCO

1330 1250

GCO

1170 5

5 15 30 45 60 75 0 Axial Distance x (mm)

Wall Temperature (K)

251

Catalytic Combustion of Syngas

15 30 45 60 75

Figure 8.19  Computed axial profiles of catalytic (C, solid lines) and gas‑phase (G, dotted lines) conversion rates of H2 and CO, and wall temperature (TW, dashed lines). H2/CO/air mixtures with two different H2/CO compositions, p = 10 bar, UIN = 2 m/s, and TIN = 600 K. The horizontal lines marked Tad provide the adiabatic equilibrium temperature. For clarity, the first 10 mm is shown in an expanded scale.

provided in Figure 8.19 for two H 2/CO compositions, TIN = 600 K and p = 10 bar. As evidenced in Figures 8.18 and 8.19, for hydrogen contents as low as 1% vol., superadiabatic surface temperatures are attained at the upstream sections of the reactor. At atmospheric pressure, the gaseous chemistry of both fuel components is noticeable only at the highest hydrogen concentration (Figure 8.18a). On the other hand, at elevated pressures the impact of the gaseous pathway extends to lower hydrogen contents (compare Figures 8.18b and c and 8.19a and b). The gas‑phase combustion of hydrogen at p = 10 bar moderates the surface temperatures along most of the reactor length, as seen in Figures 8.19a and 8.18b. The peak temperature remains relatively unaffected because the light‑off length is somewhat shorter in the high‑pressure cases, as manifested by the corresponding higher hydrogen catalytic conversion rates at x ≈ 0. In addition, the gas‑phase combustion of CO accelerates considerably with increasing pressure: its presence, however, does not affect the surface temperatures. For reactors designed to operate without excessive heat losses, the gaseous combustion of CO does not pose a thermal management concern. In addition, the homogeneous consumption of CO at high pressures may be desirable in accomplishing the conversion of this species at rates faster (and hence at shorter reactor lengths) than those dictated by heterogeneous transport limitations. Finally, the high diffusivity of hydrogen leads to a significant catalytic consumption of this species at much shorter distances than those required for CO (see Figure 8.18b–d). It is thus possible (e.g., by selecting an appropriate reactor length) to preferentially oxidize hydrogen and leave most of CO unconverted, in a way similar to the rich‑quick‑lean (RQL) concept discussed in Section 8.2. The volumetric substitution of H2 by CO lowers the surface temperatures (Figure 8.18), despite the fact that the molar exothermicity of CO is higher than that of hydrogen. Although the moderation of the surface temperatures by CO addition may be an advantage for steady reactor operation, it nonetheless impacts the dynamic

252

Synthesis Gas Combustion: Fundamentals and Applications

catalytic ignition characteristics, as discussed in the next section. At steady operation and for the high surface temperatures in Figures 8.18 and 8.19, the catalytic and gas‑phase pathways of CO and H2 are decoupled from each other as clarified in the foregoing section. Therefore, the practical measures to moderate the surface temperature are the same as those mentioned for hydrogen combustion, that is, increasing the channel hydraulic diameter or applying passive cooling with alternately coated channels. 8.6.4.3 Light-Off Temperatures Transient simulations are used to study the ignition characteristics of CO and H2 mixtures. Three H2/CO/air mixtures are considered, with a total volumetric H2 and CO content of 7.75%. Hydrogen comprises 0.5%, 1.0%, and 3.75% vol. of the mixture. Additional computations are carried out with H2*/CO, whereby the H2 has been replaced by a chemically inert fictitious species H2* that has the same thermodynamic and transport properties as H2 but does not participate in any reaction. Inlet temperatures of 620, 650, and 700 K and pressures of 1 and 10 bar are examined. For TIN = 650 K, ignition is achieved in all cases (H2 or H2*) and pressures, while for TIN = 620 K ignition is not possible, again for all cases. It is thus evident that the addition of even sizable amounts of hydrogen cannot lower the ignition temperatures of H2/CO mixtures to the corresponding values of pure hydrogen (360 to 380 K, as discussed in Section 8.6.2.2). In other words, the addition of CO clearly inhibits the catalytic ignition of hydrogen. This inhibition has its origin in the transition temperature of ~700 K, below which the effect of CO surface blocking commences. As a result, the H2/CO mixtures exhibit catalytic ignition characteristics in a manner similar to those of pure CO, irrespective of the amount of added hydrogen. In earlier steady‑state stagnation flow simulations over a platinum surface, using a surface reaction mechanism slightly modified from that of Table 8.1, Chao et al. (2003) concluded that the addition of 2.7% vol. H2 in 3.6% vol. CO marginally reduced the requirement for ignition inlet temperature by 14 to 19 K, depending on the strain rate. To study whether such results are consistent with the mechanism of Table 8.1, catalytic ignition delay times have initially been computed in a batch reactor. This approach also allows decoupling of pure kinetic effects from reactor parameters (heat loss mechanisms, properties of solid, etc.). The results are presented in Figure 8.20 for TIN = 700 K, p = 10 bar, 0.5% vol. H2 (or H2*), and 7.25% vol. CO. The catalytic ignition delay is longer in the CO/H2 than in the CO/H2* mixture (Figure 8.20a). Additional computations have shown that this result is irrespective of hydrogen content or pressure, clearly demonstrating that hydrogen inhibits the catalytic ignition of CO. The underlying reason is that the surface hydrogen, H(s), reduces the O(s) coverage, which is in turn needed for the CO(s) oxidation (see Figure 8.20b,c). The O(s) profile in Figure 8.20b actually points to a two‑stage ignition, first of H2 at t ≈ 5.5 s and then of CO at t ≈ 12 s. The former is a pseudoignition since H2 conversion starts already at t = 0 (Figure 8.20a). However, the sharp drop of H(s) and rise of O(s) at t ≈ 5.5 s, which are induced by the decreasing H2 and increasing temperature levels, are reminiscent of a typical hydrogen ignition (Deutschmann et al., 1996).

253

Catalytic Combustion of Syngas 1400

CO

(a)

0.06

CO

10

0.04

1200 1000

×H

2

T

0.02

Surface Coverage

100 10–1 10–2 10–3 10–4 10–5 10–6 10–7

800

τig

0.00 100 10–1 10–2 10–3 10–4 10–5 10–6 10–7

T

(b)

τig

Temperature (K)

Mole Fraction

0.08

600

CO(s) Pt(s) H(s)

O(s)

OH(s) H2O(s)

CO(s)

(c) Pt(s) O(s)

0

2

4

6 8 Time (ms)

10

12

14

Figure  8.20  Computed time histories in a batch reactor with p = 10 bar, TIN = 700 K, surface-to-volume ratio of 33.3 cm–1, and composition of 0.5% H2 and 7.25% CO vol. in air. (a) Major gas-phase species and temperature (black lines, H2; gray lines, chemically inert H2*), (b) major surface species coverage for H2 addition, (c) major surface species coverage for H2* addition. In (a), the ignition delay times are indicated by τig.

The aforementioned inhibition may appear contentious, but it is nonetheless very modest. It can potentially lead to a maximum increase of the preheat required to achieve ignition by 30 K (from 620 to 650 K) when adding hydrogen in CO. In conclusion, the addition of H2 in CO does not have a clear benefit for the catalytic ignition of CO. Depending on the employed catalytic reaction mechanism, it either aids the ignition of CO by lowering the preheat requirements by a meager 14 to 19 K or inhibits CO ignition by increasing the preheat by a few tens of degrees. In either case, this difference is small and does not seriously impact the reactor design. Detailed experiments are needed to resolve this apparent controversy.

254 (a)

1300 1200

1.6

1000

s 1.0

800

0.4 s

0.2 s

0.4 s

0.6 s

Steady state

0.2 s

0.2 s 0.2 s 0.2 s

0.4 s

0.6 s

0.4 s

0.8 s

0.8 s

0.4 s

(c)

0.8 s

100 10–1 10–2 10–3 10–4

0.6 s

0.8 s

Steady state

0.8 s

(d)

0

0.6 s 0.4 s 0.2 s

O(s)

s 0.8 0.6 s

(b)

10–1

10–1 10–2 10–3 10–4 10–5 10–6 10–7

tate)

3.0 s

s

900

10–2

CO(s)

steady s

2.0 s

1100

700 100

H(s)

4.5 s (~

0.6 s

Surface Temperature (K)

Synthesis Gas Combustion: Fundamentals and Applications

15

Steady state

30 45 Axial Distance x (mm)

60

75

Figure 8.21  Computed axial profiles at different times during light‑off of a 0.5% H2 and 7.25% CO vol. mixture in air, p = 10 bar, UIN = 2 m/s, TIN = 700 K: (a) wall temperatures, (b–d) surface species coverage. In (b–d) results are presented at the early phases (up to 0.8 s) and at steady state. The black lines refer to H2 content and the gray lines in (a–c) to chemically inert H2*.

Typical transient computations in the geometry of Figure  8.8 are shown in Figure 8.21 for syngas with 0.5% vol. H2 and TIN = 700 K. Axial profiles are provided (black lines, H2; gray lines, H2*) for the wall temperatures and selected surface species coverage, at various time intervals. The coverage is provided at early times (up to 0.8 s) and also at steady state. The wall heat‑up commences at the rear of the channel and then propagates upstream (Figure 8.21a); at the same time, the main surface coverage shifts from CO(s) to O(s) (Figure 8.21b and c). At t < 1.0 s, the propagation of the front is faster in the H2* than in the H2 dilution case (Figure 8.21a). For t > 2 s, however, H2 catalytic ignition is accomplished and the heat‑up of the solid is faster in the H2 dilution case due to the added heat generated from the hydrogen conversion. The total time required to reach steady state is roughly the same in both cases (~4.5 s; see Figure 8.21a). The inhibition due to hydrogen addition at the initial stages of Figure 8.21a follows much the same path described

Catalytic Combustion of Syngas

255

in Figure 8.20: H(s) is formed at early times at the front section of the reactor and upon hydrogen ignition it drops to the low steady‑state levels (Figure 8.21d). For TIN = 650 K, 3.75% H2 and 4.0% CO vol. content, p = 10 bar, and UIN = 2 m/s, the light‑off times increase. Ignition is still attained for both H2 and H2* dilutions and the steady states are reached at 13 and 9 s, respectively (Mantzaras, 2008). Again, during the initial phase of the light‑off event, the inhibition of the added hydrogen is strong. In practical systems, the relevant parameter that determines the minimum preheat requirements for ignition is mainly the initial phase of light‑off wherein hydrogen plays an inhibiting role. Also of interest in syngas combustion is the potential appearance of kinetically driven oscillatory phenomena (Imbihl and Ertl, 1995). Oscillatory behavior has been observed during the catalytic oxidation of both hydrogen and carbon monoxide over noble metals (Yamamoto et al., 1995; Yakhnin and Menzinger, 2002). It can be thus plausibly assumed that for certain operating regimes of power generation systems, such phenomena may also occur. This behavior is undesirable for practical burners and may require specific start‑up procedures and well‑defined operational envelopes in order to circumvent unstable combustion modes.

8.7 Conclusions Fuel‑lean and fuel‑rich catalytic combustion for syngas‑based and natural gas fuels have been reviewed with emphasis on applications for power generation systems. The adopted methodologies entail partial catalytic fuel conversion with consumption of the remaining fuel in a follow‑up homogeneous combustion zone. Reactor thermal management issues have been outlined and the chemical/transport mechanisms controlling the surface temperatures of practical reactors have been identified. It was shown that in both fuel‑lean and fuel‑rich combustion of hydrogen‑rich syngas fuels, the diffusional imbalance of hydrogen greatly impacts the attained surface temperatures. Suitable catalysts and reactor structures for the combustion of various syngas fuels have been presented. Fundamental studies of the hetero‑/homogeneous combustion of fuel‑lean H2 and CO over Pt were presented. Two‑dimensional steady and transient simulations were carried out for syngas compositions with varying H2 and CO contents (including pure H2 and CO fuels), pressures in the range of 1 to 15 bar, and linear velocities relevant to power generation applications. It was shown that despite the large geometrical confinements typical of honeycomb catalytic reactors, the homogeneous combustion of both H2 and CO could not be neglected, particularly at elevated pressures and temperatures. Above a transition temperature of about 700 K, which is roughly independent of pressure and syngas composition, there is no chemistry coupling between the heterogeneous pathways of CO and H2. Moreover, at sufficiently high (but still acceptable for catalytic operation) temperatures (T > 1150 K) and for pressures p > 10 bar, the gaseous reaction pathways of both CO and H2 are important, with the former crucially dependent on the radical pool provided by the latter. At those temperatures, the chemical coupling between the heterogeneous pathway of one fuel component and the homogeneous pathway of the other is minimal.

256

Synthesis Gas Combustion: Fundamentals and Applications

In syngas catalytic combustion, the diffusional imbalance of hydrogen can lead (depending on the hydrogen content) to an excessively large superadiabatic surface temperature, which may endanger the catalyst integrity and cause reactor meltdown. The diffusional imbalance of hydrogen also confines its gaseous combustion to regimes near the hot catalytic wall. The presence of gaseous combustion moderates the superadiabatic wall temperature by shielding the catalyst from the ­hydrogen‑rich channel core. Strategies for reactor thermal management are presented, which include reactors with smaller geometrical confinements (larger channel radii) so as to promote homogeneous combustion of hydrogen at the expense of catalytic combustion. Other appropriate thermal management strategies include combustion in alternately coated monolithic reactors. Below the transition temperature of ~700 K, the chemical coupling between the CO and H2 catalytic pathways is strong. The catalyst is predominantly covered by CO that, in turn, inhibits the catalytic conversion of both fuel components. The catalytic ignition temperatures of H2/air and CO/air fuels are 360 to 380 K and 650 to 700 K, respectively, over a range of reactor and flow parameters relevant for power generation applications. While the addition of CO in H2 clearly inhibits the catalytic ignition of the latter, there is no clear improvement in the ignition characteristics of CO by adding H2 due to the dominant CO surface blocking at temperatures below 700 K. Depending on the catalytic chemical reaction scheme employed, the catalytic ignition temperatures for CO/H2/air mixtures can either drop (compared to those of CO/air mixtures) by a marginal 14 to 19 K or increase by a few tens of degrees. On the other hand, in syngas combustion, the nearly neutral transport properties of CO moderate the superadiabatic surface temperatures, thus simplifying the reactor design and its thermal management.

Acknowledgments Support was provided by the Swiss Federal Office of Energy (BFE), the Swiss Commission of Technology and Innovation (KTI) under Contract 8457.2, ALSTOM Power of Switzerland, and CompactGTL.

References Appel, C., Mantzaras, J., Schaeren, R., Bombach, R., and Inauen, A. (2005b). Turbulent catalytically stabilized combustion of hydrogen/air mixtures in entry channel flows. Combust. Flame 140:70. Appel, C., Mantzaras, J., Schaeren, R., Bombach, R., Inauen, A., Kaeppeli, B., Hemmerling, B., and Stampanoni, A. (2002). An experimental and numerical investigation of homogeneous ignition in catalytically stabilized combustion of hydrogen/air mixtures over platinum. Combust. Flame 128:340. Appel, C., Mantzaras, J., Schaeren, R., Bombach, R., Inauen, A., Tylli, N., Wolf, M., Griffin, T., Winkler, D., and Carroni, R. (2005a). Partial catalytic oxidation of methane to synthesis gas over rhodium: In situ Raman experiments and detailed simulations. Proc. Combust. Instit. 30:2509.

Catalytic Combustion of Syngas

257

Arnby, K., Toerncrona, A., Andersson, B., and Skoglundh, M. (2004). Investigation of Pt/γ‑Al2O3 catalysts with locally high Pt concentrations for oxidation of CO at low temperatures. J. Catal. 221:252. Beebe, K. W., Cairns, K. D., Pareek, V. K., Nickolas, S. G., Schlatter, J. C., and Tsuchiya, T. (2000). Development of catalytic combustion technology for single-digit emissions from industrial gas turbines. Catal. Today 59:95. Berg, M., Johansson, E. M., and Jaras, S. G. (2000). Catalytic combustion of low heating value gas mixtures: Comparison between laboratory and pilot scale tests. Catal. Today 59:117. Boehman, A. L., and Dibble, R. W. (2000). Experimental and numerical investigation on the influence of temporal fuel/air unmixedness on NOX emissions of lean premixed catalytically stabilized and non-catalytic combustion. Catal. Today 59:131. Bui, P. A., Vlachos, D. G., and Westmoreland, P. R. (1996). Homogeneous ignition of hydrogen/air mixtures over platinum. Proc. Combust. Instit. 26:1763. Burch, R., and Southward, B. W. L. (2000). Clean catalytic combustion of nitrogen-bearing gasified biomass. Chem. Commun. 8:703. Carroni, R., Griffin, T., Mantzaras, J., and Reinke, M. (2003). High-pressure experiments and modeling of methane/air catalytic combustion for power generation applications. Catal. Today 83:157. Carroni, R., Schmidt, V., and Griffin, T. (2002). Catalytic combustion for power generation. Catal. Today 75:287. Chao, Y.-C., Chen, G.-B., and Hsu, H.-W. (2003). Catalytic ignition of multifuels on platinum: Effect of strain rate. Catal. Today 83:97. Chao, Y. C., Chen, G. B., Hsu, H. W., and Hsu, J. R. (2004). Catalytic combustion of gasified biomass in a platinum monolith honeycomb reactor. Appl. Catal. A Gen. 261:99. Dalla Betta, R. A., Schlatter, J. C., Nickolas, S. G., Lodewyk, A., Shojii, T., and Sasaka, M. (1993). New catalytic combustion technology for very low emission gas turbines. Paper presented at the Proceedings of the EPRI/EPA Conference on Low NOX Combustion, Miami, FL, May. Deutschmann, O., Maier, L. I., Riedel, U., Stroemman, A. H., and Dibble, R. W. (2000). Hydrogen assisted catalytic combustion of methane on platinum. Catal. Today 59:141. Deutschmann, O., Schmidt, R., Behrendt, F., and Warnatz, J. (1996). Numerical modeling of catalytic ignition. Proc. Combust. Instit. 26:1747. Dogwiler, U., Benz, P., and Mantzaras, J. (1999). Two-dimensional modelling for catalytically stabilized combustion of a lean methane-air mixture with elementary homogeneous and heterogeneous chemical reactions. Combust. Flame 116:243. Eguchi, K., and Arai, H. (1996). Recent advances in high temperature catalytic combustion. Catal. Today 29:379. Eriksson, S., Wolf, M., Schneider, A., Mantzaras, J., Raimondi, F., Boutonnet, M., and Jaras, S. (2006). Fuel rich catalytic combustion of methane in zero emissions power generation processes. Catal. Today 117:447. Ersson, A. G., Persson, K., Adu, I. K., and Jaras, S. G. (2006). A comparison between hexa­ aluminates and perovskites for catalytic combustion applications. Catal. Today 112:157. Etemad, S., Smith, L. L., and Burns, K. (2004). System study of rich catalytic/lean burn (RCL) catalytic combustion for natural gas and coal-derived syngas combustion turbines. DOE Final Report DE-FG26-02NT41521, Precision Combustion. Farrauto, R. J., Hobson, M. C., Kennelly, T., and Waterman, E. M. (1992). Catalytic chemistry of supported palladium for combustion of methane. Appl. Catal. A Gen. 81:227. Furuya, T., Hayata, T., Yamanaka, S., Koezuka, J., Yoshine, T., and Ohkoshi, A. (1987). Hybrid catalytic combustion for stationary gas turbine: Concept and small scale test results. ASME Paper 87-GT-99. Glassman, I. (1996). Combustion. London: Academic Press.

258

Synthesis Gas Combustion: Fundamentals and Applications

Griffin, T., Winkler, D., Wolf, M., Appel, C., and Mantzaras, J. (2004). Staged catalytic combustion method for the advanced zero emissions gas turbine power plant. ASME 2004-54101. Groppi, G., Belloli, A., Tronconi, E., and Forzatti, P. (1996). Catalytic combustion of CO‑H2 mixtures on manganese-substituted hexaaluminates. Catal. Today 29:403. Hickman, D. A., and Schmidt, L. D. (1993). Steps in CH4 oxidation on Pt and Rh surfaces: High-temperature reactor simulations. AIChE 39:1164. Imbihl, R., and Ertl, G. (1995). Oscillatory kinetics in heterogeneous catalysis. Chem. Rev. 95:697. Johansson, E. M., Danielsson, K. M. J., Ersson, A. G., and Jaras, S. G. (2002). Development of hexaaluminate catalysts for combustion of gasified biomass in gas turbines. J. Eng. Gas Turbines Power 124:235. Johansson, E. M., Danielsson, K. M. J., Pocoroba, E., Haralson, E. D., and Jaras, S. G. (1999). Catalytic combustion of gasified biomass over hexaaluminate catalysts: Influence of palladium loading and ageing. Appl. Catal. A Gen. 182:199. Kajita, S., Tanaka, Y., and Kitajima, J. (1990). Evaluation of a catalytic combustor in a gas turbine generator unit. ASME Paper 90-GT-89. Karim, H., Lyle, K., Etemad, S., Smith, L., Pfefferle, W. C., Dutta, P., and Smith, K. (2002). Advanced catalytic pilot for low NOX industrial gas turbines. Paper presented at Proceedings of the ASME Turbo Expo, Amsterdam, June 3–6. Krill, W., and Kesselring, J. (1978). The development of catalytic combustors for stationary source applications. Paper presented at the 3rd Workshop on Catalytic Combustion, Asheville, NC, October. Kusar, H. M. J., Ersson, A. G., and Jaras, S. G. (2003). Catalytic combustion of gasified refusederived fuel. Appl. Catal. B Environ. 45:1. Kusar, H. M. J., Ersson, A. G., and Vosecky, M. (2005). Selective catalytic oxidation of NH3 to N2 for catalytic combustion of low heating value gas under lean/rich conditions. Appl. Catal. B Environ. 1/2:25. Lampert, J. K., Kazi, M. S., and Farrauto, R. J. (1997). Palladium catalyst performance for methane emissions abatement from lean burn natural gas vehicles. Appl. Catal. B Environ. 14:211. Li, J., Zhao, Z., Kazakov, A., and Dryer, F. L. (2003). An updated comprehensive kinetic model for H2 combustion. Paper presented at the Fall Technical Meeting of the Eastern States Section of the Combustion Institute, Pennsylvania State University, October 26–29. Lietti, L., Cristiani, C., Groppi, G., and Forzatti, P. (2000). Preparation, characterization and reactivity of Me- hexaaluminate (Me=Mn, Co, Fe, Ni, Cr) catalysts in the catalytic combustion of NH3-containing gasified biomasses. Catal. Today 59:191. Lyubovsky, M., Smith, L. L., Castaldi, M., Karim, H., Nentwick, B., Etemad, S., Lapierre, R., and Pfefferle, W. C. (2003). Catalytic combustion over platinum group catalysts: Fuel‑lean versus fuel‑rich operation. Catal. Today 83:71. Mantzaras, J. (2006). Interplay of transport and hetero-/homogeneous chemistry. In Focus on combustion research, ed. S. Z. Jiang, 241. New York: Nova Publishers. Mantzaras, J. (2008). Catalytic combustion of syngas. Combust. Sci. Technol. 180:1137. Mantzaras, J., and Benz, P. (1999). An asymptotic and numerical investigation of homogeneous ignition in catalytically stabilized channel flow combustion. Combust. Flame 119:455. Mantzaras, J., Bombach, R., and Schaeren, R. (2008). Hetero-/homogeneous combustion of hydrogen/air mixtures over platinum at pressures of up to 10 bar. Proc. Combust. Instit. 32:1937. McCarty, J. G., and Wise, H. (1990). Perovskite catalysts for methane combustion. Catal. Today 8:231. Mhadeshwar, A. B., and Vlachos, D. G. (2005). A thermodynamically consistent surface reaction mechanism for CO oxidation on Pt. Combust. Flame 142:289.

Catalytic Combustion of Syngas

259

ONSE (Onsite Energy Corporation). (1999). Cost analysis of NOX control alternatives for stationary gas turbines. Pfefferle, W. C. (1974). Catalytically supported thermal combustion. Bel. Patent 814752. Pfefferle, W. C., and Pfefferle, L. D. (1986). Catalytically stabilized combustion. Prog. Energy Combust. Sci. 12:25. Pocoroba, E., Johansson, E. M., and Jaras, S. G. (2000). Ageing of palladium, platinum and manganese-based combustion catalysts for biogas applications. Catal. Today 59:179. Reinke, M., Mantzaras, J., Bombach, R., Schenker, S., and Inauen, A. (2005). Gas phase chemistry in catalytic combustion of methane/air mixtures over platinum at pressures of 1 bar to 16 bar. Combust. Flame 141:448. Reinke, M., Mantzaras, J., Schaeren, R., Bombach, R., Inauen, A., and Schenker, S. (2004). High-pressure catalytic combustion of methane over platinum: In situ experiments and detailed numerical predictions. Combust. Flame 136:217. Richards, G. A., Mcmillian, M. M., Gemmen, R. S., Rogers, W. A., and Cully, S. R. (2001). Issues for low-emission, fuel-flexible power systems. Prog. Energy Combust. Sci. 27:141. Schlegel, A., Benz, P., Griffin, T., Weisenstein, W., and Bockhorn, H. (1996). Catalytic stabilization of lean premixed combustion: Method for improving NOX emissions. Combust. Flame 105:332. Schlegel, A., Buser, S., Benz, P., Bockhorn, H., and Mauss, F. (1994). NOX formation in lean premixed non-catalytic and catalytically stabilized combustion of propane. Proc. Combust. Inst. 25:1019. Schmitz, I., Paul, T., Wirtz, S., Scherer, V., and Carroni, R. (2005). Lay-out of the post‑catalyst section of a novel gas turbine burner. Paper presented at the Proceedings of the 6th International Workshop on Catalytic Combustion, Ischia, Italy, September 11–14. Schneider, A., Mantzaras, J., and Eriksson, S. (2008). Ignition and extinction in catalytic partial oxidation of methane-oxygen mixtures with large H2O and CO2 dilution. Combust. Sci. Technol. 180:89. Schneider, A., Mantzaras, J., and Jansohn, P. (2006). Experimental and numerical investigation of the catalytic partial oxidation of CH4/O2 mixtures diluted with H2O and CO2 in a short contact time reactor. Chem. Eng. Sci. 61:4634. Shiau, C.-Y., Ma, M. W., and Chuang, C. S. (2006). CO oxidation over CeO2-promoted Cu/γ‑Al2O3 catalyst: Effect of preparation method. Appl. Catal. A Gen. 301:89. Smith, L. L., Karim, H., Castaldi, M. J., Etemad, S., and Pfefferle, W. C. (2006). Rich-catalytic leanburn combustion for fuel-flexible operation with ultra-low emissions. Catal. Today 117:438. Smith, L. L., Karim, H., Castaldi, M. J., Etemad, S., Pfefferle, W. C., Khanna, V. K., and Smith, K. O. (2005). Rich-catalytic lean-burn combustion for low-single digit NOX gas turbines. J. Eng. Gas Turbines Power 127:27. Tham, Y. F., and Chen, J. Y. (2005). Numerical study of rich-quick-lean (RQL) combustion of syngas. Paper presented at the 4th Joint Meeting of the U.S. Sections of the Combustion Institute, Philadelphia, March 20–23. Thevenin, P. O., Ersson, A. G., Kusar, H. M. J., Menon, P. G., and Jaras, S. G. (2001). Deactivation of high temperature combustion catalysts. Appl. Catal. A Gen. 212:189. Veser, G., and Frauhammer, J. (2000). Modelling steady state and ignition during catalytic methane oxidation in a monolith reactor. Chem. Eng. Sci. 55:2271. Veser, G., Ziauddin, M., and Schmidt, L. D. (1999). Ignition in alkane oxidation on noble‑metal catalysts. Catal. Today 47:219. Warnatz, J. (2005). Personal communication. Warnatz, J., Dibble, R. W., and Maas, U. (1996). Combustion, physical and chemical fundamentals, modeling and simulation. New York: Springer-Verlag, p. 69. Williams, F. A. (1985). Combustion theory. Menlo Park, CA: Benjamin/Cummings, p. 48. Witton, J. J., Noordally, E., and Przybylski, J. M. (2003). Clean catalytic combustion of low heat value fuels from gasification processes. Chem. Eng. J. 91:115.

260

Synthesis Gas Combustion: Fundamentals and Applications

Yakhnin, V., and Menzinger, M. (2002). Stationary and travelling hot spots in the catalytic combustion of hydrogen in monoliths. Chem. Eng. Sci. 57:4559. Yamamoto, S. Y., Surko, C. M., Maple, M. B., and Pina, R. K. (1995). Spatio-temporal dynamics of oscillatory heterogeneous catalysis: CO oxidation on platinum. J. Chem. Phys. 102:8614. Yee, D. Y., Lundberg, K., and Weakley, C. K. (2001). Field demonstration of a 1.5 MW industrial gas turbine with a low emissions catalytic combustion system. J. Eng. Gas Turbines Power 123:550. Zwinkels, M. F. M., Jaras, S. G., Menon, P. G., and Griffin, T. A. (1993). Catalytic materials for high‑temperature combustion. Catal. Rev. Sci. Eng. 35:319.

Issues 9 Operability Associated with Steady Flowing Combustors Tim Lieuwen, Vincent McDonell, Domenic Santavicca, and Thomas Sattelmayer Contents 9.1 Introduction................................................................................................... 261 9.2 Blowout.......................................................................................................... 262 9.3 Flashback....................................................................................................... 268 9.3.1 Turbulent Flame Propagation in the Core Flow................................ 268 9.3.2 Combustion Pulsation-Induced Flashback........................................ 270 9.3.3 Flashback in the Boundary Layer...................................................... 270 9.3.4 Vortex Breakdown-Driven Flame Propagation in the Core of Swirling Flows................................................................................... 271 9.4 Combustion Instability.................................................................................. 273 9.5 Autoignition................................................................................................... 278 9.6 Conclusions....................................................................................................284 References............................................................................................................... 285

9.1 Introduction A number of important practical problems must be dealt with in developing a system capable of combusting syngas, particularly if the system must also emit low levels of CO and NOX emissions (Richards et al., 2001). This chapter focuses upon combustor operability issues, associated with having the combustor reliably hold the flame so that it neither flashes back nor blows out, and burns the fuel in a quiet, steady fashion. These operability issues generally involve complex, poorly understood inter­actions between swirling flow dynamics, flow field alterations induced by volumetric expansion across the flame, and flame propagation. The objective of this chapter is to review understanding of the manner in which syngas fuel composition influences these operability issues in steady flowing combustors, such as gas turbines, boilers, and furnaces. The four most critical of these operability issues, all of which are strongly influenced by fuel properties, are blowout, flashback, combustion instability, and autoignition.

261

262

Synthesis Gas Combustion: Fundamentals and Applications

Blowout refers to situations where the flame becomes detached from the location where it is anchored and is physically blown out of the combustor. Blowout is often referred to as the static stability limit of the combustor. Blowout involves the inter­ actions between the reaction and propagation rates of strained flames in a high-shear flow. Blowout events can require a lengthy and often expensive system shutdown, purge cycle, and restart. A second issue is flashback, where the flame propagates upstream of the combustor and into premixing passages that are not designed for high temperatures. Flashback involves turbulent flame speed propagation in a highly inhomogeneous, swirling flow. Flashback is a serious safety risk because of overheat and subsequent failure of nozzle components. Combustion instability refers to damaging pressure oscillations associated with fluctuations in the combustion heat release rate. These oscillations cause wear and damage to combustor components and, in extreme cases, can cause liberation of pieces into the hot gas path, damaging downstream turbine components. Autoignition refers to the ignition of the reactive mixture upstream of the combustion chamber. Similar to flashback, it results in chemical reactions and hot gases in premixing sections, but its physical origins are quite different from those of flashback. Rather than the flame propagating upstream into the premixing section, autoignition involves spontaneous ignition of the mixture in the premixing section. Understanding these operability issues requires understanding of more fundamental combustion properties. The objective of this chapter is to compile known results and discuss their implications on each of these operability issues. This is intended to provide an overview of the underlying processes that must be considered when evaluating how a given combustor’s operability will be affected with syngas fuel.

9.2 Blowout Developing physics-based correlations of blowout behavior is complicated by lack of understanding of the detailed phenomenology of the blowout process, such as the dynamics of near-blowout flames or the flame characteristics at the stabilization point (Durbin and Ballal, 1996). For example, there is disagreement on whether premixed flames in high turbulent intensity gas turbine environments have flamelet, “thickened” flamelet, or well-stirred reactor (WSR) type properties. This has implications on blowout modeling because the appropriate physical model clearly changes depending on whether the reaction zone exhibits flame sheet or volumetric characteristics. Several different theories or physical considerations have been used in past blowout correlation studies. For example, Longwell et al. (1953) suggested that blowout occurs when it is not possible to balance the rate of entrainment of reactants into the recirculation zone, viewed as a well-stirred reactor regime, and the rate of burning of these gases. A similar idea relates to an energy balance between heat supplied by the hot recirculating flow to the fresh gases and that released by reaction (Williams, 1966; Kundu et al., 1977). For bluff body, shear stabilized flames, Zukoski (1997) proposed that the contact time between the combustible mixture and hot gases in the shear layer must exceed a chemical ignition time. Finally, several studies have

Operability Issues Associated with Steady Flowing Combustors

263

proposed a flamelet-based description based upon local extinction by excessive flame stretch, controlled by the relative values of the straining of the flame in the shear layers and the extinction strain rate (Yamaguchi et al., 1985; Shanbogue et al., 2008). As noted by Glassman (1996) and Shanbogue et al. (2008), these lead to similar correlations that relate the blowout limits to a Damköhler number, that is, ratio of a residence and chemical kinetic time, τres/τchem.

Da =



τ res τ chem

(9.1)

However, it is difficult to determine which of the above descriptions most accurately describes the controlling processes based upon analyses of Da correlations alone, because the different velocity, length, and chemical time scales generally lead to comparable groupings of the data. Nonetheless, it is clear that the chemical kinetic time scale, τchem, plays a very significant role in controlling blowout limits. Figure 9.1 plots the dependence of a calculated flame propagation chemical time, defined as τchem = α/SL2, upon the H2/CO/CH4 ratio, where α and SL denote the thermal diffusivity and laminar flame speed, respectively. Each point in the composition space corresponds to a fixed adiabatic flame temperature of 1500 K; that is, the mixture stoichiometry is adjusted for each composition such that the mixture has the given temperature. Note the order of magnitude variation in chemical time from the fast H2 mixtures to the slower CO mixtures. The above observation is very consistent with experimental findings that the key parameter that influences the blowout/extinction characteristics of syngas is the percentage of hydrogen. Numerous studies have shown that the fuel/air ratio at which blowout/extinction occurs monotonically decreases as the percentage of hydrogen in the fuel increases, whether it is CH4/H2 mixtures (Schefer, 2003), CO/H2 mixtures H2 0.5

1.5

5 CO

2.5

2

2.5 3 CH4

Figure  9.1  Dependence of chemical time (ms), calculated using GRI-Mech 3.0 as the kinetic mechanism, upon fuel composition at fixed adiabatic flame temperature, 1500 K at 1.7 atm with 300 K reactants temperature.

264

Synthesis Gas Combustion: Fundamentals and Applications 0.40 0.35

at LBO

0.30 0.25 0.20 0.15 0.10 0.05 0

20

40

% H2

60

80

100

Figure 9.2  Measured dependence of equivalence ratio of CH4/H 2/CO mixtures at LBO upon H 2 mole fraction in a premixed, swirling combustor (nozzle exit velocity of 59 m/s, reactants temperature 458 K, and combustor pressure 4.4 atm.

(Vagelopoulos and Egolfopoulos, 1994), or other hydrogen-blended fuels. For example, the data in Figure 9.2 was obtained from Zhang et al. (2005) and plots the dependence of the fuel/air ratio at blowout of H2/CH4/CO mixtures upon the percentage of H2 in the fuel. These plots show the well-known result that, in general, mixtures can be stabilized with lower equivalence ratios as the H2 concentration increases. While this graph focuses on fuel/air ratio at blowout, these data can be replotted to illustrate similar trends for adiabatic flame temperature or laminar flame speed, whose blowout values also monotonically decrease with fuel/air ratio at blowout. While clearly there are important issues such as appropriate choice of length and velocity scale, Damköhler number scalings have been found to capture blowout trends across a wide range of fuel compositions up to about 50% H2, as illustrated by Figure 9.3. These data consist of a number of permutations of CO/H2/CH4 fuel blends, ranging from pure fuels to various combinations. The Damköhler number is defined as

Da =

τ res τ Blowoff

=

D /U b τ Blowoff

(9.2)

where D and Ub denote the width of the combustor and the burned gas flow speed, respectively. The chemical time scale τBlowoff equals the calculated residence time at blowout of a well-stirred reactor model (using the GRI-Mech 3.0 mechanism), which correlates well with the time scale used in Figure 9.1 for hydrogen levels less than about 50% (Zhang et al., 2005). While scatter is present in the data, the results show that blowout occurs at a roughly constant value of the Damköhler number. However, as noted by Noble et al. (2006), the blowout Damköhler number based upon a PSR calculation changes by

265

Operability Issues Associated with Steady Flowing Combustors

DaB = (D/Ub)/tBlowoff

102

101

100

10–1

10–2

0

10

20

% H2

30

40

50

Figure  9.3  Ratios between residence time and chemical time in a premixed, swirling combustor at constant nozzle exit velocity of 59 m/s, combustor pressure of 1.7 atm, and 300 K reactant temperature.

four orders of magnitude at blowout with 50 to 100% H2. Noble et al. (2006) suggested that this variation might be a manifestation of thermal diffusive effects, with the result that the fuel/air ratio, and therefore the chemical time, with which to characterize the mixture, ϕmod , was not the globally averaged fuel/air ratio, ϕave , but a modified fuel/air ratio, ϕmod = ϕave + f (Df /Dox ), where Df and Dox denote the fuel and oxidizer diffusivity. Data from Zhang (2008) support this idea. They note substantial variations in chemical time relationships for CH4/H2 mixtures, with H2 levels greater than about 20%, particularly between extinction strain rate–based time scales and unstrained flame or PSR-based time scales. Note that the latter is purely a kinetic time scale, while the former two time scales also depend upon diffusive processes. Given the substantial difference in diffusivity of the fuel relative to oxidizer with increasing H2 levels, it then follows that appropriate choice of time scale becomes increasingly important with high-H2 fuels. To restate, systematic differences between different kinetic time scales can be anticipated when comparing over a range of fuels with different diffusivities. Continuing this point, Zhang (2008) argued that blowout correlated best with an extinction strain–based time scale. Figure 9.4 compares three calculated chemical times for near-blowout CH4/H2 flames. In this experiment, nozzle velocity and geometry were fixed, so it is reasonable to presume relatively constant fluid mechanic time scale. As such, the y-axis on this graph is inversely proportional to the Damköhler number. These data show that the extinction-based kinetic scale is roughly constant across the whole range of H2 levels at blow-off, while the PSR- and flame propagation–based scales are not. However, such efforts to correlate blowout limits with a single, constant time scale pass over key physics. For example, the residence time parameter, assumed

266

Synthesis Gas Combustion: Fundamentals and Applications 100 α/SL2 τBlowoff

Time Scale (s)

10–1

1/κext

10–2 10–3 10–4

0

20

40 % H2

60

80

Figure 9.4  Comparison of three chemical time scales for near-blowout H2/CH4 flames in a premixed, swirling flame.

constant above, would certainly be expected to change somewhat as the underlying fluid mechanics of these flames change due to the variation in burned gas temperature, and therefore burned gas flow velocity and Reynolds number. Furthermore, observations by Zhang et al. (2005) suggest that the physical mechanisms of blowout change with hydrogen levels. For mixtures with H2 levels below about 50% by ­volume, the blowout event occurs abruptly with a small change in fuel composition, although sometimes preceded by slight liftoff of the flame from the burner. However, for high-H2 mixtures, the blowout and liftoff events were quite distinct. Usually, the flame became visibly weaker, lifted off from the holder, and moved progressively downstream with decreases in equivalence ratio before blowing out for good. In addition, studies by Muruganandam et al. (2005) and Nair and Lieuwen (2005) have shown that flames do not generally blow out in a completely discontinuous manner. Rather, as blowout is approached, the flame becomes increasingly unsteady, lifts off the burner, and moves downstream. These observations were quantified by acoustic or optical measurements of the chemiluminescence/sound radiated by the flame, which show increasingly large fluctuations, characterized by time-localized events in the signal as the blowout boundary is approached. These fluctuations in chemiluminescence apparently are associated with axial fluctuations in the leading edge of the flame and localized extinction in the flame, in which holes in the flame sheet occur in response to high localized stretch. The unburned fuel passes through the hole and either passes out of the combustor or is burned in an abrupt reignition event downstream. This alters the local fluid mechanics, which in turn influences the stretch rate that the flame is subjected to at a later instant of time. Measurements of Zhang et al. (2007b) have shed some light on the underlying mechanisms for this unsteadiness in a CH4/H2-fueled swirl combustor. In their facility, the flame was nominally attached to an annular center body under stable conditions (see Figure 9.5). Much of the near-blowout dynamics were associated with the flame periodically

267

Operability Issues Associated with Steady Flowing Combustors 16 12

60 Axial Location (mm)

Axial Location (mm)

60

40

8 4

40

20

0 –4

20

–8 –12

0

0

(a)

–20 0 20 Radial Location (mm)

–16

(b)

30 m/s 60 Axial Location (mm)

Axial Location (mm)

60

40

20

0

–20

0

20

Radial Location (mm)

(c)

40

20

0

–20 0 20 Radial Location (mm)

(d)

Figure  9.5  Snapshots of instantaneous flow field and flame front for 50%CH4/50% H2 premixed, swirling flame near blowout; bottom right schematic shows interrogation window.

detaching from the center body and convecting downstream. This process repeated itself in a chaotic fashion—images showing the flow field and flame at four instances of time are shown in Figure 9.5. In a nonreacting swirling flow such as this one, it is known that the flow field is characterized by complicated, three-­dimensional helical structures and precessing vortices (Syred, 2006). Note that when the flame is firmly attached, the downstream flow field exhibits essentially none of these features. When the flame is detached from the center body, these flow instabilities develop, as clearly shown in the flow field. In this case, the flame is buffeted by a much more complex, vortical flow field. Presumably, this detachment of the flame from the center body is due to the high shear at this location, which, below some equivalence ratio, is larger than the extinction strain rate that the flame can withstand. Higher-hydrogen flames, being able to withstand a higher strain rate than CH4 at the same flame temperature (Sankaran and Im, 2003; Ren et al., 2001), could remain attached to the center body for a broader range of fuel/air ratios; this partially explains why they can persist at lower fuel/air ratios. However, due to vortex breakdown, an alternative flame stabilization point also exists downstream. For reasons that were not understood, the

268

Synthesis Gas Combustion: Fundamentals and Applications

low-H2 flames did not persist in this downstream region, whereas the much higherH2 flames could. Shanbogue et al. (2008) hypothesized a relationship between these observations of highly dynamic, near-blowout flames and steady-state, single time scale Damköhler number correlations for bluff body flames. First, they note that, as described above, blowout is preceded by events associated with (spatio) temporally localized extinction that occurs sporadically on near-blowout flames. They emphasize that these extinction events are distinct from blowout—in fact, under certain conditions the flame can persist indefinitely with certain levels of local extinction, consistent with those made by Zhang (2008) in swirl flames. They propose that Damköhler number correlations contain the essential physics describing this “initial” stage of blowout; that is, they are correlations for the conditions where local extinction on the flame begins, but do not fundamentally describe the ultimate blowout condition itself. However, such correlations are reasonably successful in correlating blowout limits because the ultimate blowout event is related to the onset of these local extinction events that precede blowout.

9.3 Flashback Flashback occurs when the turbulent displacement speed exceeds the flow velocity along some streamline, allowing the flame to propagate upstream into the premixing section. While flashback is a classical topic that has been extensively investigated, the complexity of the topic increases substantially in swirling flows. In particular, experimental investigations have revealed four different flashback mechanisms, which may lead to upstream flame propagation, depending on the specific burner design and operating point: turbulent flame propagation in the core flow, flashback due to combustion instabilities, flashback in the boundary layer, and flashback in the core flow due to alteration of vortex breakdown dynamics (Kröner et al., 2003; Fritz et al., 2004; Kiesewetter et al., 2003; Thibaut and Candel, 1998). The first three types can occur in swirling as well as nonswirling premix burners, whereas the fourth mechanism requires a swirling flow in the mixing zone. Importantly, fuel composition effects influence these mechanisms very differently. Each mechanism is considered separately below.

9.3.1 Turbulent Flame Propagation in the Core Flow In a well-designed burner, flame propagation into the burner is prevented by high axial flow velocities. In principle, the flame will be able to propagate upstream in all zones of the burner with flow velocities below the turbulent burning velocity of the mixture. Flashback occurs when the turbulent flame speed exceeds the flow velocity along some streamline, allowing the flame to propagate into the premixing section (Plee and Mellor, 1978). This fact leads to the simple design rule that the flow field must not have local velocity deficits and that the axial flow velocity must be substantially above the turbulent flame speed, ST . Given the typical combustor pressure drop of ∆ptot = 2 to 3%, absolute velocities are on the order of uabs ≈ 90 to 120 m/s in the burner mixing zone. In the case of strong swirl, the axial velocity component

Operability Issues Associated with Steady Flowing Combustors

269

drops to approximately uax ≈ 65 to 85 m/s for the same burner pressure drop. The average turbulence level depends mainly on the swirl number in the mixing zone, although there is also some influence of the specific burner design; for example, high swirl designs often used in natural gas–fueled systems have turbulence levels up to 10 to 15%. Hence, typical velocity fluctuations in premix burners are u′/uax­ ≈ 14 to 22% or u′ ≈ 9 to 18 m/s. Assuming that the flow field does not have local wakes with low axial velocity and with excessive turbulence, this leads to the important conclusion that flame propagation in the turbulent burner core flow can only occur if the turbulent burning velocity is substantially higher than the characteristic turbulent velocity fluctuation. For the worst case of maximum swirl, the criterion for flame propagation is then ST /u′ ≈ 4.5 to 7, and the reduction of the swirl number leads to even higher values. Laminar flame speeds of high-hydrogen syngas fuels are substantially higher than those of natural gas, and decrease with pressure. For engines without recuperation, pressure ratios above 15, and adiabatic flame temperatures of Tad < 1900 K, the laminar flame speed of syngas does not exceed SL = 2 m/s, whereas recuperated engines with low pressure ratios of approximately 4 may reach SL = 4 m/s. It has been shown that the turbulent flame speed, ST, exhibits a dependency upon fuel composition (Lipatnikov and Chomiak, 2005). For this reason, calculation of ST by extrapolating data from other fuels that have similar laminar flame speeds, SL , and from experiments with similar turbulence intensities, u′/SL , is not applicable. For example, Kido et al. (2002) measured the turbulent flame speed for a variety of H2, CH4, and C3H8 mixtures with nominally the same laminar flame speeds, but found wide variations in ST that approached a factor of 10. The reason for these fuel effects is uncertain. Some workers have suggested that that they can be correlated with thermodiffusive effects. For example, differences in the relative rates of mass diffusion of the deficient species or thermal diffusion affect the local laminar flame speed and the tendency of the flame to become spontaneously wrinkled, even in the absence of turbulent fluctuations. If differential diffusion processes are significant, then this could be expected to be very significant in syngas fuels, because of the large differences in diffusivity of the various fuel and oxidizer components. We refer the reader to the comprehensive review on this subject by Lipatnikov and Chomiak (2005) for more discussion of this topic. Worst-case estimates can probably be provided without considering preferential diffusion, as highly turbulent flow fields provide the worst-case scenarios for flashback due to turbulent flame propagation in the core flow. If the simple relationship ST ≈ SL + u′ is used, which does not account for the above effects, estimates for the turbulent flame speed ST can be derived. This leads to ST /u′ < 1.3 in the nonrecuperated case and to ST /u′ < 1.5 for the worst case with intense recuperation. Since these values are substantially lower than the ST /u′ ≈ 4.5–7 required for flame propagation against the main flow velocity even in the highly swirling case (see above), there is no indication that the drop of the flow speed below the turbulent burning velocity is the most critical cause of flashback. However, high turbulence levels, which are beneficial because they improve the fuel/air mixing and lead to shorter flames, clearly deteriorate the safety margin against flashback. For fuels with low laminar flame speeds, SL , the margin is sufficiently large because the

270

Synthesis Gas Combustion: Fundamentals and Applications

turbulent flame speed, ST , does not substantially exceed the characteristic turbulent velocity fluctuation, u′, whereas this margin is smaller for the combustion of syngas. This drop may become critical if the velocity field has strong-wake, high-turbulence regions from, for example, swirler vanes, upstream separation zones, and fuel jets. Consequently, a reduction of the swirl below the level usually employed in natural gas burners appears beneficial for the combustion of fuels with high hydrogen content because this measure reduces turbulence and the turbulent flame speed, ST . In summary, a major design criterion for nozzle aerodynamics is that the axial velocity must be as high and as uniform as possible and free of strong wakes. As this criterion is important in the entire mixing zone and not only near the burner exit, designs with constant or slightly conical and accelerating airflow paths downstream of the swirler are the preferred solution. Strong acceleration of the flow bears the danger of flame stabilization upstream near the fuel injector in stoichiometric zones near the fuel jets, in the event that the flame can propagate through the high-velocity area downstream, such as during compressor surge.

9.3.2 Combustion Pulsation-Induced Flashback A second flashback mechanism occurs through velocity fluctuations in the burner associated with combustion instabilities. At high pulsation levels the velocity field in the burner is substantially modulated. This modulation leads to the periodic drop of the flow velocity below the time average, and the generation of large-scale vortices. If the frequency is low enough, the flame will propagate upstream. Although this basic mechanism is independent of fuel type per se, the pulsation level at which it becomes significant is a function of the steady-state flashback margin described above. As such, this critical pulsation amplitude decreases with increases in hydrogen concentration of the syngas. However, since high pulsation levels must be avoided for other reasons, mentioned in Section 9.4, flashback due to the second mechanism should not occur in regular, stable combustor operation. Its significance stems from the fact that unexpected combustion instabilities could lead to catastrophic burner failure due to flashback triggered by the pulsations.

9.3.3 Flashback in the Boundary Layer Flashback in laminar boundary layers is a classical topic that has been extensively investigated (Lewis and von Elbe, 1987; Wohl, 1952; Putnam and Jensen, 1948). An investigation of boundary layer flashback in laminar, syngas-fueled Bunsen flames has been detailed by Davu et al. (2005). Near the wall, the low velocities, as well as the boundary layer turbulence, promote flame propagation upstream. These effects compete with flame quenching due to heat loss at the burner wall and flame stretch. As flashback limits in laminar flows clearly correlate with the velocity gradient at the wall, the concept of the critical velocity gradient has been developed in the past. In laminar flow, this gradient g f correlates with the laminar burning velocity SL and a quenching distance dp:

g f ∝ SL d p

(9.3)

Operability Issues Associated with Steady Flowing Combustors

271

The quenching distance dp is the characteristic length scale for the region at the wall, where the chemical reaction quenches due to heat losses. Scaling this quenching distance by a quantity on the order of the flame thickness, α/SL , we obtain the following criterion for assessing the influence of the fuel on flashback in the boundary layer:

g f ∝ SL2 α

(9.4)

This expression shows that an increase of the laminar flame speed has a substantial influence on the critical velocity gradient required for flashback prevention. Moreover, the influence of pressure can be estimated; note that α ~ 1/p. A quantitative evaluation of Equation 9.4 shows that the required velocity gradient for syngas with high hydrogen content is approximately one order of magnitude higher than that for natural gas. This indicates that boundary layer flashback concerns are much more critical for syngas than natural gas. Whether the critical wall gradient in turbulent boundary layers is higher than in the laminar case depends on the thickness of the quenching distance with respect to the laminar sublayer (Wohl, 1952; Schäfer et al., 2005). If the quenching distance is smaller, then the situation is similar to the laminar case. In the opposite, not yet extensively investigated case, an increase may be observed because thermal diffusion normal to the burner wall is increased by turbulence. Although flame propagation in turbulent boundary layers is an ongoing area of research, there are indications that proper aerodynamic burner designs produce substantially larger velocity gradients than required to avoid flashback for low flame speed fuels. However, the same conclusion cannot be made for fuels with high hydrogen content. In conclusion, burners with flow fields suited for the reliable premixing of natural gas may be prone to flashback in the boundary layer with high-H2 fuels. Another difficulty is that the addition of small amounts of air along the wall using an effusion technique, which has proven to be an effective measure against flashback for natural gas, may not dilute the mixtures outside the lean flammability limit in the critical near-wall zones for hydrogen-containing syngas. Even with dilution, the flame speed near the wall may be substantially higher than that for natural gas without dilution. Keeping the boundary layers as thin as possible is an essential design criterion for syngas burners, and even more important, local separation zones in the mixing zone must be avoided. Particularly critical are diffuser sections near the burner exit, which lead to a rapid increase of the wall boundary layers.

9.3.4 Vortex Breakdown-Driven Flame Propagation in the Core of Swirling Flows As opposed to boundary layer flashback, which is largely driven by flame propagation processes, a fourth mechanism is largely driven by the interaction of the heat release with swirling flows, which leads to a transition of the vortex breakdown characteristics (Kröner et al., 2003; Fritz et al., 2004). Since in gas turbine combustion, swirling flows are almost exclusively used for flame stabilization, this fourth mechanism is of major relevance for premixed syngas burners.

272

Synthesis Gas Combustion: Fundamentals and Applications

Stable flame CIVB

Uf Swirl generator

Z

Mixing tube Combustion chamber (a)

(b)

Figure 9.6  Flashback due to combustion-induced vortex breakdown (CIVB): stable flame (a), flame moving upstream with the breakdown bubble (b).

A related mechanism has been discussed within the framework of studies on flame acceleration in spinning tubes (Umemura and Tomita, 2001). Essentially the same phenomenon, referred to as combustion-induced vortex breakdown (CIVB), was observed in a tubular premix burner without center body (Figure 9.6), and it was found to be responsible for flame flashback in swirling flows with high axial velocities (Fritz et al., 2004). A similar effect is observed in burners with a center body. In such burners, the recirculation zone jumps suddenly back over the tip of the center body and propagates upstream, forming an annular bubble. For a given geometry, the dependence of the breakdown conditions depends upon swirl number. No breakdown occurs for low swirl numbers, S < ~0.5, and only vortex breakdown states are present for high swirl numbers, S > ~1. However, these two regimes are separated by an intermediate hysteresis regime where either flow state is possible. As such, for intermediate swirl values that are typical of those used in practical systems (e.g., S ~ 0.6–1.2) the system has two possible states: no vortex breakdown or vortex breakdown (Brown and Lopez, 1990; Wang and Rusak, 1997). Nominally, no breakdown occurs in the nozzle, but combustion can provide the finite amplitude perturbation required to move the system from one flow state (no breakdown) to the breakdown state. The basic phenomenon leading to the sudden flow transition is that the flame contributes to vortex breakdown, and therefore generates a region of low or negative flow velocity ahead of it. The flame advances forward, causing the location of the vortex breakdown region to advance farther upstream into the mixing zone. This process continues as the flame proceeds farther and farther upstream. In this case, flashback can occur in the core of vortical flows even if ST is everywhere much less than the flow velocity in the isothermal case. This basic effect depends upon the distribution of gas dilatation along the flame front, which in turn depends upon the heat release; that is, it is largely independent of chemical kinetic details (Noble et al., 2006). This is because gas expansion across the flame perturbs the approach flow and postflame streamlines, with a magnitude that is proportional to the density ratio across the flame and the relative inclination angle of the flame and approach flow (Noble et al., 2006).

Operability Issues Associated with Steady Flowing Combustors

273

However, kinetic effects enter through quenching of the flame as it propagates upstream. Kröner et al. (2007) measured the flashback limits of mixtures of natural gas and hydrogen and found a strong dependence on hydrogen content. They concluded that CIVB-induced flashback is a more severe problem for strain-resistant fuels, such as high-hydrogen syngas. Design rules for minimizing CIVB-induced flashback are detailed in Burmburger et al. (2006).

9.4 Combustion Instability Combustion instabilities are characterized by large-amplitude pressure oscillations that are driven by unsteady heat release (Lieuwen and Yang, 2005). A necessary, but not sufficient, condition for an instability to occur is that the unsteady pressure and heat release oscillations must be in phase (or, more precisely, that their phase difference is less than 90 degrees). Under these conditions, the heat release adds energy to the perturbation field (Rayleigh, 1945). Syngas composition variations primarily affect combustion instabilities by altering this phase angle. In order to understand how variations in fuel composition affect the phase difference between pressure and heat release fluctuations, it is necessary to consider the specific mechanism responsible for the instability. Two mechanisms are known to be particularly significant in premixed systems: fuel/air ratio oscillations and vortex shedding (Ducruix et al., 2005; Zinn and Lieuwen, 2005). In the former, acoustic oscillations in the premixer section cause fluctuations in the fuel or air supply rates, thus producing a reactive mixture whose equivalence ratio varies periodically in time. The resulting mixture fluctuation is convected to the flame, where it produces heat release oscillations after a certain convective time delay, τconv. The coupling of the premixer acoustics with the fuel system is also affected by the pressure drop across the fuel injector (which in turn can vary with the fuel’s volumetric heating value). The vortex shedding mechanism is due to large-scale, coherent vortical structures. These structures are the result of flow separation from flameholders and rapid expansions, as well as vortex breakdown in swirling flows. These vortices are convected by the flow to the flame, where they distort the flame front after a convective time delay, τconv, and thereby cause the rate of heat release to oscillate. A computed image showing this process is illustrated in Figure 9.7. Fuel/air ratio oscillations and vortex shedding become important when the resulting heat release perturbation is in phase with the pressure fluctuation. Basically, the sequence of physical processes involved in this feedback loop is as follows: a pressure fluctuation in the combustor results in a velocity and pressure fluctuation in the nozzle, which causes a perturbation in the fuel flow rate or the vorticity. This perturbation is convected to the flame, where it produces a fluctuation in the rate of heat release and, in turn, a pressure fluctuation. While we refer the reader to other references for details (Lieuwen and Yang, 2005), this can be expressed by the following relationship:

τ conv + τ chem = kT

(9.5)

274

Synthesis Gas Combustion: Fundamentals and Applications

Figure 9.7  Computed image of swirling flame distorted by vortical structures. (Courtesy of Huang and Yang, 2004.)

where τconv refers to the time required for either the equivalence ratio perturbation or the vortex to convect from its point of formation to the center of mass of the flame, τchem refers to the chemical delay time, T refers to the acoustic period, and k is an integer constant whose value depends upon the combustion chamber acoustics (Lieuwen et al., 2001; Gonzalez-Juez et al., 2005). The center of mass of the flame front, at least in terms of its phase response to perturbations, is in general a complex function of flame shape, flame length, flow velocity, and frequency. Only in the cases of low Strouhal numbers, defined roughly as the product of frequency and flame length, divided by the flow velocity, does this definition correspond to its general geometrical usage. While we raise this issue for the reader to be aware of it (see focused treatments on flame response transfer functions for a further discussion of this issue; Lieuwen, 2005), we can nonetheless understand the leading order effects of syngas fuel composition by simply tracking their effect upon the flame location. Variations in fuel composition impact the phase relationship expressed by Equation 9.5 by affecting both the convective and chemical times. The effect of fuel composition on the chemical time is clear. The effect on the convective time can be better understood from the following equation, which expresses the convective time as the sum of the convective time in the premixer (τpm) and the convective time in the combustor (τcomb):

τ conv = τ pm + τ comb

(9.6)



τ conv = [ L pm u pm ]  +  [ L f ucomb ]

(9.7)

where Lpm refers to the distance from the point of origin of the disturbance to the entrance to the combustor, upm refers to the mean convective velocity in the premixer, Lf refers to the distance the perturbation travels from the combustor entrance

275

Operability Issues Associated with Steady Flowing Combustors

0.65

Equivalence Ratio 0.70

0.75

4.0 3.5

45

3.0 2.5

40

2.0 35

1.5 1.0

30 25 1839

Prms/Pmean (%)

Combustor Length (in)

0.60 50

0.5 1922 2001 Adiabatic Flame Temperature (K)

2076

Figure 9.8  Stability map for methane-fueled swirl burner with 200oC reactants and 84 m/s inlet velocity. (From Gonzalez-Juez et al., 2005. With permission.)

to the center of mass of the flame, and ucomb refers to the mean convective velocity in the combustor. We illustrate these points with a stability map obtained at a fixed fuel composition by Gonzalez-Juez et al. (2005). Figure 9.8 shows a sample stability map (natural gas) for a swirl combustor at the inlet velocity of 84 m/s and at a preheat temperature of 200°C. This combustor is of variable length, allowing for a systematic mapping of the stability boundary dependence upon natural frequency of the combustor, 1/T. This plot shows that the instability band shifts to lower combustor lengths (i.e., higher frequencies) as fuel/air ratio increases. This trend can be understood from the above equations—increasing fuel/air ratio increases flame speed, decreasing τconv. Because instabilities occur at certain values of τconv /T, this implies that instability regions will shift to locations of proportionally smaller T, that is, shorter combustors. This result illustrates an important point that generally cannot be observed in fixed geometry systems—variations in operating conditions generally do not make a system more or less unstable from a global perspective. Rather, they shift instability islands within the parameter space. This point is important to our arguments below on fuel composition influences on combustion instabilities. In the same way, the effect of variations in fuel composition on the convective time is primarily exercised through its influence upon the location of the flame center of mass. For example, increasing the percentage of hydrogen in a syngas fuel will increase the flame speed and therefore change the location of the flame center of mass. Of course, other parameters affecting the flame speed, such as inlet temperature and equivalence ratio, will also affect the flame location. To illustrate, we present in Figure 9.9 a set of data obtained by Figura et al. (2007) from a combustor that can be varied in length from 30 to 45 inches, corresponding to a range of acoustic frequencies from approximately 300 to 400 Hz. The data are presented in the form of

276

Synthesis Gas Combustion: Fundamentals and Applications

Φ = 0.65

Φ = 0.65

Prms/Pmean (%)

Prms/Pmean (%)

4 3 2 1 0

0.60 30

35

40

45 Combustor Length (in) (a) 100% Natural gas

0.70

1.6 1.2 0.8

0.50

0.4 0.0

0.60 30

35

40

45

Combustor Length (in) (b) 75% Natural gas 25% hydrogen

Figure  9.9  Instability maps and corresponding two-dimensional chemiluminescence flame structure images at an inlet temperature of 200°C and an inlet velocity of 75 m/s with (a) 100% natural gas fuel and (b) 25% hydrogen and 75% natural gas fuel. (From Figura et al., 2007. With permission.)

a two-dimensional stability map, which is a plot of the normalized root mean square (rms) pressure fluctuation versus the equivalence ratio and the combustor length. Results are shown for two fuels: (a) 100% natural gas and (b) 25% hydrogen–75% natural gas, and for a fixed inlet velocity and temperature of 75 m/s and 200°C. The stability maps show that instabilities occur over a narrow range of combustor lengths and equivalence ratios for both fuels. In the case of 100% natural gas, the strongest instability occurs at an equivalence ratio of 0.65 and a combustor length between 38 and 39 inches, and has a frequency of 364 Hz. In the 25%-75% case, the strongest instability occurs at an equivalence ratio of 0.60 and a combustor length of 39 inches, and has a frequency of 352 Hz. Referring back to Equation 9.7, and noting that the inlet velocity and the instability frequencies are nearly the same for both fuels, we can see that the distance to the flame center of mass must be nearly the same for both fuels at the operating condition where the instabilities occur. Confirmation of this is given in Figure  9.9, where the two-dimensional chemiluminescence flame images are shown for both fuels for the equivalence ratios where the instabilities occur. These images were taken under stable conditions, which were achieved by decreasing the length of the combustor to 30 inches. The flame images show that, as expected, the shape of the flame and the location of the flame center of mass (indicated by •) are very nearly the same for both fuels. As discussed above, this can be explained by the fact that the increase in the flame speed, when hydrogen is added to the fuel, is offset by the fact that the instability occurs at a lower equivalence ratio. Similar data and points have been made by Russ et al. (2007) and Richards et al. (2007). Russ et al. (2007) analyzed the phase of the flame response of a forced flame

Operability Issues Associated with Steady Flowing Combustors

277

excited by vortical instabilities, and showed that the variation in phase of the response with fuel (for CH4, C2H6, and prevaporized kerosene) could be interpreted as a difference in time delay that directly correlated with the differing flame speeds, and therefore flame lengths (i.e., directly related to the above center of mass). Similarly, Richards et al. (2007) measured stability maps and heat release distributions of natural gas and a natural gas/propane mixture. Their data clearly show that the shift in stability associated with adding propane to the natural gas can be directly related to the corresponding change in flame center of mass. Prediction of combustion instabilities is complex due to need for predicting relative values of driving and dissipation processes. Moreover, as shown in a number of recent studies, the flame response to perturbations, which determines the energy addition rate to the acoustic field by the flame, is quite complex. However, we suggest that the key leading order effects of fuel composition can be understood from the above relatively simple characteristic time argument. In other words, understanding fuel composition effects upon combustion instability requires an understanding of the flame length, flame attachment points, and possible downstream flame standoff location. Even with these simplifications, however, understanding these effects is complex. To illustrate, consider Figure 9.10, which plots three possible flame shapes typically observed in swirling flows—note that each flame configuration will have a different center of mass. In configuration (a), the flame is stabilized on the rapid expansion and center body. If the shear at these regions exceeds the extinction strain rate of the flame at one or both locations, the flame will locally extinguish, detach, and move downstream. Configuration (b) represents a situation where the flame can still stabilize at the center body, but the shear at the rapid expansion is too high. In configuration (c), the flame strain is too high at both points and the flame is stabilized by the vortex breakdown bubble farther downstream. Clearly, fuel composition will have significant impacts on which flame configuration is present because of the strong dependence of extinction strain rate on hydrogen levels in the fuel. Furthermore, for a given flame configuration, the turbulent flame speed will have an important influence upon the flame length. In addition, the fluid mechanics of the flow also exercise strong influences on the flame location. For example, the stand-off location of the flame in configuration (c) will depend upon the vortex breakdown bubble location. Finally, even relatively small variations in flame length, standoff location, or overall configuration can exert significant influences upon instabilities in low Mach

Reactants

(a)

(b)

(c)

Figure  9.10  Schematic showing three possible flame configurations in a swirling flow, such as found in typical gas turbine combustors.

278

Synthesis Gas Combustion: Fundamentals and Applications

number flows—this is due to the fact that the pressure–heat release phase is proportional to changes in the absolute value, as opposed to the relative value, in the ratio of the convective time to acoustic period, τ/T. To illustrate, note that a 45-degree pressure–heat release phase shift, τ/T = 0.12, is induced by a variation in flame length or standoff distance of only 1.5 cm, assuming an axial velocity and frequency of 50 m/s and 400 Hz, respectively. This sensitivity grows with increases in frequency and decreases in velocity.

9.5 Autoignition Many steady flowing combustion systems have high inlet temperatures or pressures that create conditions favorable for spontaneous ignition of the fuel/air mixture. Exacerbating this situation is the lean premixed design used in most low NOX combustion systems. In order to achieve uniform mixing for improved emissions performance, longer mixing times are desired; however, this conflicts with the need to avoid autoignition. Thus, designing the premixer to avoid autoignition requires knowledge of the reactive mixture’s ignition delay time and a residence time. For gas turbine type applications, the time scale associated with physical mixing in modern low emissions combustors is on the order of 1 to 5 ms, based on bulk velocities and premixer volumes. For other applications, such as boilers or furnaces, 10 to 30 ms is more typical, as the design constraints relative to compactness are not as severe. In some applications, highly preheated, diluted reactants can be used to potentially achieve low emissions and high overall efficiency (e.g., Cavaliere et al., 2008). However, it should be recognized that, in any of these applications, the complex aerodynamics associated with swirl, separation, and strong gradients can make it difficult to assign a single time scale to represent the physical premixing time in a given system. Even if a very small fraction of the mixture has longer residence times, spontaneous ignition of this “packet” can lead to ignition of the entire mixture within the premixer. As a result, careful design of the aerodynamic flow in the premixer is critical relative to the consideration for autoignition. Not coincidently, design strategies to minimize pressure drop and mitigate flashback will also help minimize risk for autoignition. The key question then is whether the ignition delay time of the fuel being used at the local premixer conditions is longer, shorter, or the same as the premixing time scale. Morever, all aspects of the operation must be considered, such as start-up or load shedding. If the ignition delay time is much longer than the typical premixer residence time, concerns about autoignition may be allayed to some degree. This question of whether the syngas/air mixture will autoignite in the premixer can be addressed both theoretically and experimentally. A number of kinetic mechanisms for hydrogen or hydrogen/carbon monoxide are available, yet there are only a handful of experimental results available for autoignition of hydrogen or hydrogen/ carbon monoxide at compressor discharge conditions to confirm the accuracy of the models. For theoretical calculations, recent mechanisms specific to hydrogen or hydrogen-containing fuels include Mueller et al. (1999a, 1999b) Baulch et al. (1994), Akbar et al. (1997), Davis et al. (2005), San Diego 2003/08/30 (Williams et al., 2003), Hydrogen 2004 (O’Conaire et al., 2004), Li et al. (2004), and Konnov, 2008.

Operability Issues Associated with Steady Flowing Combustors

279

Table 9.1 Chemical Reaction Mechanisms for Fuels Containing H2 or CO Mechanism

Species

Reactions

Notes

Mueller et al. (1999b)

13

  34

GRI-Mech 3.0 (Smith et al., 1999) Baulch et al. (1994) San Diego 2003/08/03 Davis et al. (2005) Hydrogen 2004 (O’Conaire et al., 2004) Li et al. (2004) Konnov (2008)

53

325

Nitrogen-containing elemental reactions not included, but is available Designed for natural gas with limited C3 content

42 39 14 10

167 173   38   21

Compiled by Kaneshige (1996) Includes up to C3 Updated versions under development No CO reactions

11

  19 21

H/O species only H/O species only; includes updated hydroxyl and hydroperoxyl enthalpies of formation

In addition, GRI-Mech 3.0 (Smith et al., 1999) is widely used for natural gas kinetic calculations. GRI-Mech is optimized for methane and natural gas and so may not be expected to provide optimal results for hydrogen or hydrogen/carbon monoxide. However, since hydrogen and carbon monoxide are intermediate species in the oxidation of methane, GRI-Mech should generally provide reasonable results for the ranges it has been validated for. The mechanisms are summarized in Table 9.1. It is noted that refinement and improvement of mechanisms is an ongoing effort at many research facilities and, in light of current interest in hydrogen/syngas fuels, is a very active field at the time of writing of this chapter. Interestingly, this collection of mechanisms generally results in similar outcomes (within a factor of 5) for ignition delay time, as shown in a number of recent comparisons (e.g., Ströhle and Myhrvold, 2007; Beerer and McDonell, 2007b; Petersen et al., 2007; Chaos and Dryer, 2008). The newer updates included in Table  9.1 (e.g., Li et al., 2004; Konnov, 2008) generally address issues in the fuel-rich region or at very high pressures (e.g., >250 atm). In the moderate-pressure lean conditions, the revisions do not substantially change the predicted ignition delay time. They do provide changes in the subsequent propagation stages, however, and so may be important relative to predicting postinduction time behavior. Due to the wide-ranging conditions associated with the full complement of continuous combustion devices that may operate on hydrogen/syngas, it is important to examine the dependency of both inlet temperature and pressure. To illustrate pressure dependencies, Figure 9.11 presents calculated results using one of the mechanisms from Table 9.1. As shown, the role of pressure is quite interesting for hydrogen air combustion, with higher pressure giving rise to shorter delay times at lower temperatures, and lower pressures yielding shortest delay times at moderate temperatures. This behavior is attributed to the different explosion limit regions associated with hydrogen/oxygen reactions (Yetter et al., 1991, 1992), in which the rates of the elementary reactions involving H2O2 and HO2 exhibit strong temperature/pressure dependencies. Such behavior is attributed

280

Synthesis Gas Combustion: Fundamentals and Applications Temperature (°F) 102

2350 2040 1790 1585

1415

1270

1145

1040

945

865

790

Ignition Delay Time (s)

101 100 10–1

Mueller et al. 1999 1 atm 3 atm

10–2 10–3

5 atm 10 atm

10–4

15 atm 20 atm

10–5 0.6

0.7

0.8

0.9

1

1.2 1.1 1000/T (1/K)

1.3

1.4

1.5

1.6

Figure 9.11  Effect of pressure on ignition delay time for hydrogen/air mixtures.

to the depletion of the OH radical pool in favor of less reactive HO2 radicals (Skinner and Ringrose, 1965). In the range of more typical preheating found in practical continuous flowing combustion systems (up to around 1000°F), pressure tends to shorten ignition delay times. As a result, one might reason that gas turbines pose the most severe challenge for ignition delay. However, as pointed out above, industrial combustion applications often have an order of magnitude longer residence times, and advanced concepts for emissions reductions can feature even higher inlet temperatures than those found in gas turbines (e.g., Cavaliere et al., 2008). In light of the tremendous amount of experience and anecdotal data on natural gas–fired industrial gas turbines using lean premixed combustion strategies (e.g., Richards et al., 2001), it is helpful to assess the relative autoignition behavior of natural gas and hydrogen. For this assessment, simulations were carried out for H2 and natural gas using appropriate mechanisms for each (Mueller et al., 1999a, 1999b; Smith et al., 1999). Typical results are shown in Figure 9.12. The composition of NG used in the example is 80% CH4, 10% C2H6, and 10% C3H8 on a volumetric basis. This result shows that ignition delay times for H2 tend to be shorter than those of natural gas. Interestingly, the difference appears relatively small at temperatures below 1000 K (above 1000/T = 1.0). As the temperature increases, however, the differences are quite significant, especially above T = 1250 K (below 1000/T = 0.8). For ignition delay in premixing ducts in nonrecuperated gas turbine systems, the temperature ranges expected tend to be less than 1000 K; thus for gas turbine applications, the result shown in Figure 9.12 appears promising because it suggests that ignition delay times for hydrogen are comparable to those for natural gas. These delay times at gas turbine premixer conditions are 100 to 1000 ms. As mentioned above, the nonlinear behavior in the hydrogen case is due mainly to the second explosion limit behavior

281

Operability Issues Associated with Steady Flowing Combustors Temperature (°F) 101

2110

1790

1540

1340

1175

1040

925

Ignition Delay Time (s)

100 10–1 10–2 10–3

H2, H2, H2, H2, NG, NG,

10–4 10–5 0.6

0.7

0.8

0.9

1 1000/T (1/K)

1.1

1.2

= 1.0 = 0.8 = 0.6 = 0.5 = 1.0 = 0.5 1.3

1.4

Figure 9.12  Calculated ignition delay time at 15 atm—pure hydrogen versus natural gas (NG) at various equivalence ratios.

(Yetter et al., 1992) of hydrogen oxidation. Hence, the subtleties of mixture composition, temperature, and pressure can become more complicated in the CH4/H2/CO system than they are for natural gas. In addition, it is known that the mechanisms commonly used for natural gas are not necessarily recommended for T < 1000 K at high pressure (e.g., Smith et al., 1999). As a result, though the ignition delay appears similar at low temperatures, this is based on calculations done at temperatures that are really outside of the limits for which the mechanisms are felt to be reliable. Hence, additional validation is in order. Because pressure effects and actual delay times are important in gas turbine and other continuous combustion applications, and because of nonlinear behavior exhibited for some species like hydrogen, as shown in Figure 9.12, considerable interest in further verifying the answers to the key question about the ignition delay time exists. While shock tubes are commonly used for such measurements (Spadaccini and Colket, 1994; Petersen et al., 1999; Huang et al., 2004), flow reactors have also been used, but on a much more limited basis (Mueller et al., 1999a; Beerer et al., 2006; Beerer and McDonell, 2007b). These two approaches tend to favor high temperature and low temperature, respectively. Experimental studies for flowing syngas mixtures are very limited. Aside from work documented in a project report (Peschke and Spadaccini, 1985), no other results for syngas ignition delay at low temperatures are available. One limited study conducted resulted in a single ignition event (Boleda et al., 1998). Recently, results for hydrogen/ carbon monoxide mixtures at low temperatures have been reported (Petersen et al., 2007). Herein, results obtained in a flow reactor are plotted against predictions using the Mueller mechanism for various pressures in Figure 9.13. Because the predicted ignition delay times determined using any number of recent mechanisms are similar

282

Synthesis Gas Combustion: Fundamentals and Applications Temperature (°F) 102

2350 2040 1790 1585 1415 1270 1145 1040

945

865

790

Ignition Delay Time (s)

101 100 10–1 Mueller et al., 1999, 1 atm Mueller et al., 1999, 3 atm Mueller et al., 1999, 5 atm Mueller et al., 1999, 10 atm Mueller et al., 1999, 15 atm Mueller et al., 1999, 20 atm Beerer, et al., 2006 Beerer, et al., 2006, adjusted to 20 atm Peschke and Spadaccini, 1985 (12–23 atm) Boleda, et al., 1999, H2/CO/CO2 (5 atm) Beerer, et al., 2006, H2 (6 atm)

10–2 10–3 10–4 10–5 0.6

0.7

0.8

0.9

1

1.2 1.1 1000/T (1/K)

1.3

1.4

1.5

1.6

Figure 9.13  Comparison of existing and current ignition delay results for hydrogen/carbon monoxide mixtures: 50% H2, 50% CO.

(Beerer and McDonell, 2007a; Petersen et al., 2007), results using a single representative mechanism are sufficient to make observations. Figure 9.13 shows the previous (Peschke and Spadaccini, 1985) and recent flow reactor data (Beerer et al., 2006), along with a plot of predicted delay times. The recent results coincide with the previous ones, which is most evident when pressure is corrected. A pressure dependency of P –0.75 is used, as reported previously (Peschke and Spadaccini, 1985). As shown, when the recent results are scaled using this relation, they lie directly on top of the previous results, which were obtained for pressures between 12 and 23 atm. While the simulated delay times show a shorter delay time for higher pressures in the lowertemperature regions, in all cases the predictions are one to two orders of magnitude longer than the measured values, and diverge farther as temperature decreases. Recent work using other ignition delay techniques also suggests that the differences observed between the models and the measurements at low temperature are real and not just a property of constant-pressure flow reactors (Petersen et al., 2007; Walton et al., 2007). The significance of these discrepancies is more clearly illustrated in Table  9.2. Ignition delay times were calculated using the pressures, temperatures, and fuel/air ratios of several gas turbine engines in commercial use today. Compressor discharge temperatures were obtained from International Turbomachinery Handbook 2006 or were estimated based on an isentropic compression to the desired pressure. Delay times were calculated using an experimentally obtained correlation (Peschke and Spadaccini, 1985), and through homogeneous ignition calculations using the Mueller mechanism (Mueller et al., 1999b). Again, the discrepancies are significant. For example, for the GE 9H engine, the predicted ignition delay time is in the range of 11.8 seconds, while the

283

Operability Issues Associated with Steady Flowing Combustors

Table 9.2 Estimated Ignition Delay Times for Representative Engine Compressor Discharge Conditions Premixer Condition

Engine GE 9Ha Solar Taurus 65 Solar Taurus 60 Solar Mercury 50b GE LM6000a Siemens V-94.3Aa Siemens V-94.2a Capstone C-60b Furnace Furnace Furnace Furnace Furnace a b c d

Estimated Ignition Delay Time of H2/CO Mixture (ms)

Pressure (atm)

Air Temp. (K)

Flow Reactor Experiments (Peschke and Spadaccini, 1985; Beerer et al., 2006)

Homogeneous Ignition Model (Mueller et al., 1999b)

23 15 12.3 9.9 35 17.7 12 4.2 1 1 1 1 1

705 670 644 880 798 665 600 833 750 798 880 1,000 1,250

85 153 221 59 35 141 336 140 300 224 151 97d 52d

11,800 —c — 4,941 34,850 — — — 95000 10,000 236 0.106 0.054

Inlet temperature estimated from ideal gas, isentropic compression. Recuperated engine. — indicates no ignition within 5 minutes. Peschke and Spadaccini expression is highly extrapolated at these conditions—these conditions have crossed the second explosion limit, and thus the kinetic calculation may be more reliable for these conditions.

experimental correlation predicts ignition in only 85 ms. Table  9.2 also includes results for atmospheric combustion conditions that might be found in highly preheated lean com­bustion systems. As shown, at around 800 K, the ignition delay time drops dramatically for the kinetic model result. This is a result of crossing the “extended” second explosion limit, wherein mild ignition transitions to strong ignition. The empirical expression based on measured data is not intended to capture this regime change and is really limited only to mild ignition conditions. As a result, for the low-pressure, high-temperature results, the kinetic model result may be more reliable. Clearly, further studies are required to clarify discrepancies between measured and predicted delay times and further underscore the need to better understand low-temperature ignition behavior. The results also point out that caution is needed when applying different tools to assess this problem. The underlying assumptions contained within the comprehensive kinetic modeling (e.g., homogeneous ignition) may not apply in the case of flowing mixtures within gas turbine premixing ducts (or flow reactors). In reality, short ignition times have been observed for various

284

Synthesis Gas Combustion: Fundamentals and Applications

experimental techniques, but often the objective of those experiments was to attain homogeneous ignition (which generally occurs at higher temperatures), and therefore the observed short times were not of direct interest (Beerer and McDonell, 2007b; Chaos and Dryer, 2008). Further complicating the analysis are impacts of contaminants, particulate, and possible surface chemistry, all of which may contribute to the ignition process in practical applications. To this end, Dryer and Chaos (2008) have taken a critical look at HO2 and H2O2 reactions. Modifications of the rate coefficients of the HO2 + OH = H2O + O2 reaction were found to improve agreement with the existing low-temperature, high-pressure results. In addition, the CO + HO2 = CO2 + OH reaction was reexamined in the context of the contribution of CO to the induction chemistry associated with ignition. It was also pointed out by Chaos and Dryer (2008) that surface chemistry associated with CO/Fe and CO/Ni can result in carbonyls and other metallic particulate that can catalyze the ignition process, which may explain the short delay times in flow reactors. Even the presence of NOX compound was shown to influence the induction chemistry.

9.6 Conclusions A number of challenges remain for future investigations to clarify these issues. Fundamental flame properties, such as laminar flame speeds, stretch sensitivities, and extinction stretch rates, are largely unknown at the conditions of interest. Fur­ thermore, more system-dependent properties, such as turbulent flame speed, are also largely uncharacterized, and critical factors such as the thermodiffusive dependencies of these mixtures require clarification. In addition, the fluid mechanics of reacting swirl flows, which critically impacts all of these operability concerns, is poorly understood and requires extensive further work. A large body of systematic work for nonreacting swirl flows exists at low Reynolds numbers, but a number of open questions remain about the swirl flow dynamics in high Reynolds number, reacting flow dynamics. Furthermore, these dynamics are strongly influenced by exothermicity impacts on the fluid mechanics. As discussed, flame location appears to be more fundamentally influenced by the flame’s ability to withstand the high-strain, high-shear regions. Relative to ignition delay, it is evident that more work is needed in the low-temperature regime. Theoretical predictions suggest autoignition is not an issue with syngas; however, recent experimental results suggest that this issue needs revisiting and indeed is subject of numerous current research activities. Thus far, extreme values of ignition delay times (at least in a one-dimensional sense) provide some estimates to help guide premixer design and suggest that this should not be a “show stopper.” However, the safety margin observed approaches a factor of 10 in some cases, and with uncertainties as to the potential role of surface chemistry and particulates or other contaminants, dismissing the potential for autoignition in lean premixed combustion of syngas remains risky. As such, these problems will continue to be a rich area requiring considerable further investigation in order to understand and predict these dynamic combustion phenomena.

Operability Issues Associated with Steady Flowing Combustors

285

References Akbar, R., Kaneshige, M., Schultz, E., and Shepherd, J. (1997). Detonations in H2-N2OCH4-NH3-O2-N2 mixtures. Explosion Dynamics Laboratory Report FM97-3, Graduate Aeronautical Laboratories, California Institute of Technology (GALCIT). Baulch, D. L., Cobos, C. J., Cox, R. A., Frank, P., Hayman, G., Just, Th., Kerr, J. A., Murrells, T., Pilling, M. J., Troe, J., Walker, R. W., and Warnatz, J. (1994). Summary table of evaluated kinetic data for combustion modeling: Supplement 1. Combust. Flame 98:59. Beerer, D. J., Greene, M. U., McDonell, V. G., and Samuelsen, G. S. (2006). Correlation of ignition delay with IGCC and natural gas fuels. Final report, Contract 03-01-SR112, prepared for South Carolina Institute for Energy Studies. Beerer, D. J., and McDonell, V. G. (2007a). Experimental study of ignition delay for application to hydrogen and syn-gas fired lean premixed gas turbine engines. Paper E01 presented at the 5th U.S. Combustion Meeting, San Diego, March. Beerer, D. J., and McDonell, V. G. (2007b). Autoignition of hydrogen and air in a continuous flow reactor with application to lean premixed combustion. J. Eng. Gas Turbine Power 130:051507-1–051-507-9. Boleda, A., McDonell, V., and Samuelsen G. S. (1998). Standalone power plant running on biomass gas—Phase 1. Final report, UCI-ARTR-98-1, prepared for Reflective Energies. Brown, G., and Lopez, J. (1990). Axisymmetric vortex breakdown. Part 2. Physical mechanisms. J. Fluid Mech. 221:553. Burmberger, S., Hirsch, C., and Sattelmayer, T. (2006). Design rules for the velocity field of vortex breakdown swirl burners. ASME Paper GT2006-90495. Cavaliere, A., de Jonnon, M., and Ragucci, R. (2008). Highly preheated lean combustion. In Lean combustion technology and control, ed. Dunn-Rankin, chap. 3. San Diego: Elsevier. Chaos, M., and Dryer, F. L. (2008). Syngas combustion kinetics and applications. Combust. Sci. Technol. 180:1053. Davis, S. G., Joshi, A. V., Wang, H., and Egolfopoulos, F. (2005). An optimized kinetic model of H2/CO combustion. Proc. Combust. Instit. 30:1283. Davu, D., Franco, R., Choudhuri, A., and Lewis, A. (2005). Investigation on flashback propensity of syngas premixed flames. AIAA Paper 2005-3585. Dryer, F., and Chaos, M. (2008). Ignition of syngas/air and hydrogen/air mixtures and low temperatures and high pressures: Experimental data interpretation and kinetic modeling implications. Combust. Flame 152:293. Ducruix, S., Schuller, T., Durox, D., and Candel, S. (2005). Combustion instability mechanisms in premixed combustors. In Combustion instabilities in gas turbine engines, ed. T. Lieuwen and V. Yang, chap. 9, 179–207. AIAA Press: Reston, VA. Durbin, M., and Ballal, D. (1996). Studies of lean blowout in a step swirl combustor. J. Eng. Gas Turbines Power 118:72–77. Figura, L., Lee, J. G., Quay, B. D., and Santavicca, D. A. (2007). The effects of fuel composition on flame structure and combustion dynamics in a lean premixed combustor. ASME Paper GT2007-27298. Fritz, J., Kröner, M., and Sattelmayer, T. (2004). Flashback in a swirl burner with cylindrical premixing zone. J. Eng. Gas Turbines Power 126:276. Glassman, I. (1996). Combustion. 3rd ed. San Diego: Academic Press. Gonzalez-Juez, E., Lee, J. G., and Santavicca, D. A. (2005). A study of combustion instabilities driven by flame-vortex interactions. AIAA Paper 2005-4330. Huang, J., Hill, P. G., Bushe, W. K., and Munshi, S. R. (2004). Shock-tube study of methane ignition under engine-relevant conditions: Experiments and modeling. Combust. Flame 136:25.

286

Synthesis Gas Combustion: Fundamentals and Applications

Huang, Y., and Yang, V. (2004). Effect of swirl on combustion dynamics in a lean-premixed swirl-stabilized combustor. Proc. Combust. Instit. 30:1771. International turbomachinery handbook 2006. (2006). Elsevier: Amsterdam. 46, 6. Kido, H., Nakahara, M., Nakashima, K., and Hashimoto, J. (2002). Influence of local flame displacement velocity on turbulent burning velocity. Proc. Combust. Instit. 29:1855–1861. Kiesewetter, F., Kirsch, C., Fritz, J., Kroner, M., and Sattelmayer, T. (2003) Two-dimensional flashback simulation in strongly swirling flows. ASME Paper GT-2003-38395. Konnov, A. A. (2008). Remaining uncertainties in the kinetic mechanism of hydrogen combustion. Combust. Flame 152:507. Kröner, M., Fritz, J., and Sattelmayer, T. (2003). Flashback limits for combustion induced vortex breakdown in a swirl burner. J. Eng. Gas Turbines Power 125:693. Kröner, M., Sattelmayer, T., Fritz, J., Kiesewetter, F., and Hirsch, C. (2007). Flame propagation in swirling flows—Effect of local extinction on the combustion induced vortex breakdown. Combust. Sci. Technol. 179:1385. Kundu, K. M., Banerjee, D., and Bhaduri, D. (1977). Theoretical analysis on flame stabilization by a bluff-body. Combust. Sci. Technol. 17:153. Lewis, B., and von Elbe, G. (1987). Combustion. Flames and explosions of gases. 3rd ed. New York: Academic Press. Li, J., Zhao, Z., Kazakov, A., and Dryer, F. L. (2004). An updated comprehensive kinetic model of hydrogen combustion. Int. J. Chem. Kinetics 36:566. Lieuwen, T. (2005). Physics of premixed combustion-acoustic wave interactions. In Combustion instabilities in gas turbine engines, ed. T. Lieuwen and V. Yang, pp. 315–368. AIAA Press: Reston, VA. Lieuwen, T., and Yang, V., Eds. (2005). Combustion instabilities in gas turbine engines. AIAA Press: Reston, VA. Lieuwen, T., Torres, H., Johnson, C., and Zinn, B. T. (2001). A mechanism for combustion instabilities in premixed gas turbine combustors. J. Eng. Gas Turbines Power 123:182. Lipatnikov, A., and Chomiak, J. (2005). Molecular transport effects on turbulent flame propagation and structure. Prog. Energy Combust. Sci. 31:1. Longwell, J., Frost, E., and Weiss, M. (1953). Flame stability in bluff-body recirculation zones. Ind. Eng. Chem. 45:1629. Mueller, M. A., Kim, T. J., Yetter, R. A., and Dryer, F. L. (1999a). Flow reactor studies and kinetic modeling of the H2/O2 reaction. Int. J. Chem. Kinetics 31:113. Mueller, M. A., Yetter, R. A., and Dryer, F. L. (1999b). Flow reactor studies and kinetic modeling of the H2/O2/NOX and CO/H2O/O2/NOX reactions. Int. J. Chem. Kinetics 31:705. Muruganandam, T. M., Nair, S., Scarborough, D., Neumeier, Y., Jagoda, J., Lieuwen, T., Seitzman, J., and Zinn, B. (2005). Active control of lean blowout for turbine engine combustors. J. Propulsion Power 21:807. Nair, S., and Lieuwen, T. (2005). Acoustic detection of blowout in premixed flames. J. Propulsion Power 21:32. Noble, D., Zhang, Q., Shareef, A., Tootle, J., Meyers, A., and Lieuwen, T. (2006). Syngas mixture composition effects upon flashback and blowout. ASME Paper GT2006-90470. O’Conaire, M., Curran, H. J., Simmie, J. M., Pitz, W. J., and Westbrook, C. K. (2004). A comprehensive modeling study of hydrogen oxidation. Int. J. Chem. Kinetics 36:603. Peschke, W. T., and Spadaccini, L. J. (1985, November). Determination of autoignition and flame speed characteristics of coal gases having medium heating values. Final report for AP-4291 Research Project 2357-1. Petersen, E. L., Davidson, D. F., and Hanson, R. K. (1999). Ignition delay times of ram accelerator CH4/O2/diluent mixtures. J. Propulsion Power 15:82. Petersen, E. L., Kalitan, D. M., Barrett, A., Reehal, S. C., Mertens, J. D., Beerer, D. J., Hack, R. L., and McDonell, V. G. (2007). New syngas/air ignition data at lower temperature and elevated pressure and comparison to current kinetic models. Combust. Flame 149:244.

Operability Issues Associated with Steady Flowing Combustors

287

Plee, S. L., and Mellor, A. M. (1978). Review of flashback reported in prevaporizing/­premixing combustors. Combust. Flame 32:193. Putnam, A., and Jensen, R. (1948). Application of dimensionless numbers to flash-back and other combustion phenomena. In 3rd Symposium on Combustion, Flame, and Explosion Phenomena, p. 89. Rayleigh, J. S. W. (1945). The theory of sound, Classics of Science and Mathematics, Vol. 2. New York: Dover. Ren, J. Y., Qin, W., Egolfopoulos, F. N., and Tsotsis, T. T. (2001). Strain-rate effects on ­hydrogen-enhanced lean premixed combustion. Combust. Flame 124:717. Richards, G. A., McMillian, M. M., Gemmen, R. S., Rogers, W. A., and Cully, S. R. (2001). Issues for low-emission, fuel-flexible power system. Prog. Energy Combust. Sci. 27:141. Richards, G. A., Straub, D. L., Ferguson, D. H., and Robey, E. H. (2007). In LNG interchangeability/gas quality: Results of the National Energy Technology Laboratory’s research for the FERC on natural gas quality and interchangeability, Section 5. NETL/DOE Report. Russ, M., Meyer, A., and Buchner, H. (2007). Scaling thermo-acoustic characteristics of LP and LPP swirl flames. ASME Paper GT2007-27775. Sankaran, R., and Im, H. G. (2003). Effects of hydrogen addition on the flammability limits of stretched methane/air premixed flames. Paper presented at the 3rd Joint Meeting of the U.S. Sections of the Combustion Institute. Schäfer, O., Koch, R., and Wittig, S. (2005). Flashback in lean prevaporized premixed combustion: Non-swirling turbulent pipe flow study. ASME Paper 2001-GT-0053, Chicago, IL. Shanbogue, S., Hussain, S., and Lieuwen, T. (2009). Lean blowoff of bluff body stabilized flames: Scaling and dynamics. Progress Energy Combust. Sci. 35(1):98–120. Schefer, R. W. (2003). Hydrogen enrichment for improved lean flame stability. Int. J. Hydrogen Energy 28:1131. Skinner, G. B., and Ringrose, G. H. (1965). Ignition delays of a hydrogen–oxygen-argon mixture at relatively low temperatures. J. Chem. Physics 42:2190–2192. Smith, G. P., Golden, D. M., Frenklach, M., Moriarty, N. W., Eiteneer, B., Goldenberg, M., Bowman, C. T., Hanson, R. K., Song, S., Gardiner, W. C., Jr., Lissianski, V. V., and Qin, Z. (1999). http://www.me.berkeley.edu/gri_mech/. Spadaccini, L. J., and Colket, M. B., III. (1994). Ignition delay characteristics of methane fuels. Prog. Energy Combust. Sci. 20:431. Ströhle, J., and Myhrvold, T. (2007). An evaluation of detailed reaction mechanisms for hydrogen combustion under gas turbine conditions. Int. J. Hydrogen Energy 32:125. Syred, N. (2006). A review of oscillation mechanisms and the role of the PVC in swirl combustion systems. Prog. Energy Combust. Sci. 32:93. Thibaut, D., and Candel, S. (1998). Numerical study of unsteady turbulent premixed combustion: Application to flashback simulation. Combust. Flame 113:53. Umemura, A., and Tomita, K. (2001). Rapid flame propagation in a vortex tube in perspective of vortex breakdown phenomenon. Combust. Flame 125:820. Vagelopoulos, C. M., and Egolfopoulos, F. N. (1994). Laminar flame speeds and extinction strain rates of mixtures of carbon monoxide with hydrogen, methane, and air. Proc. Combust. Instit. 25:1317. Walton, S. M., He, X., Zigler, B. T., and Wooldridge, M. H. (2007). An experimental investigation of the ignition properties of hydrogen and carbon monoxide mixtures for syngas turbine applications. Proc. Combust. Instit. 31:3147. Wang, S., and Rusak, Z. (1997). The dynamics of a swirling flow in a pipe and transition to axisymmetric vortex breakdown. J. Fluid Mech. 340:177. Williams, F. A., et al. (2003). http://maeweb.ucsd.edu/combustion/cermech/. Williams, F. A. (1966). Flame stabilization of premixed turbulent gases. Appl. Mechanics Surveys 1966:1157–70.

288

Synthesis Gas Combustion: Fundamentals and Applications

Wohl, K. (1952). Quenching, flash-back, blow-off theory and experiment. In 4th Symposium (International) on Combustion, 69–89, Boston, MA. Yamaguchi, S., Ohiwa N., and Hasegawa, T. (1985). Structure and blow-off mechanism of rod-stabilized premixed flame. Combust. Flame 62:31. Yetter, R. A., Rabitz, H., and Hedges, R. M. (1991). A combined stability-sensitivity analysis of weak and strong reactions of hydrogen/oxygen mixtures. Int. J. Chem. Kinetics 23:251. Yetter, R. A., Dryer, F. L., and Golden, D. M. (1992). In Major research topics in combustion, ed. M. Y. Yussaini, A. Kumar, and R. G. Voigt, 309. New York: Springer-Verlag. Zhang, Q., Noble, D. R., Meyers, A., Xu, K., and Lieuwen, T. (2007). Characterization of fuel composition effects in H2/CO/ch4 mixtures upon lean blowout. J. Eng. Gas Turbines Power 129:688. Zhang, Q., Noble, D. R., Shanbhogue, S. J., and Lieuwen, T. (2007). PIV measurements in H2/CH4 flames under near blowoff conditions. Paper presented at the 5th U.S. Combustion Meeting, San Diego, CA. Zhang, Q. (2008). Lean blowoff characteristics of swirling H2/CO/CH4 flames. Ph.D. thesis, Georgia Institute of Technology. Zinn, B. T., and Lieuwen, T. (2005). Combustion instabilities: Basic concept. In Combustion instabilities in gas turbine engines, ed. T. Lieuwen and V. Yang. AIAA Press. Zukoski, E. E. (1997). Afterburners. In Aerothermodynamics of gas turbine and rocket propulsion, ed. G. Oates, chap. 2, 47–147. New York: Dover.

of 10 Combustion Syngas in Internal Combustion Engines Melanie K. Fox, Gregory K. Lilik, André L. Boehman, and Olivier Le Corre Contents 10.1 Introduction...................................................................................................290 10.1.1 Hydrogen-Assisted Combustion of Spark Ignition (SI) Fuels...........290 10.1.1.1 Early Use of Hydrogen in SI Engines................................. 290 10.1.1.2 Impacts of Hydrogen on Methane and Gasoline Combustion......................................................................... 291 10.1.2 Hydrogen-Assisted and Dual-Fuel Combustion of Compression Ignition (CI) Fuels............................................................................. 296 10.1.2.1 Ignition Delay Correlations for Dual-Fuel Combustion..... 298 10.1.2.2 Performance and Emissions in Dual-Fuel CI Combustion with Hydrogen................................................300 10.2 Experimental.................................................................................................302 10.2.1 Hydrogen-Assisted Methane and Natural Gas Combustion..............302 10.2.1.1 Experimental Configuration for Spark Timing Study for Hydrogen-Assisted Methane Combustion.....................302 10.2.1.2 Experimental Configuration for HCNG Studies................. 303 10.2.2 Dual-Fuel Combustion with Hydrogen and Diesel and with Syngas and Diesel..............................................................................304 10.2.2.1 Experimental Configuration for Dual-Fuel Combustion with Hydrogen and Diesel...................................................304 10.2.2.2 Experimental Configuration for Dual-Fuel Combustion with Syngas and Diesel.......................................................304 10.3 Results and Discussion..................................................................................306 10.3.1 Hydrogen-Assisted Methane and Natural Gas Combustion..............306 10.3.1.1 Spark Timing Study for Hydrogen-Assisted Methane Combustion.........................................................................306 10.3.1.2 HCNG Combustion Studies................................................ 311 10.3.2 Dual-Fuel Combustion with Hydrogen and Diesel and with Syngas and Diesel.............................................................................. 314

289

290

Synthesis Gas Combustion: Fundamentals and Applications

10.3.2.1 Dual-Fuel Combustion with Hydrogen and Diesel............. 314 10.3.2.2 Dual-Fuel Combustion with Syngas and Diesel................. 317 10.4 Conclusions.................................................................................................... 324 Acknowledgments................................................................................................... 325 References............................................................................................................... 325

10.1 Introduction Advanced power systems that are projected to achieve high efficiency and low emissions rely on synthesis gas as a key intermediate energy carrier (Rao et al., 2002; U.S. DOE, 2003). In such systems, coal or other fuels are converted to synthesis gas (composed mostly of hydrogen and carbon monoxide) via gasification or partial oxidation. Some targets that have been considered for advanced power plant designs are 75% thermal efficiency for natural gas–fueled plants on a lower heating value (LHV) basis and 60% for coal-fueled plants on a higher heating value (HHV) basis while producing electricity, without CO2 capture and sequestration or coproduction of any transportation fuels. A target for coal-based plants producing H2 or transportation fuels only is a minimum fuel utilization efficiency of 75% on a LHV basis (Rao et al., 2002). Analyses of system configurations and efficiency limitations have suggested that the only means to achieve such efficiency targets requires gas turbines integrated with solid oxide fuel cells (SOFCs) in hybrid power systems. Because of their role in distributed energy (DE) production and their combination of high efficiency and low cost, advanced reciprocating engines are another potential means of converting synthesis gas into power. Mixtures of hydrogen and carbon monoxide have high antiknock behavior and therefore could serve as spark ignition (SI) fuels and as homogeneous charge compression ignition (HCCI) fuels (Shudo and Takahashi, 2004; Shudo, 2006). However, addition of hydrogen to carbon monoxide or to methane tends to increase combustion temperatures and increases NO emissions under stoichiometric SI combustion (Li and Karim, 2005). So, such mixtures may be more appropriate in lean-burn applications where combustion temperatures are moderated by excess air. An example of hydrogen–natural gas or hydrogen– methane mixtures is the study of hydrogen-enriched natural gas (known as HCNG or hythane). HCNG shows significant benefits over conventional natural gas combustion in spark ignition engines, as will be described in some detail in the next section. Hydrogen can also be used to assist in SI combustion of gasoline.

10.1.1 Hydrogen-Assisted Combustion of Spark Ignition (SI) Fuels 10.1.1.1 Early Use of Hydrogen in SI Engines The first use of hydrogen in SI engines dates to 1820 with the Reverend W. Cecil (Norbeck et al., 1996). According to Norbeck et al., between 1860 and 1870, Otto burned a fuel containing 50% hydrogen. Over the last 40 years, hydrogen addition to different fuels has been considered (e.g., Stebar and Parks, 1974; Parks, 1976; Varde, 1981). Results showed lower CO and HC emissions for a given air–fuel ratio (AFR) but higher NO emission in comparison with pure fuels. Thermal efficiency is linked

291

Combustion of Syngas in Internal Combustion Engines

to AFR rather than the fuel blend. Finegold (1976) and Houseman and Hoehn (1974) confirmed those tendencies. However, a later work by Al-Janabi and Al-Baghdadi (1999) indicated that NOX emissions could be decreased by increasing the AFR. Some of these past observations are presented in Tables  10.1 and 10.2, based on single-cylinder and multicylinder engine studies. 10.1.1.2 Impacts of Hydrogen on Methane and Gasoline Combustion The unique characteristics of hydrogen can alter the combustion of other fuels when hydrogen is blended or burned with other fuels. Some relevant properties for hydrogen, methane, and gasoline are presented in Table 10.3, which highlights the differences between hydrogen and other more conventional fuels. Hydrogen can be a good additive for natural gas or gasoline for the following reasons (e.g., Bauer, 1999; Bauer and Forest, 2001):





1. Operating conditions can be very lean, wherein the fuel equivalence ratio, ϕ, can reach a limit of 0.1. Natural gas and gasoline have lean limits of 0.53 and 0.7, respectively. 2. The laminar flame speed for a stoichiometric hydrogen–air mixture (265­ to 325 cm/s) is around seven times higher than that for methane or gasoline. This property of hydrogen leads to a very fast rate of heat release and a decrease of the wall heat transfer, which is between 17 and 25% of the primary fuel energy for hydrogen, compared to 22 to 33% for natural gas or 30 to 42% for gasoline. 3. The flame temperature for a stoichiometric hydrogen–air mixture is around 2320 K, while for natural gas it is approximately 2150 K and for gasoline around 2470 K. 4. The autoignition temperature is around 860 K for a stoichiometric hydrogen-air mixture, 810 K for methane, and 750 K for gasoline at standard pressure (1 atm). This property raises important theoretical consideration of the compression ratio, CR, and thermal efficiency, η, defined as follows: CR =

Vc + Vd Vc

(10.1)

where Vc is the clearance volume and Vd the displaced volume, and

 1  η = 1−    CR 

γ−1

 1  or η = 1 −    CR 

n−1



(10.2)

where γ is the ratio of specific heats and n is the polytropic coefficient considering the real gases (air and fuel) and heat transfer. Norbeck et al. (1996) reported the development of an SI engine fueled by hydrogen with an efficiency between 42 and 46% with NO emissions lower than 100 ppm. 5. Emissions from hydrogen-fueled engine exhaust are neither toxic nor photochemical reactive (e.g., Hoekstra et al., 1996).

Comments

Power  modif. Spark advance (SA)

 Obj.: Equivlent Zero-Emission Vehicle (E.Z.E.V. )





CO

CO2

Power 





















HC

Fixed 20CA

82.5 114.3 612 4–18

1

Bade and Karim (1999) Waukesha CFR

NOX

73 77 322 8.5 1800

1

Apolescu and Chiriac (1996) Dacia

High efficiency

82.55 92.08 492.8 14.04 1700

1

GN—H2 1

125 130 1,595 11 Fixed Variable

Collier et al. (1996) Hoekstra et al. (1996) CFR

Nagalingam et al. (1983) 523.001

Efficiency

Engine specification Fuel Number of cylinder Bore Stroke Displaced volume Compression ratio Speed (rev/min) Spark ignition CA

Reference

Table 10.1 Single-Cylinder Engine Research on Hydrogen-Assisted Combustion

Specific fuel consumption—14%

–26%

Near null

–50%

+30%

610 8.5

1

Bauer (1999) Bauer and Forest (2001) CFR

Operating conditions ϕ = 0.4 – 0.6

Lower/pure fuel

Greater/pure fuel

Variable 1500

76.2 110

1

Al-Baghadi and Al-Janabi (1999, 2003) E6/U.S. CFR

292 Synthesis Gas Combustion: Fundamentals and Applications

Comments

Hyperbole HC/NOX

   

NOX HC CO CO2

  

8.8

9/8.8

Best Efficiency Spark Advance (B.E.S.T.)

4 89 84 2100

Cattelan and Wallace (1995) Chevrolet

4/4 85/85 86/70 1981/1588

Swain et al. (1993) Nissan/Toyota



Var.

2500

5700

Hythane 8

Raman et al. (1996) GM

Efficiency

Engine specifications Fuel Nb cylinders Bore Stroke Displaced volume Compression ratio Speed (rev/min) Spark advance CA

Reference

    Obj. E.Z.E.V.

1700

8.5

8 90.2 90 4600

Collier et al. (1996) Hoeskstra et al. (1996) V8

Divided by 4 on NOX

Benefit 1 point absolute

3800

9

8V 107.95 101.6 7410

 50%  58% Weak decreasing

Var.

10.5

6 102 120 5900

Vandenborre and Sierens (1996) Munshi et al. Verhelst and Sierens (2001a, 2001b) (2004) Crusader T7400 GM 454 AVL List

Table 10.2 Multicylinder Commercial Engine Research on Hydrogen-Assisted Combustion

Benefit 1 point absolute 0.1 g/kWh 2 g/kWh 1.5 g/kWh

11,000

6

Collier et al. (2005) Daewoo

Combustion of Syngas in Internal Combustion Engines 293

294

Synthesis Gas Combustion: Fundamentals and Applications

Table 10.3 Relevant Physical and Chemical Properties of Hydrogen, Methane, and Gasoline Hydrogen H2

Natural Gas CH4

Gasoline C8H18

Carbon-hydrogen ratio Density, kg/m3 Lower heating value, kJ/kg Lower heating value, kJ/m3 Stoichiometric air–fuel ratio, kg/kg

0 0.0893 119,930 10,708 34.20

0.25 0.7143 50,020 35,730 17.19

0.44 4.4 44,500 195,800 15.08

Equivalence ratio, ϕ

0.1

0.53

0.70

Laminar speed of flame, cm/s Temperature of flame, K Heat wall, %

265–325

37–45

37–43

2,318 17–25

2,148 23–33

2,470 30–42

Reference Peschka (1993) Peschka (1993)

Das (1990), Nagalingam et al. (1983) Das (1990), Nagalingam et al. (1983) Peschka (1993) Peschka (1993)

Hydrogen also suffers disadvantages for the following reasons:







1. The lower heating value of hydrogen is 10,700 kJ/m3, compared to 35,730 kJ/m3 for methane and 195,800 kJ/m3 for gasoline. One should note that on a mass basis the lower heating value of hydrogen is 119,930 kJ/kg, three times that of methane and gasoline. The density of hydrogen, however, is only 0.0893 kg/m3 at standard temperature and pressure, giving a much lower heat value on a volumetric basis. 2. If hydrogen is inducted with the intake air, it will occupy a significant fraction of the in-cylinder volume (around 29%, compared to 9.5% for natural gas and 1.8% for gasoline, under lean conditions), thus decreasing the volumetric efficiency. 3. Hydrogen has a tendency to autoignition. Very little energy is needed to initiate the combustion of a stoichiometric hydrogen–air mixture, 0.02 mJ compared to 0.29 mJ for CH4 and 0.24 mJ for gasoline (Norbeck et al., 1996). As a consequence, equivalence ratio hydrogen can be ignited from hot spots or from residual gases at almost any equivalence ratio (Das, 1996a). 4. The quenching distance is thinner for hydrogen than for methane or gasoline (0.064 cm for H2, 0.203 cm for CH4, and 0.2 cm for gasoline), which tends to increase cylinder heat loss (Das, 1996b).

Blending of hydrogen and natural gas provides an opportunity to gain environmental advantages. Natural gas is comprised of light alkanes (CH4, C2H6, C3H8, C4H10, and C5H12) and inert gases (CO2 and N2). Assuming that its main constituent, CH4, approximates natural gas, the ideal combustion equation is written as

295

Combustion of Syngas in Internal Combustion Engines

(1 − f ) CH 4 + f H 2 + (2 − 1.5 f )(O2 + 3.76 N 2 ) ⇒



(1 − f ) CO2 + (2 − f ) H 2O + 3.76 (2 − 1.5 f ) N 2



(10.3)

where f is the volume fraction of hydrogen in the H2–CH4 blend. The stoichiometric mass air–fuel ratio (AFRs) is defined by AFRs =



(2 − 1.5 f )( M O2 + 3.76 M N 2 ) (1 − f ) M CH 4 + f M H 2

(10.4)

where M is the molecular weight. The fuel equivalence ratio, ϕ, is commonly defined as φ=



1 m blend AFRs m air

(10.5)

For a generic fuel with the formula CHy, we have y=



CH y +

nO2 (O2 + 3.76 N 2 ) ⇒ φ

4 (1 − f ) + 2 f 1− f

(10.6)

(10.7)

n p ( xCH y CH y + xCO CO + xCO2 CO2 + x N 2 N 2 + x NO NO + x H 2O H 2O + x H 2 H 2 ) The higher molecular diffusivity of hydrogen than of hydrocarbons yields better mixing of fuel and oxidizer, therefore affecting flame characteristics by increasing the burning rate (Andrea et al., 2004; Sobiesiak et al., 2002). The increased burning rate decreases cycle-to-cycle variations. Furthermore, burning ends earlier in the expansion stroke and the chance for misfire is reduced, thereby effectively improving combustion stability and combustion product composition (Sita Rama Raju et al., 2000). The laminar flame velocity of natural gas is about 50 to 60% lower than that of gasoline (Karim, 2002). A slow burning rate increases conduction-governed heat losses, therefore leading to a cooler flame temperature. The resultant formaldehyde contributes to cycle-to-cycle variations (Karim et al., 1996) and also toxic emission formation from natural gas combustion (Li and Karim, 2005). The next sections provide details on the specific role of hydrogen in CH4 (or natural gas) combustion through experimental data on combustion with hydrogen blending. Two different experiments are described to shed light on how hydrogen addition affects, or assists, spark ignition combustion.

296

Synthesis Gas Combustion: Fundamentals and Applications

10.1.2 Hydrogen-Assisted and Dual-Fuel Combustion of Compression Ignition (CI) Fuels Hydrogen and carbon monoxide mixtures could also serve in dual-fuel engines that operate under compression ignition (CI) using a pilot injection of diesel fuel. While there is little published work on the use of synthesis gas as a fuel for IC engines, there has been a substantial effort in Bilcan’s thesis (2003) on the use of various gaseous fuels, including synthesis gas, in dual-fuel compression ignition engines. Bilcan’s work (2001) is one example of the large body of activity on pilot-ignited dual-fuel diesel engines that operate on a combination of diesel fuel and natural gas. Other engine studies related to synthesis gas include the work by Karim and coworkers (Karim and Moore, 1990; Karim and Wierzba, 1992) and McMillian and Lawson (2006), who examined the possibility of synthesis gas production via a natural gas– fueled “partial oxidation” engine. In the work by McMillian and Lawson, a sparkignited engine was operated at equivalence ratios of 1.3 to 1.6 and was shown to yield H2 concentrations as high as 11 vol.% in a spark ignition mode. They estimated that hydrogen concentrations as high as 20 vol.% could be achievable by operating in a HCCI operating mode. Since there has been little reported work on the operation of internal combustion engines on synthesis gas, the rest of the CI engine sections of this chapter will focus on the dual-fuel application of synthesis gas and hydrogen as internal combustion (IC) engine fuels. Dual-fuel engines have been employed in a wide range of applications to utilize gaseous fuels. They are most commonly modified diesel engines and can achieve very low emission levels, particularly smoke and particulates. Benefits with the dualfuel conversion include smoother and quieter operation, significantly longer engine life between overhauls, fuel savings, and enhanced safety. The gaseous fuel, which is called the primary fuel, provides most of the energy input. This is inducted along with air and compressed. At full load, around 80% of the total energy could be contributed by the primary fuel. The pilot fuel is usually diesel and, in fact, is used to ignite the gaseous fuel–air charge. The injection of the pilot fuel takes place near top dead center (TDC), like in the diesel engine. The pilot fuel self-ignites and forms multiple ignition centers from which the combustion of the primary fuel is initiated (Poonia et al., 1998, Liu and Karim, 1997). Finally, the gaseous fuel and pilot fuel burn together in the combustion chamber. The combustion process in a dual-fuel engine tends to exhibit a combination of features of both diesel and spark ignition engines. For CI engines, the ignition of the primary fuel (i.e., which is typically the gaseous fuel in dual-fuel CI combustion) is activated by the in-cylinder conditions. Some fuels do not have good enough ignition quality to enable ignition. Therefore, two fuels must be used, as shown in Figure 10.1. First, a pilot fuel, which could be, for example, diesel fuel, is injected, resulting in ignition and a subsequent temperature rise in the combustion chamber. Then, the second fuel, which could be, for example, syngas, is injected and ignites as the chamber temperature increases. In dual-fuel engines, the energy released by combustion comes partly from the combustion of gaseous alternative fuel, while the diesel fuel continues to provide, through timed cylinder injection, the remaining part of the energy released. Ideally, in

Combustion of Syngas in Internal Combustion Engines Pilot fuel Inlet

Spray

Exhaust gas

297

Figure 10.1  Conceptual diagram of dual-fuel CI engine. (Reprinted with permission from SAE 2005-01-1731, © 2005 SAE International.)

Combustion

Air and syngas

relation to the gaseous alternative supplied to the engine, there is a need to determine the optimum diesel fuel quantity at a particular engine operating condition, so as to provide the best performance over the desired load range. The main aim is to minimize the use of diesel fuel due to environmental considerations and maximize its substitution by alternative fuels throughout the load and speed ranges. The dual-fuel engine is an ideal multifuel engine that can operate effectively on a wide range of fuels with the flexibility of operating as a conventional diesel engine. The typical combustion process in a dual-fuel engine consists of four stages: an ignition delay period, premixed combustion of the pilot fuel, premixed combustion of the gaseous fuel, and diffusion combustion of the gaseous fuel together with the combustion of the remaining pilot fuel, as shown in Figure 10.2. Of interest is the determination of the ignition delay, which is the time delay between the injection of the pilot fuel and the initiation of chemical heat release, and correlation of the ignition delay for various fuel combinations. During the ignition delay period, complex chemical reactions take place. The ignition delay can be correlated by using an Arrhenius equation (see Hardenberg and Hase (1979)), which has been modified by Prakash et al. (1999) for biogas-diesel systems. This relation takes into account several effects, such as oxygen concentration and variations of polytropic coefficient. Bilcan et al. (2001) have proposed an expression of the polytropic coefficient for different gaseous fuels, which has been validated for a syngas-diesel engine by Garnier et al. (2005). The ignition delay for the pilot ignition of various gaseous fuels has also been measured by Kavtaradze et al. (2005), who compared ignition delays for diesel fuel, natural gas, and mixtures referred to as synthesis gas (70% N2 + 30% CH4 and 60% H2 + 20% CH4 + 20% N2). While these synthesis gas mixtures are substantially different from the synthesis gas expected within an advanced technology power plant, the results are nonetheless instructive. Kavtaradze et al. found that for

298

Synthesis Gas Combustion: Fundamentals and Applications 40% load 78.7% substitution

Pilot fuel premixed combustion peak

60 (2)

ROHR, (J/CA)

50 40

(3)

Gaseous fuel premixed combustion peak

30

Start of combustion 20

Pilot fuel injection

(4 )

10 (1)

–40 –30 –20 –10

0

–10

0

10

20

30

40

50

60

70

80

Crank angle (CA)

Figure 10.2  Characteristic stages of the rate of heat release for combustion in a dual-fuel engine. (Reprinted with permission from SAE 2005-01-1731, © 2005 SAE International.)

pilot ignition of various gaseous fuels, including the use of exhaust gas recirculation, the following formula was effective for correlating ignition delay.

E  τ = c K pθ−inj1.3 exp  R   Tθinj   

(10.8)

where c is an empirically determined constant with a unit of time (s); K is an empirically determined parameter equal to 0.9z –0.09, where z is the percentage of exhaust recirculation; p is the in-cylinder pressure (bar); θinj is the crank angle at the fuel injection; E is the activation energy (J/mol); and R is the ideal gas constant (J/mol-K). In comparing this correlation of the ignition for different gaseous fuels and for various levels of exhaust gas recirculation, it was observed that the impact of hydrogen on pilot-ignited diesel combustion is to shorten both the ignition delay and the duration of combustion. 10.1.2.1 Ignition Delay Correlations for Dual-Fuel Combustion Two methods exist in correlating the ignition delay for dual-fuel combustion. The first is based on the Livengood and Wu (1955) integral, and the second on an Arrhenius equation. 10.1.2.1.1 Integral Approach of Livengood and Wu Hountalas and Papagiannakis (2000) studied the ignition delay with the following equation:

Combustion of Syngas in Internal Combustion Engines

299

Θ



I (Θ) =

∫αp 0

1

−k

E  exp  a  T 

dθ

(10.9)

where θ is the current crank angle, p(θ) the in-cylinder pressure, T(θ) the in-cylinder temperature, and Θ a specific crank angle. The integral is evaluated from the ignition angle, taken as the reference; that is, the lower limit of integration is 0. The ignition delay τ is obtained when the integral is equal to 1; that is, when the Livengood and Wu integral equals 1, the specific crank angle is by definition the ignition delay τ, that is, I(τ) ≡ 1. The parameters are fitted to obtain good agreement with experiments for a given fuel. 10.1.2.1.2 Arrhenius Equation Approach The ignition delay τ models for diesel or dual fuels are based on the Arrhenius equation (c.f. Aligrot et al., 1997; Assanis et al., 1999; Heywood, 1988; Liu and Karim, 1995; Ramos, 1989).

 E  τ = A p − k exp  a  RT 

(10.10)

where τ is the ignition delay (ms), p and T are the pressure (bar) and the temperature (K) averaged over the ignition process, Ea is the activation energy (J/mole), R is the universal gas constant (J/mole/K), and A and k are kinetic parameters. Hiroyasu (1985) correlated ignition delay in an Arrhenius form as follows:

 E τ = A p − k GFAR− n exp  a RT

  

(10.11)

where GFAR denotes the global fuel–air ratio. Mansour et al. (2001) incorporated the engine speed N and the inlet temperature Tin in the above equation, yielding:

 E  τ = A N − a Tin−b p − k exp  a  RT 

(10.12)

The Prakash model (Prakash et al., 1999) was formulated to integrate the engine speed (i.e., the piston speed), the activation energy depending on the cetane number, the oxygen concentration, and thermodynamic conditions (i.e., pressure and temperature) at TDC. Garnier et al. (2005) found that the Prakash model, shown in the following formula and substantially more complex than that used by Kavtaradze et al. (2005) in Equation 10.8, provided an effective correlation:

(

)

τ = AC f Ock exp Ea D + Q 0.63

(10.13)

300

Synthesis Gas Combustion: Fundamentals and Applications

Thus, an effective correlation of the ignition delay for dual-fuel combustion can be established. Compared with other gaseous fuels (e.g., natural gas), syngas tends to shorten the ignition delay and the duration of combustion in a dual fuel. 10.1.2.2 Performance and Emissions in Dual-Fuel CI Combustion with Hydrogen Hydrogen has been used in conjunction with diesel fuel to power CI engines. This dual-fuel combustion is often called diesel pilot-ignited hydrogen combustion. Diesel pilot-ignited hydrogen combustion with a small quantity of hydrogen is beneficial since the diesel fuel is replaced by hydrogen, which may stretch the supply of hydrocarbon fuels. There is a range of emissions results reported on diesel pilot-ignited hydrogen combustion. The size, power output, and rotational speed of the engine used in the studies skew these emission results. Lower emissions are achievable at the cost of de-rating an engine, to the point of unusable road performance. Varde and Varde (1984) conducted some of the earliest studies on hydrogen substitution. The work was performed on a 4.75 kW, single-cylinder direct-injection, naturally aspirated diesel engine. A pilot injection of diesel fuel was fixed at 22° before TDC (BTDC), and engine speed was fixed at 2400 revolutions per minute (rpm). The hydrogen was fumigated into the air intake. In this work, propane, natural gas, and hydrogen were compared in dual-fuel combustion. The flows of the gases were reported on the basis of the overall H/C ratio, which included the diesel fuel. The maximum flow rate of hydrogen was sufficient to provide 15% of the total fuel energy. A reduction of smoke was reported when hydrogen was introduced at fullrated load. A 50% reduction of smoke was reported at part load when 15% of the total fuel energy was hydrogen. Increasing the hydrogen at part load beyond 15% of total fuel energy was shown to increase soot levels, due to insufficient oxygen. NOX was seen to increase with hydrogen substitution at both part load and full load. An increase in NOX emissions of 30% was reported at full-rated load with hydrogen addition at 15% of the total fuel energy. Hydrocarbon emissions were also seen to increase with increased flow rates of hydrogen at part load and full load. Lambe and Watson (1993) conducted a study in which they optimized a CI engine for hydrogen combustion with a diesel pilot. A Petter PH1W, 6 kW open-chamber, naturally aspirated, direct-injection diesel engine was used in the study. A delayedport admission system was used to supply hydrogen. The system piloted hydrogen via a secondary valve that allowed the flow of hydrogen into the cylinder when the intake valve opened. Under high loads, an atomized water jet was employed to prevent knock. Hydrogen comprised 65 to 95% of the fuel energy. To accomplish such high fuel substitution levels, a minimum pilot diesel fuel quantity was first found. At low loads, lower efficiency was observed running in a dual-fuel mode than the situation with diesel alone. At higher loads with 75% of the maximum output, the dualfuel operation was more efficient than the diesel-only operation. Exhaust emissions were taken at 1000 and 1500 rpm with unclear quantities of hydrogen (somewhere between 65 and 95%) and were compared to diesel-only combustion. At 1000 rpm

Combustion of Syngas in Internal Combustion Engines

301

and full load, smoke was reported to be reduced by 82% with dual-fuel combustion. At 1500 rpm and full load, smoke was reported to be reduced by 20%. NOX tended to increase at 1000 rpm under light loads, with dual-fuel combustion. At 1500 rpm, NOX increased under dual-fuel combustion. CO2 emissions decreased for all loads under dual-fuel combustion. At 1000 rpm, CO2 decreased by 20%. At 1500 rpm, CO2 decreased by 85%. For both speeds, CO emissions decreased overall. Hydrocarbon emissions increased for both speeds. Nitrogen oxides were reduced by up to 70% in some cases. It was observed that combustion under dual-fuel operation is controlled by flame propagation rather than autoignition. Tomita and coworkers (2001) investigated diesel and hydrogen dual-fuel combustion using a four-stroke, single-cylinder diesel engine. Injection timing was altered over a wide range of crank angles from 67.7º BTDC to 3.2° after TDC (ATDC). Testing was conducted at 1000 rpm. Hydrogen was aspirated into the intake air. A pilot injection of diesel was used to ignite the hydrogen. Smoke was seen to decrease to near-zero levels at all injection timings and at all equivalence ratios of hydrogen. NOX emissions dropped to zero at injection timings of 40° BTDC and earlier for all equivalence ratios of hydrogen. However, at timings later than 40° BTDC, NOX emissions increased over diesel-only combustion. HC emissions only modestly decreased with increasing hydrogen. CO2 emissions decreased with increasing levels of hydrogen. Thermal efficiency was found to increase significantly at injection timings of 30° BTDC and earlier. Kumar et al. (2003) investigated the performance increase of hydrogen on vegetable oil in a CI engine. In this work, hydrogen-diesel combustion was also studied for comparison. The study was conducted on a Kirloscar AV1, single-cylinder, four-stroke CI engine with a power rating of 3.7 kW at 1500 rpm. The tests were conducted at 1500 rpm at 80 and 100% of maximum output. The diesel fuel was injected at 27 BTDC. The hydrogen was inducted into the air intake. Hydrogen flow rate was reported as hydrogen mass share, given in Equation 10.14. The hydrogen mass share used in their testing ranged between 0 and 30%, with 5% reported to be the optimum hydrogen mass share. Justification for this was unclear. Brake thermal efficiency increased by 1.7% at 100% of maximum output. At 40% maximum output, 5% hydrogen mass share caused a 1.5% reduction of brake thermal efficiency. The lower efficiency at 40% of maximum output was reported to be due to insufficient diesel fuel to ignite the hydrogen. The smoke reduced from 3.9 BSU (Bosch smoke unit) to 2.7 BSU at 5% mass share and at 100% of maximum output. At 40% of maximum output, smoke dropped from 1.5 BSU to 1 BSU at 5% of hydrogen mass share. HC emissions reduced from 100 ppm to 70 ppm at 100% of maximum output at 5% hydrogen mass share. At 40% of maximum output, HC emissions decreased from 30 ppm to ~25 ppm, at 5% hydrogen mass share. CO emissions reduced from 20% to 0.14% at 100% of maximum output and 5% hydrogen mass share. At 40% of maximum output, CO emissions decreased from 0.9% to 0.6%, at 5% hydrogen mass share. NO emissions increased from 775 ppm to 895 ppm at 100% of maximum output and 5% hydrogen mass share. At 40% of maximum output, NO showed no significant variation at 5% hydrogen mass share.

302



Synthesis Gas Combustion: Fundamentals and Applications

Hydrogen mass share =

mH 2 mH 2 + m f

(10.14)

In a fundamental study, Lu and coworkers (2004) conducted spectral analysis and chemiluminescence imaging during hydrogen addition to a high-speed direct injection (HSDI) engine under conventional and low-temperature combustion (LTC). The work was conducted on a rapid compression machine (RCM) with optical access, which operated at ~1000 rpm. Hydrogen was supplied to the cylinder along with the air. Hydrogen was added at a rate of 0%, 5%, 10%, and 15% of the energy released. The LTC mode was based on 25% and 50% EGR and late injection timing. They reported that under mixing-controlled diesel combustion, small amounts of hydrogen had no significant effects on soot temperature, soot concentration, or peak pressure. Under LTC, having 15% of the fuel energy from hydrogen led to increased soot concentration and soot temperature. The study also examined the OH radical to determine if it would oxidize part of the soot formed during early combustion. At 10% energy release from hydrogen, in LTC condition, the OH radical was found to further reduce soot concentration due to long residence timings.

10.2 Experimental 10.2.1 Hydrogen-Assisted Methane and Natural Gas Combustion Studies from two different spark ignition engine experiments will be presented so both facilities and methodologies are described here. 10.2.1.1 Experimental Configuration for Spark Timing Study for Hydrogen-Assisted Methane Combustion In this experiment, a Lister-Petter engine was adapted for hydrogen and natural gas operation and controlled by a motor generator. Table 10.4 provides specifications for the engine retrofitted for use in these experiments. Spark ignition engines are well suited for stationary applications, and particularly for natural gas–fired cogeneration plants. For such applications, energetic and environmental performance evaluations are of great importance. In this study, four parameters are considered:

1. Air–fuel equivalence ratio 2. Spark advance 3. Engine load 4. Effect of hydrogen This study corresponds to a two-level factorial design:



1. Two spark advances (12 and 16 CA before top dead center [BTDC]) 2. Two loads (80% and 100% full load) 3. Two equivalence ratios (0.741 and 0.645) 4. Four blends of hydrogen in natural gas: 0, 10, 15, and 20 vol.% H2

Combustion of Syngas in Internal Combustion Engines

303

Table 10.4 Technical Specifications of the Engine Retrofitted for Spark Ignition Studies of Hydrogen-Natural Gas Blends Spark Ignition Engine Number of cylinder Engine speed (rpm) Bore × stroke (mm) Connecting rod length (mm) Compression ratio Spark timing Valve timing (CA)

Lister-Petter Engine Retrofit 1 1,500 95.3 × 88.9 165.3 12.4 Variable BTDC Inlet open: 36 ABDC inlet close: 69 BBDC exhaust open: 76 BTDC exhaust close: 32

10.2.1.2 Experimental Configuration for HCNG Studies Spark ignition studies were conducted in a 0.5 L single-cylinder Ricardo Hydra research engine, coupled to a motoring/absorbing dynamometer. Specifications for the engine are provided in Table 10.5. Compressed natural gas (CNG) or HCNG was delivered to the engine from a 3600 psi storage tank, and controlled with a series of pressure regulators and Sierra Model 820 mass flow controller, producing an atmospheric supply to the intake air downstream of the throttle.

Table 10.5 Technical Specification for the 0.5 L Ricardo Hydra Single-Cylinder Engine Used for the Studies of Hydrogen-Enriched Natural Gas (HCNG, ~30 vol.% H2) Engine type Bore Stroke Number of cylinders Swept volume Compression ratio Aspiration Rated speed Minimum speed Maximum speed Coolant outlet temperature Oil inlet temperature

Four stroke 86.00 mm 86.00 mm 1 0.50 L 10.44:1 Normally aspirated 6000 rpm 1000 rpm 6500 rpm 85˚C 85˚C

304

Synthesis Gas Combustion: Fundamentals and Applications

Engine parameters were selected to simulate a highway cruising condition typical of a 2.2 L engine: 2000 rpm, 1.5-bar brake mean effective pressure (BMEP) at stoichiometric AFR, and maximum brake torque (MBT) timing. This condition was the starting point for each fuel, after which fueling rates were maintained, and the throttle opened incrementally to attain increasingly lean stoichiometry until the lean limit for each fuel (or fuel blend) was reached. Spark timing was adjusted for MBT at each new throttle position. In this manner, the variation between MBT timings for HCNG and CNG, and when compared with the heat release profile, the start of combustion (SOC), and combustion duration, can be determined.

10.2.2 Dual-Fuel Combustion with Hydrogen and Diesel and with Syngas and Diesel 10.2.2.1 Experimental Configuration for Dual-Fuel Combustion with Hydrogen and Diesel In the hydrogen-assisted diesel combustion studies, a DDC/VM Motori 2.5 L, fourcylinder, turbocharged, common rail, direct-injection, light-duty diesel engine was used for steady-state testing. Engine specifications are given in Table 10.6 and the general engine layout is given in Figure 10.3. The engine control unit (ECU) is the computer that controls engine operation. An unlocked ECU was used to modify and control main injection and pilot injection timings, as well as EGR valve position and fuel rail pressure. This computer interface permits modification of the engine calibration settings in the ECU in real time. Other experimental details for this test facility can be found in Zhang and Boehman (2007). 10.2.2.2 Experimental Configuration for Dual-Fuel Combustion with Syngas and Diesel For the study of dual-fuel combustion syngas and diesel, a single-cylinder, directinjection, air-cooled stationary diesel engine was adapted to work in dual-fuel mode, using syngas as the primary fuel. The main engine specifications are presented in Table 10.6 Engine Specifications for the VM Motori/DDC 2.5 L Turbodiesel Engine Used in the Hydrogen-Assisted Diesel Combustion Studies Engine Displacement Bore Stroke Compression ratio Connecting rod length Rated power Peak torque Injection system Value train

DDC 2.5L TF DI-4V automotive diesel engine 2.5 L 92 mm 94 mm 17.5 159 mm 103 KW @ 4000 rpm 340 Nm @ 1800 rpm Bosch electronically controlled common-rail injection system 4 valves/cylinder

305

d ol e Co

Engine

Diesel fuel

Intake manifold

EGR valve

Exhaust manifold

Turbo

Hot EGR

Exhaust

EGR cooler

Hot boosted air

Air

EG R

Combustion of Syngas in Internal Combustion Engines

H2 Charge air cooler

H2

Cooled boosted air

Figure 10.3  Engine configuration for the hydrogen-assisted diesel combustion studies.

Table 10.7. Atmospheric temperature during the experiments varied from 20°C up to 27°C. Additionally, the atmospheric pressure changed from 1002 mbar to 1023 mbar, and the relative humidity from 47% to 69%. The engine is connected to an electrical dynamometer. To provide the gaseous fuel, a mixing system was composed of nine pure gases provided by bottles (methane, ethane, propane, butane, nitrogen, carbon dioxide, oxygen, hydrogen, and carbon monoxide). The syngas used in this work is composed of H2 (10%), CH4 (4%), CO2 (12%), CO (25%), and N2 (49%). The thermodynamic conditions due to the autoignition and combustion of the pilot fuel (diesel) allow this low energy content gas (LHV ~ 4.7 MJ/kg) Table 10.7 Technical Specifications of the Dual-Fuel Engine Engine

Lister-Petter Diesel Engine

Number of cylinders Diesel nominal power (W) Engine speed (rpm) Bore × stroke (mm) Connecting rod length (mm) Compression ratio Injection timing Valve timing (CA)

1 2800 1500 95.3 × 88.9 165.3 18 Fixed, 20CA BTDC Inlet open: 36 BTDC Inlet close: 69 ABDC Exhaust open: 76 BBDC exhaust close: 32 BTDC

306

Synthesis Gas Combustion: Fundamentals and Applications

to be ignited along the front flame. Consequently, the pilot fuel has been kept above a minimum value of 5% to allow ignition. The experimental curves of the rate of heat release (ROHR), which are obtained from pressure versus crank angle data (with a resolution of 0.1 CA) averaged over a hundred cycles, are derived from a thermodynamic model. The so-called one-zone model (see Heywood, 1988; Thyagarajan and Babu, 1985) describes the behavior of a quasi-perfect gas, with a homogeneous temperature and pressure within the combustion chamber. These curves are used to determine the energy conversion rates during different phases of the combustion process by application to the experimental data. The amount of gas injected at constant load is defined as the substitution rate of diesel fuel expressed by (Prakash et al., 1999)

ds =

ddiesel − ddual − fuel ddiesel

(10.15)

where ds is the defined diesel substitution, ddiesel is the quantity of diesel injected in diesel mode, and ddual-fuel is the quantity of diesel injected in dual-fuel mode. Since this definition for the diesel substitution is based solely on the amounts of diesel fuel used with and without substitution, it represents both the mass and the energy fraction of the substitution on a percentage basis. The mass flow rates of air, diesel, and syngas were measured as well as the exhaust gas temperature. Substitutions between 10 and 70% were measured, but the main interest of this engine remains for substitutions beyond 30 or 40%. Besides, diesel substitution has to be significant in order to bring out the several advantages of dualfuel technology. Additional experimental details and a description of the analysis methodology are available elsewhere (Garnier et al., 2005).

10.3 Results and Discussion 10.3.1 Hydrogen-Assisted Methane and Natural Gas Combustion 10.3.1.1 Spark Timing Study for Hydrogen-Assisted Methane Combustion Figures 10.4 and 10.5, corresponding to spark timings of 12 CA BTDC and 16 CA BTDC, respectively, show the electric power output from the retrofitted ListerPetter engine operating hydrogen and methane blends. Air throttle opening was modified to obtain the different equivalence ratios (ϕ = 0.645 and ϕ = 0.741), while simultaneously the fuel flow rate (blend of natural gas and hydrogen) was adjusted to maintain the electric power. The efficiency results in Figure 10.6 show three clear trends: greater spark advance leads to greater efficiency, a greater amount of hydrogen leads to a greater efficiency, and more lean operating conditions lead to decreased efficiency. Figure  10.6 also indicates that the efficiency is nearly the same between two of the operating conditions (0% H2 – SA = 12 CA – ϕ = 0.741) and (20% H2 – SA = 16 CA – ϕ = 0.645). This result is consistent with observations in the literature that hydrogen addition leads to enhanced burn rate and thereby permits operation at more lean conditions.

307

Combustion of Syngas in Internal Combustion Engines 3.5 3 Pelec (kW)

2.5 2 1.5 1

φ = 0.741 φ = 0.645

0.5 0

0

5

10

% Vol H2

15

20

25

Figure  10.4  Electric power output from the SI engine motor generator experiment for different hydrogen blends in methane with spark timing of 12 CA before top dead center (BTDC) at equivalence ratios of 0.741 and 0.645. 3.5 3 Pelec (kW)

2.5 2 1.5 1

φ = 0.741 φ = 0.645

0.5 0

0

5

10

15

20

25

% Vol H2

Figure 10.5  Electric power output from the SI engine motor generator experiment for different hydrogen blends in methane with spark timing of 16 CA BTDC at equivalence ratios of 0.741 and 0.645.

The emissions data show the beneficial and detrimental impacts of hydrogen blending with methane, depending upon the operating condition. Figure 10.7 shows CO2 emissions and indicates the following: CO2 emissions are independent of spark timing. Lean operation dilutes the fuel–air mixture and reduces the volume fraction of CO2 in the exhaust, while blending with hydrogen dilutes the carbon content of the reactions and reduces the volume fraction of CO2 in the exhaust. Emissions of oxides of nitrogen, NOX, are well known to be dependent primarily on flame temperature. As a consequence, the trends in Figure  10.8 are consistent with expectations. An earlier spark advance gives greater NOX emissions. Weaker combustion conditions, in which more inert gases are incorporated in the reactants,

308

Synthesis Gas Combustion: Fundamentals and Applications 32

Efficiency (%)

30 28 26 SA = 12 CA φ = 0.741

24

SA = 16 CA φ = 0.741 SA = 12 CA φ = 0.645

22 20

SA = 16 CA φ = 0.645

0

5

10

15

20

25

% Vol H2

Figure  10.6  Efficiency from the SI engine motor generator experiment for different hydrogen blends in methane with spark timings of 12 and 16 CA BTDC at equivalence ratios of 0.741 and 0.645. 10 9 8 CO2 (% vol)

7 6 5

SA = 12 CA φ = 0.741

4

SA = 16 CA φ = 0.741

3 2

SA = 12 CA φ = 0.645

1

SA = 16 CA φ = 0.645

0

0

5

10

% Vol H2

15

20

25

Figure 10.7  Carbon dioxide emissions from the SI engine motor generator experiment for different hydrogen blends in methane with spark timings of 12 and 16 CA BTDC at equivalence ratios of 0.741 and 0.645.

will limit the heat release, thus limiting the flame temperature and the temperature for burned gases in which nitrogen oxides form during SI combustion. Also, with a greater amount of hydrogen in the reactants, the NOX emissions increase, since the flame temperature of hydrogen is higher than that of natural gas. Moreover, the thickness of the flame extinction zone at the periphery of the cylinder is thinner with increasing H2 content, which will also contribute to greater burned gas temperature.

309

Combustion of Syngas in Internal Combustion Engines 3500 3000

NOX (ppm)

2500 2000 SA = 12 φ = 0.741

1500

SA = 16 φ = 0.741 SA = 12 φ = 0.645

1000

SA = 16 φ = 0.645

500 0

0

5

10

% Vol H2

15

20

25

Figure 10.8  NOX emissions from the SI engine motor generator experiment for different hydrogen blends in methane with spark timings of 12 and 16 CA BTDC at equivalence ratios of 0.741 and 0.645.

SA = 12 CA φ = 0.741 SA = 16 CA φ = 0.741 SA = 12 CA φ = 0.645 SA = 16 CA φ = 0.645

3500

Unburned HC (ppm)

3000 2500 2000 1500 1000 500 0

0

5

10

% Vol H2

15

20

25

Figure 10.9  Hydrocarbon emissions from the SI engine motor generator experiment for different hydrogen blends in methane with spark timings of 12 and 16 CA BTDC at equivalence ratios of 0.741 and 0.645.

Figure  10.9 shows the hydrocarbon emissions results. For weaker combustion conditions with more inert gas in the reactants, higher HC emissions are observed. More dilute combustion conditions are not favorable to the flame propagation and consequently lead to emissions of unburned fuel. However, the greater the amount of hydrogen in the reactants, the less HC emissions are observed, indicating that either

310

Synthesis Gas Combustion: Fundamentals and Applications 800 700

CO (ppm)

600 500 400

SA = 12 CA φ = 0.741

300

SA = 16 CA φ = 0.741

200

SA = 12 CA φ = 0.645

100 0

SA = 16 CA φ = 0.645 0

5

10

% Vol H2

15

20

25

Figure  10.10  Carbon monoxide emissions from the SI engine motor generator experiment for different hydrogen blends in methane with spark timings of 12 and 16 CA BTDC at equivalence ratios of 0.741 and 0.645.

the higher combustion temperatures or enhanced burn rate with hydrogen contribute to better methane ignition and more complete burning. Also, earlier spark timing is shown to lead to greater HC emissions. Figure  10.10 shows the hydrocarbon emissions results. The authors listed in Tables 10.1 and 10.2 have observed that CO emissions decrease with the amount of H2 in the reactants. In the present work, under the more lean conditions (ϕ = 0.645) hydrogen does not provide much benefit on CO emissions, but closer to stoichiometric operation (ϕ = 0.741) the trend with H2 addition is consistent with that found by the authors cited in Tables 10.1 and 10.2. Taking into account the concern over greenhouse gas emissions and climate change, adding hydrogen to natural gas can be a partial solution to reducing CO2 and CH4 emissions from natural gas–fired engines. As shown in Table 10.8, the differences between more typical common operating conditions (spark advance (SA) = 12 CA BTDC – ϕ = 0.741 and 0% H2) and those with hydrogen (SA = 16 CA BTDC – ϕ = 0.645 and 20% H2), which has the minimum CO2 emissions in two-level factorial designs, could lead to the following comments. While the thermal efficiencies are more or less the same between these conditions, CO2 reduction is very significant, 22.7% between the two sets of operating conditions. NOX emissions are also reduced, which is supported by the findings of Collier et al. (2005) and Cattelan and Wallace (1995), mainly due to the effect of the air–fuel ratio. Nevertheless, HC emissions are significantly increased. The global warning potential (GWP) of HC emissions from natural gas combustion is approximately twenty-four times CO2 GWP (expecting that these HC emissions are largely CH4). Nonetheless, even in taking into account the contribution of the CH4 emissions, the reduction of the net GWP of the emissions is still significant.

311

Combustion of Syngas in Internal Combustion Engines

Table 10.8 Comparison of Emissions and Efficiency for Retarded Spark Timing (Spark Advance (SA) = 12 CA) and Less Lean Conditions (ϕ = 0.741) without H2 versus for Advanced Spark Timing (SA = 16 CA) and More Lean Conditions (ϕ = 0.645) with H2

Efficiency CO2 NOX HC CO

Operating OS1

Conditions OS2

SA = 12 CA ϕ = 0.741 0% H2 30.3% 8.8% 730 ppm 1360 ppm 560 ppm

SA = 16 ϕ = 0.645 20% H2 29.1% 6.8% 420 ppm 2290 ppm 705 ppm

Difference (%)

–3, 9 –22, 7 –42, 5 68, 4 25, 9

10.3.1.2 HCNG Combustion Studies In the studies of HCNG combustion, wherein roughly 30 vol.% H2 was blended with compressed natural gas, the intention was to observe how hydrogen addition affects the burn rate to subsequently sustain lean combustion. In addition, these experiments are intended to highlight the relative effects of swirl, which can enhance in-cylinder turbulence intensity and thereby increase turbulent burning velocities. The effect of hydrogen, which clearly will increase the laminar flame speed, can be expected to increase the turbulent burning velocity as well. The term quiescent is used to describe the engine head without any modifications. This configuration yields very little if any intake-induced swirl, and is denoted in the data labels by “nS.” Swirl is used to describe the configuration where one intake port is blocked fully, generating some amount of intake-induced swirl, and denoted by “S” in the data labels. Figure 10.11 shows the trend of maximum brake torque (MBT) spark timing with equivalence ratios for CNG and HCNG with and without intake-induced swirl. As seen in Figure 10.11, the spark timing trends as a function of ϕ for each fuel and swirl condition show that swirl enables retarded spark timings, which was also found by Swain et al. (1993) and Cattelan and Wallace (1995). Note that the MBT timing for CNG with intake-induced swirl in the stoichiometric range aligns closely with that for HCNG without swirl, which shows that (regarding MBT timing) hydrogen supplants swirl for CNG. These data also show that the swirl affects CNG more significantly than HCNG in terms of enabling retarded spark timings, particularly for equivalence ratios higher than 0.7. As seen in Figure  10.12, burn duration is decreased with intake-induced swirl for both CNG and HCNG combustion, as expected. The trend is generally more

312

Synthesis Gas Combustion: Fundamentals and Applications

Spark Timing (CAD Relative to TDC)

–10

–20

–30

–40

–50 1

0.9

0.8 0.7 0.6 Equivalence Ratio

0.5

0.4

0.3

Figure  10.11  Spark timing for maximum brake torque as a function equivalence ratio (ϕ) for compressed natural gas (CNG) and HCNG (~30 vol.% H2 in CNG) fueling with and without intake-induced swirl. ○, HCNG no swirl; , HCNG swirl; , CNG no swirl; ■, CNG swirl.

Burn Duration

40

30

20

10

1

0.9

0.8 0.7 0.6 Equivalence Ratio

0.5

0.4

0.3

Figure 10.12  Burn duration (based on crank angle duration between 10 and 90% mass fraction burned) at maximum brake torque (MBT) spark timing as a function equivalence ratio (ϕ) for CNG and HCNG (~30 vol.% H2 in CNG) fueling with and without intake-induced swirl. ○, HCNG no swirl; , HCNG swirl; , CNG no swirl; ■, CNG swirl.

significant for CNG than for HCNG. Between ϕ of 0.9 and 0.6, the burn duration for CNG with swirl is about the same as for HCNG without swirl. The fastest burn durations occur for HCNG with swirl, similar to the findings of Tunestal et al. (2002), and are significantly faster than the other three cases. At 2000 rpm, hydrogen and swirl both contribute to the combustion process in terms of increasing the burn rate.

313

Combustion of Syngas in Internal Combustion Engines

Coefficient of Variance of Indicated Mean Effective Pressure (%)

3

2.5

2

1.5

1

1

0.8

0.9

0.7

0.6

0.5

0.4

Equivalence Ratio (a)

Coefficient of Variance of Indicated Mean Effective Pressure (%)

3

2.5

2

1.5

1

1

0.9

0.8

0.7

0.6

0.5

Equivalence Ratio (b)

Figure  10.13  Coefficient of variance of indicated mean effective pressure (COV% of IMEP) as function of equivalence ratio for (a) CNG and (b) HCNG (~30 vol.% H2 in CNG) with and without intake-induced swirl. ○, HCNG no swirl; , HCNG swirl; , CNG no swirl; ■, CNG swirl.

The cycle-to-cycle variation for each test is shown in Figure  10.13. The trend shows that the coefficient of variance (COV%) of the indicated mean effective pressure (IMEP) is generally lower for CNG with and without swirl than for HCNG, until ϕ becomes leaner than 0.6. It is only in the leanest regions where HCNG improves COV% consistently compared to CNG, as in Figure 10.14. Note that hydrogen essentially increases the lean limit of combustion of CNG, as CNG cannot be burned at ϕ < 0.55.

314

Synthesis Gas Combustion: Fundamentals and Applications

Coefficient of Variance of Indicated Mean Effective Pressure (%)

90 80 70 60 50 40 30 20 10 0

1

0.9

0.8 0.7 0.6 Equivalence Ratio

0.5

0.4

Figure  10.14  COV% of IMEP for CNG and HCNG (~30 vol.% H2 in CNG) with and without intake-induced swirl. ○, HCNG no swirl; , HCNG swirl; , CNG no swirl; ■, CNG swirl.

10.3.2 Dual-Fuel Combustion with Hydrogen and Diesel and with Syngas and Diesel 10.3.2.1 Dual-Fuel Combustion with Hydrogen and Diesel In this section the results of experimentation on hydrogen substitution of conventional diesel combustion modes are discussed. The hydrogen for diesel substitution rate was defined on the percentage energy basis. Hydrogen was substituted for diesel in increments up to 15% on a fuel energy basis, in four unique modes. A maximum of 15% hydrogen substitution was also chosen because it is below the lower explosion limit of hydrogen in air, 4.1% volume, for all modes tested. Table 10.9 lists the operating conditions and engine injection parameters for the four test cases considered, which represent combinations of low and high speeds and loads. Figure  10.15 shows a comparison of the apparent heat release rate for one of the four test conditions, at low speed (1800 rpm) and high load (75% of maximum output). This plot is indicative of the results for the other tests. One sees a similar premixed ignition as hydrogen is added, and a much more intense early combustion of the gas phase fuel, followed by the ignition of the main diesel fuel injection pulse. This final phase of combustion is less intense when hydrogen is present because less diesel fuel is injected due to replacement of the diesel fuel energy by hydrogen. Thus, the heat release profile is significantly altered by the early and largely premixed combustion of the premixed gaseous hydrogen in the intake air. Figure 10.16 shows the trend in NOX emissions with hydrogen addition, which indicates little overall change in the aggregate emissions of oxides of nitrogen. However, despite this apparent lack of significant trend, the split between NO and NO2 varies

315

Combustion of Syngas in Internal Combustion Engines

Table 10.9 Parameters of Baseline and Hydrogen-Assisted Operation of the 2.5 L Turbodiesel Engine

Mode

Load (kW)

EGR (%)

Boost (bar)

Intake Manifold Gas (°C)

1800 rpm @ 25% max. output 1800 rpm @ 75% max. output 3600 rpm @ 25% max. output 3600 rpm @ 75% max. output

15.7 46.5 26.1 78.2

10.5   0.7   1.4   1.0

0.2 0.7 0.9 1.1

56.3 42.0 70.3 76.5

Exhaust Gas (°C)

Pilot Inj. (°BTDC)

Main Inj. (°BTDC)

332.5 476.9 247.9 476.4

17.4 38.3 56.8 58.1

–2.9 6.2 12.3 13.6

Apparent Heat Release Rate (J/deg)

100

80

60

40

20

0 –20 –40

–30

–20 –10 0 10 Crank Angle (°ATDC)

20

30

Figure  10.15  Apparent heat release rate at 1800 rpm at 75% maximum output, with 0% (——), 2.5% (—  —), 7.5% (– – – – ), and 15% (-----) hydrogen substitution on an energy basis.

dramatically, with a high percentage of the total NOX being comprised of NO2 at higher levels of hydrogen addition. This trend is consistent with an increasing role of HO2, which serves to oxidize the NO present to NO2, as indicated in Equation 10.16, with the population of HO2 increasing with increasing amounts of H2 addition.

NO + HO 2 ← → NO 2 + OH

(10.16)

316

Synthesis Gas Combustion: Fundamentals and Applications

Brake Specific NOX Emissions (g/kW-hr)

6

5

4

3

2

1

0

1800 @ 25%

1800 @ 75%

3600 @ 25%

3600 @75%

Figure 10.16  Brake-specific NOX emissions at 0%, 2.5%, 7.5%, and 15% hydrogen substitution on an energy basis at 1800 rpm at 75% maximum output.

The HO2 is produced from the increased levels of hydrogen via Equation 10.17. Glassman (1996) states that the most probable initial step in the combustion of oxygen and hydrogen is

H 2 + O 2 ← → HO 2 + H

(10.17)

Figure 10.17 elaborates on the hydrogen-induced NO-to-NO2 shift and shows that as the hydrogen substitution increases, so does the percentage of NO2 in the total NOX emissions. The effect was seen to be further enhanced in low-load operation modes in which the air–fuel ratio was fuel lean. In a recent study, Bika et al. (2008) observed similar results in which NO2 was seen to increase as a percentage of total NOX. The NO-to-NO2 shift was observed to occur in petroleum-based diesel and biodiesel, with a more significant ratio shift at less than 10% hydrogen addition on an energy basis. This study confirmed the modest NOX emission reductions observed during hydrogen aspiration. Upatnieks et al. (2005) conducted a study on an optically accessible, heavyduty DI diesel engine in which intake oxygen was diluted via nitrogen as simulated EGR, which resulted in an increased NO2-to-NO ratio. The increase of NO2 and decrease of NO was attributed to an increased quenching of the NO2-to-NO reaction (Equation 10.18) due to decreasing flame temperatures (Upatnieks et al., 2005).

NO 2 + O ← → NO + O 2

(10.18)

317 11

96

10

95

9

94

8

93

7

92

6

91

5

90

4

Percentage of Brake Specific NO in NOX Composition (%)

97

89 3 –5 0 5 10 15 20 % Hydrogen Substitution on the Energy Basis

Percentage of Brake Specific NO2 in NOX Composition (%)

Combustion of Syngas in Internal Combustion Engines

Figure 10.17  Percent of brake-specific NO (—— ) in NOX and NO2 (– – – – ■) in NOX emissions vs. energy percent from hydrogen fuel for 1800 rpm at 75% of maximum output.

The observation reported by Upatnieks et al. (2005) corresponds to the NO and NO2 emissions that resulted from hydrogen substitution via aspiration, in that the HO2 radical converted NO to NO2, then the NO2-to-NO reaction was quenched, resulting in increased NO2 emissions. Hydrogen substitution was observed to result in increased levels of NO2 emissions at fuel-lean air–fuel ratios in which combustion temperatures were decreased, quenching of the NO2-to-NO reaction. Furthermore, premixed hydrogen combustion has been shown to have higher cooling loss to the combustion chamber wall, which would further quench the NO2-to-NO reaction (Shudo et al., 2000). Despite this trend in NO2 concentration in the exhaust with H2 addition, the role of H2 is largely to replace diesel fuel without deleterious effect. Again, this represents a potential role for H2 in our energy supply over the long term, as a replacement or supplement for traditional transportation fuels. 10.3.2.2 Dual-Fuel Combustion with Syngas and Diesel In this section, we survey some of the experimental observations of Garnier et al. (2005) and others to provide an indication of the unique impact of syngas on dualfuel diesel combustion. Ignition delay (τ) is an essential parameter to determine for all diesel fuels and has a significant impact on combustion, energy conversion efficiency, and pollutant formation. As mentioned in the introduction, the determination of ignition delay for a dual-fuel engine takes on the additional complication that the vaporizing pilot fuel spray is injected not solely into air but into an air–fuel mixture. Thus, correlation of

318

Synthesis Gas Combustion: Fundamentals and Applications

the ignition delay for dual-fuel engines requires consideration of the impact of the gaseous fuel on the stoichiometry in the cylinder. As mentioned in the introduction, Garnier et al. (2005) found that the Prakash model, shown in Equation 10.19, provided an effective correlation:

(

)

τ = AC f Ock exp Ea D + Q 0.63

(10.19)

where A is as defined in Equation 10.20 with units of (m/s) and Mps is the average piston speed (m/s),

A = 0.36 + 0.22 M ps

(10.20)

the activation energy (J/mol) is linked to the cetane number (CN) of the diesel fuel,

Ea =

618840 CN + 25



(10.21)

and Oc represents the oxygen concentration in the gas, whereby the air–fuel ratio is introduced:

Oc =

[O ] charge [O ] air

(10.22)

In-cylinder conditions are set by Equations 10.22 and 10.23:



D=

1 1 − RTinj 17190

(10.23)

21.2 Pinj − 12.4

(10.24)

Q=

In Equations 10.23 and 10.24, D has units of (mol/J), Q is dimensionless, and Tinj and Pinj are the temperature (K) and pressure (bar) in the cylinder when the diesel spray is injected, as introduced by Bilcan et al. (2001) for calculating an injection compression ratio (ICR), as follows:

ICR =

Vcyl Vinj

(10.25)

where Vinj is the volume of the combustion chamber at the exact moment of injection and Vcyl is the volume of the cylinder at BDC. In-cylinder temperature and pressure at the moment of injection are as follows:

Combustion of Syngas in Internal Combustion Engines



Tinj = TBDC ⋅ ( ICR )

ndf −1

pinj = pBDC ⋅ ( ICR )

ndf

319



(10.26)



(10.27)

where ndf is the polytropic index of the gaseous charge. In Equation 10.26 and 10.27 ndf and nair are the polytropic coefficients of the compression phase, in dual-fuel mode and diesel mode, respectively. fp is the gaseous concentration in the combustion chamber.

ndf = nair + α f p

(10.28)



α=

d γ dual fuel df p

(10.29)

The unknown parameters needed to determine τ are Cf, k, and α. The coefficient α is obtained Table 10.10 using experimental data of gaseous concentration Compared Values of α fp and theoretical values of calorific capacities of for Different Fuels air and gas. The slope of the curve ndf = f (fp) at Gaseous Fuel α [–] different loads leads to values of α. The average 0.19 Biogas (63% CH4, 37% CO2) value gives α = 0.26 and is compared with other NG (100% CH4) 0.18 gaseous fuels defined by Bilcan et al. (2001) in 0.31 LPG (30% C3H8, 70% C4H10) Table 10.10. The optimal pair of factors (Cf = 1.784, k = 0.542) is found by minimizing the difference between experiment and theory, using the least squares method for ln(τ) curves. Experimental data of ignition delay are then compared to predicted data (Figure 10.18). The ignition delay is well predicted from diesel substitutions above 30%: the uncertainty does not exceed 0.5° CA. Dual-fuel engines typically operate at diesel substitutions above 50%. The overall process of dual-fuel combustion can be described by a superposition of multiple Wiebe’s functions, as shown by Liu and Karim (1997), who superposed two functions to characterize the premixed and the diffusion combustion processes. Bilcan et al. (2001) have developed a procedure using three Wiebe’s laws, one for each combustion stage. These three laws describe the ROHR for biogas (from landfill)diesel fuel engines (Figure 10.19). In order to determine the onset of the diffusion combustion phase, the predicted curve of the premixed pilot fuel combustion is subtracted from the experimental curve of ROHR. The end of the diffusion combustion is considered to occur when the burned fraction is 99.9%. Figure 10.19 illustrates an effective decomposition of the phases of the combustion process. As shown in Figure 10.20a, the peak value of the ROHR during the premixed combustion of the pilot fuel is not significantly affected by the variation of the diesel substitution, until a certain limit is reached. This limit, at 45 to 50% substitution, is

320

Synthesis Gas Combustion: Fundamentals and Applications 12 10

ID (CA)

8 6

ID pre – 40% load ID exp – 40% load

4

ID pre – 50% load ID exp – 50% load

2 0

0

20

40 Substitution (%)

60

80

Figure 10.18  Predicted and experimental ignition delay at 40% load and 50% load.

represented by the quantity of diesel that can be burned during the premixed phase. If the total quantity of diesel introduced inside the cylinder during the ignition delay becomes smaller than this limit, the maximum value of ROHR for the premixed combustion of the pilot fuel decreases. As seen in Figure 10.20b, the ROHR during the second phase of the combustion process becomes significant only after the limit in diesel substitution, around 45 to 50%. For lower levels of diesel substitution (below roughly 20%), since the fuel-toair ratio in the gaseous mixture is quite small, the second premixed peak (for the syngas) is almost imperceptible, and as a consequence, the energy released in the second phase of the combustion process is modest. Emissions of NOX show an expected trend of increasing with diesel substitution, since the increasing H2 content as a percentage of total fuel energy leads to an increase in adiabatic flame temperature, as shown in Figure 10.21. This is one reason why the use of H2 fumigation in dual-fuel combustion may also require the use of exhaust gas recirculation to prevent excessive NOX emissions. As mentioned already, there has been little consideration in the literature of the combustion of syngas in reciprocating engines, but there has been a significant amount of work on the fumigation of the intake air of diesel engines with hydrogen alone. At low concentrations of hydrogen in the intake air (less than 20% diesel substitution), one might consider this hydrogenassisted diesel combustion, while at higher levels of diesel substitution (greater than 20%), the process is the same dual-fuel combustion process considered here. In the case of hydrogen-assisted combustion in CI engines, Varde and Frame (1983) reported some of the earliest work. They observed that at low levels of hydrogen addition (e.g., 5 to 10% of the total energy injected as hydrogen in the intake air)

Combustion of Syngas in Internal Combustion Engines

321

ROHR, J/CA

Premixed combustion of gaseous fuel

Premixed combustion of pilot fuel

Crank Angle, Deg Start of the premixed combustion of pilot fuel

ROHR, J/CA

(a)

End of the premixed combustion of pilot fuel

Diffusion combustion

Crank Angle, Deg Start of the diffusion combustion

End of the diffusion combustion (b)

Figure 10.19  Effective decomposition of the rate of heat release for the different phases of combustion.

322

Synthesis Gas Combustion: Fundamentals and Applications 300 250

Qpd (J)

200 150 100 Load 40%

50 0

Load 50% 0

20

40 Substitution (%)

60

80

(a) Premixed combustion of the pilot diesel fuel 300 250

Qpg (J)

200 150 100 Load 40%

50 0

Load 50% 0

20

40

60

80

Substitution (%) (b) Premixed combustion of the gaseous fuel (Syngas)

Figure 10.20  Heat released for each part of the combustion process. (a) Premixed combustion of the pilot diesel fuel. (b) Premixed combustion of the gaseous fuel (syngas). (c) Diffusion combustion.

there were reductions in smoke emissions at part load. At higher loads and higher levels of hydrogen addition, NOX emissions increased. Kumar et al. (2003) examined hydrogen-assisted combustion of a vegetable oil fuel from the jatropha plant and showed mild efficiency improvements and reduced smoke at 5 to 7% hydrogen mass-to-diesel mass ratio. They observed a NOX increase, however, when substituting hydrogen for the jatropha oil. Lu et al. (2004) examined the impact of hydrogen addition to DI diesel combustion through the use of a rapid compression machine

323

Combustion of Syngas in Internal Combustion Engines 350 300

Qd (J)

250 200 150 100

Load 40% Load 50%

50 0

0

20

40 Substitution (%) (c) Diffusion combustion

60

80

Figure 10.20  (Continued.)

4000 3500 3000

NOX (ppm)

2500 2000 1500 1000

Load 40% Load 50%

500 0

0

20

40

60

80

Substitution (%)

Figure 10.21  Emissions of NOX as a function of load and diesel substitution.

324

Synthesis Gas Combustion: Fundamentals and Applications

with optical access. They observed increased OH radical intensity during the premixed ignition process and reduced soot intensity in the spray flame with hydrogen addition from 5 to 10% of the total energy. At higher levels of hydrogen addition, soot intensity increased. Tomita et al. (2001) showed that at fairly high diesel substitution levels, moderate load, and advanced diesel pilot injection timing, both smoke and NOX could be brought to very low levels with hydrogen addition (for pilot injections before 40° BTDC). They suggest that a well-mixed diesel fuel-hydrogen mixture provides a mild and distributed combustion throughout the cylinder, which represents a premixed charge compression ignition (PCCI) mode of combustion. One would expect that these trends in performance of dual-fuel diesel combustion with hydrogen will be indicative of the performance with syngas, since hydrogen will likely dominate the ignition characteristics of the gaseous fuel and carbon monoxide will contribute to thermal energy release as combustion proceeds to completion, just as it does in the late stages of combustion in premixed flames. Therefore, the observations by Garnier et al. (2005) for dual-fuel combustion with syngas and those of various authors for dual-fuel combustion with hydrogen together provide an indication of how IC engines could perform with the use of syngas for stationary power generation.

10.4 Conclusions The aim of this chapter is to survey the published work on the combustion of syngas in reciprocating engines and to survey the published work and present original work on hydrogen-assisted combustion. An emphasis was placed on hydrogen-assisted SI combustion of methane and natural gas and CNG, and on the dual-fuel combustion of syngas in compression ignition engines. The consideration of hydrogen-assisted SI and CI combustion also shows how H2 can play a unique and potentially valuable role in IC engines. From the experimental work and the literature reviewed here, the following conclusions can be drawn: • Hydrogen addition to methane and natural gas can have positive effects on the combustion of methane, reducing burn duration and increasing burn rates, thereby improving efficiency and reducing emissions. • Hydrogen addition to conventional CI combustion through fumigation in the intake air of a diesel engine can serve to substitute for diesel fuel, without significant detrimental effects, up to as much as 15% energy substitution from hydrogen to stay below the lower explosion limit for hydrogen. • Syngas addition (i.e., substitution of syngas for diesel fuel) tends to shorten the ignition delay and shorten the duration of combustion in dual-fuel operation. • Syngas addition tends to increase NOX emissions, presumably from the increase of adiabatic flame temperature due to the hydrogen in the syngas. • Based upon observations of hydrogen-assisted CI combustion by Tomita et al. (2001), modest amounts of syngas addition in combination with advanced injection timing of the diesel pilot can lead to effective low-temperature PCCI combustion, yielding both reduced PM and reduced NOX.

Combustion of Syngas in Internal Combustion Engines

325

Acknowledgments The authors thank the U.S. DOE and program manager Robie Lewis for their support of the hydrogen-assisted CNG and hydrogen-assisted diesel work under Contract DE-FC26-04FT42233.

References Al-Baghdadi, M. A. R. S., and Al-Janabi, H. A.-K. S. (2003). A prediction study of a spark ignition supercharged hydrogen engine. Energy Conv. Mgmt. 44:3143. Aligrot, C., Champoussin, J. C., Guerrassi, N., and Claus, G. (1997). A correlative model to predict autoignition delay of diesel fuels. SAE Technical Paper 970638. Al-Janabi, H. A.-K. S., and Al-Baghdadi, M. A.-R. S. (1999). A prediction study of the effect of hydrogen blending on the performance and pollutants emission of a four stroke spark ignition engine. Int. J. Hydr. Energy 24:363. Andrea, T. D., Henshaw, P. F., and Ting, D. S.-K. (2004). The addition of hydrogen to a gasoline fuelled SI engine. Int. J. Hydr. Energy 29:1541. Apostolescu, N., and Chiriac, R. (1996). A study of hydrogen-enriched gasoline in a spark ignition engine. SAE Technical Paper 960603. Assanis, D. N., Filipi, Z. S., Fiveland, S. B., and Syrimis, M. (1999). A predictive ignition delay correlation under steady-state and transient operation of a direct injection diesel engine. In ICE Division of ASME, Fall Technical Conference, Vol. 33-2, 99-ICE-231, pp. 95–104. Bade, S. O., and Karim, G. A. (1999). Hydrogen as an additive to methane for spark ignition engine applications. Int. J. Hydr. Energy 24:577. Bauer, C. (1999). The effect of hydrogen on the performance of methane-fuelled S.I. engines. Thesis, Faculty of Graduate Studies and Research, Edmonton, Alberta. Bauer, C., and Forest, T. W. (2001). Effect of hydrogen addition on the performance of methane-fuelled vehicles. I. Effect on SI engine performance. Int. J. Hydr. Energy 26:55. Bika, A. S., Franklin, L., and Kittelson, D. B. (2008). Emissions effects of hydrogen as a supplemental fuel for diesel and biodiesel. SAE Technical Paper 2008-01-0648. Bilcan, A. (2003). Contribution to the study of the thermodynamic cycle of dual-fuel. Ph.D. thesis, Nantes University, France. Bilcan, A., Tazerout, M., Le Corre, O., and Ramesh, A. (2001). Ignition delay in dual-fuel engines: An extended correlation for gaseous fuels. Paper presented at the ICE Division of ASME, Spring Technical Conference, Philadelphia, April 29–May 2. Cattelan, A., and Wallace, J. (1995). Exhaust emission and energy consumption effects from hydrogen supplementation of natural gas. SAE Technical Paper 952497. Collier, K., Hoekstra, R. L., Mulligan, N., Jones, C., and Hahn, D. (1996). Untreated exhaust emissions of a hydrogen-enriched CNG production engine conversion. SAE Technical Paper 960858. Collier, K., Mulligan, N., Shin, D. S., and Brandon, S. (2005). Emission results from the new development of a dedicated hydrogen-enriched natural gas heavy duty engine. SAE Technical Paper 2005-01-0235. Das, L. M. (1990). Hydrogen engines: A view of the past and a look into the future. Int. J. Hydr. Energy 15:425. Das, L. M. (1996a). Fuel induction techniques for a hydrogen operated engine. In Hydrogen fuel for surface transportation, 27–36. Society of Automotive Engineers, Warrendale, PA. Das, L. M. (1996b). Hydrogen–oxygen reaction mechanism and its implication to hydrogen engine combustion. Int. J. Hydr. Energy 21:705.

326

Synthesis Gas Combustion: Fundamentals and Applications

Finegold, J. G. (1976). Hydrogen: Primary or supplementary fuel for automotive engines. SAE Technical Paper 760609. Garnier, C., Bilcan, A., Le Corre, O., and Rahmouni, C. (2005). Characterisation of a syngasdiesel fuelled CI engine. SAE Technical Paper 2005-01-1731. Glassman, I. (1996). Combustion. 3rd ed. New York: Academic Press. Hardenberg, H. O., and Hase, F. W. (1979). An empirical formula for computing the pressure rise of a fuel from its cetane number and from relevant parameters of direct injection diesel engine. SAE Technical Paper 790493. Heywood, J. B. (1988). Internal combustion engine fundamentals. New York: McGraw-Hill Book Company. Hiroyasu, H. (1985). Diesel engine combustion and its modeling: Diagnostics and modeling of combustion in reciprocating engines. In Proceedings of the COMODIA Symposium, Tokyo, pp. 53–75. Hoekstra, R. L., Van Blarigan, P., and Mulligan, N. (1996). NOX, emissions and efficiency of hydrogen, natural gas, and hydrogen/natural gas blended fuels. SAE Technical Paper 961103. Hountalas, D. T., and Papagiannakis, R. G. (2000). Development of a simulation model for direct injection dual-fuel diesel-natural gas engines. SAE Technical Paper 2000-01-0286. Houseman, J., and Hoehn, F. W. (1974). A two-charge engine concept: Hydrogen enrichment. SAE Technical Paper 741169. Karim, G. A. (2002). Hydrogen as spark ignition engine fuel. Chem. Ind. 56:256. Karim, G. A., and Moore, N. P. W. (1990). The production of hydrogen by the partial oxidation of methane in a dual-fuel engine. SAE Technical Paper 901501. Karim, G. A., and Wierzba, I. (1992). Safety measures associated with the operation of engines on various alternative fuels; Reliability Eng. Syst. Safety 37:93. Karim, G. A., Wierzba, I., and Al-Alousi, Y. (1996). Methane-hydrogen mixtures as fuels. Int. J. Hydr. Energy 21:625. Kavtaradze, R. Z., Zeilinger, K., and Zitzler, G. (2005). Ignition delay in a diesel engine utilizing different fuels. High Temp. Apparat. Struct. 43:951. Kumar, M. S., Ramesh, A., and Nagalingam, B. (2003). Use of hydrogen to enhance the performance of a vegetable oil fuelled compression ignition engine. Int. J. Hydr. Energy 28:1143. Lambe, S. M., and Watson, H. C. (1993). Optimizing the design of a hydrogen engine with pilot diesel fuel ignition. Int. J. Veh. Design 14:370. Li, H., and Karim, G. A. (2005). Exhaust emissions from an SI engine operating on gaseous fuel mixtures containing hydrogen. Int. J. Hydr. Energy 30:1491. Livengood, J. C., and Wu, P. C. (1955). Correlation of autoignition phenomenon in internal combustion engines and rapid compression machines. In 5th Symposium (International) on Combustion, pp. 347–56. Liu, Z., and Karim, G. A. (1995). The ignition delay period in dual-fuel engines. SAE Technical Paper 950466. Liu, Z., and Karim, G. A. (1997). Simulation of combustion process in gas-fuelled diesel engine. Proc. Inst. Mech. Eng. A:211. Lu, P.-H., Xie, X.-B., and Lai, M.-C. (2004). Spectral analysis and chemiluminescence imaging of hydrogen addition to HSDI diesel combustion under conventional and low­temperature combustion. SAE Technical Paper 2004-01-2919. Mansour, C., Bounif, A., Aris, A., and Gaillard, F. (2001). Gas-diesel (dual-fuel) modeling in diesel engine environment. Int. J. Therm. Sci. 40:409. McMillian, M. H., and Lawson, S. A. (2006). Experimental and modeling study of hydrogen/ syngas production and particulate emissions from a natural gas-fueled partial oxidation engine. Int. J. Hydr. Energy 31:847.

Combustion of Syngas in Internal Combustion Engines

327

Munshi, S., Nedelcu, C., Harris, J., Edwards, T., Williams, J. R., Lynch, F., Frailey, M. R., Dixon, G., Wayne, S., and Nine, R. (2004). Hydrogen-blended natural gas operation of a heavy-duty turbocharged lean-burn spark ignition engine. SAE Technical Paper 2004-01-2956. Nagalingam, B., Duebel, F., and Schmillen, K. (1983). Perfonnance study using natural gas, hydrogen supplemented natural gas and hydrogen in AVL research engine. Int. J. Hydr. Energy 8:715. Norbeck, J. M., Heffel, J. W., Durbin, T. D., Tabbara, B., Bowden, J. M., and Montano, M. C. (1996). In Hydrogen fuel for surface transportation, 1–26. Society of Automotive Engineers, Warrendale, PA. Parks, F. B. (1976). A single-cylinder engine study of hydrogen rich fuels. SAE Technical Paper 760099. Peschka, W. (1992). Liquid hydrogen—Fuel of the future. Berlin: Springer-Verlag. Poonia, M. P., Ramesh, A., and Gaur, R. R. (1998). Effect of intake air temperature and pilot fuel quantity on the combustion characteristics of a LPG–diesel dual-fuel engine. SAE Technical Paper 982455. Prakash, G., Ramesh, A., and Shaik, A. B. (1999). An approach for estimation of ignition delay in a dual-fuel engine. SAE Technical Paper 1999-01-0232. Raman, V., Hansel. J., Fulton. I., Lynch. F., and Bruderly. D. (1996). Hythane—An ultraclean transportation fuel. In Hydrogen fuel for surface transportation, 47–56. Society of Automotive Engineers. Ramos, J. I. (1989). Internal combustion engine modeling. Hemisphere Publishing Corporation. Rao, A. D., Samuelsen, G. S., Robson, F. L., and Geisbrecht, R. A. (2002). Power plant system configurations for the 21st century. In ASME Int. Gas Turb. Inst., Turbo Expo IGTI, Vol. 1, p. 831. Shudo, T. (2006). An HCCI combustion engine system using on-board reformed gases of methanol with waste heat recovery: Ignition control by hydrogen. Int. J. Veh. Design 41:206. Shudo, T., Nakajima, Y., and Futakuchi, T. (2000). Thermal efficiency analysis in a hydrogen premixed combustion engine. Soc. Automotive Eng. Jap. Rev. 21:177. Shudo, T., and Takahashi, T. (2004). Influence of reformed gas composition on HCCI combustion engine system fueled with DME and H2-CO-CO2 which are onboard-reformed from methanol utilizing engine exhaust heat. Trans. JSME B 70:2663. Sita Rama Raju, A. V., Ramesh, A., and Nagalingam, B. (2000). Effect of hydrogen induction on the performance of a natural-gas-fuelled lean-burn SI engine. J. Inst. Energy 73:143. Sobiesiak, A., Uykur, C., and Ting, D. S.-K. (2002). Hydrogen/oxygen additives influence on premixed iso-octane/air flame. SAE Technical Paper 2002-01-1710. Stebar, R. F., and Parks, F. B. (1974). Emission control with lean operation using hydrogensupplemented fuel. SAE Technical Paper 740187. Swain, M. R., Yusuf, M. J., Dulger, Z., and Swain, M. N. (1993). The effects of hydrogen addition on natural gas engine operation. SAE Technical Paper 932775. Thyagarajan, V., and Babu, M. K. G. (1985). A combustion model for a dual-fuel direct injection diesel engine. In Proceedings of COMODIA Symposium on Diagnostics and Modelling of Combustion in Reciprocating Engines, Tokyo, pp. 607–614. Tomita, E., Kawahara, N., Piao, Z., Fujita, S., and Hamamoto, Y. (2001). Hydrogen combustion and exhaust emissions ignited with diesel oil in a dual-fuel engine. SAE Technical Paper 2001-01-3503. Tunestal, P., Christensen, M., Einewall, P., Andersson, T., Johansson, B., and Jonsson, O. (2002). Hydrogen addition for improved lean burn capability of slow and fast burning natural gas combustion chambers. SAE Technical Paper 2002-01-2686.

328

Synthesis Gas Combustion: Fundamentals and Applications

Upatnieks, A., Mueller, C. J., and Martin, G. C. (2005). The influence of charge-gas dilution and temperature on DI diesel combustion processes using a short-ignition-delay, oxygenated fuel. SAE Technical Paper 2005-01-2088. U.S. Department of Energy, Office of Fossil Energy. (2003). FutureGen—A sequestration and hydrogen research initiative. Available at http://fossil.energy.gov/programs/ powersystems/futuregen/futuregen_factsheet.pdf (accessed December 30, 2006). Vandenborre, H., and Sierens, R. (1996). Greenbus: A hydrogen fuelled city bus. Int. J. Hydr. Energy 21:521. Varde, K. S. (1981). Combustion characteristics of small spark ignition engines using hydrogen supplemented fuel mixtures. SAE Technical Paper 810921. Varde, K. S., and Frame, G. A. (1983). Hydrogen aspiration in a direct injection type diesel engine—Its effects on smoke and other engine performance parameters. Int. J. Hydr. Energy 8:549. Varde, K. S., and Varde, L. K. (1984). Reduction of soot in diesel combustion with hydrogen and different H/C gaseous fuel. Paper presented at the 5th World Hydrogen Energy, Toronto. Verhelst, S., and Sierens, R. (2001a). Aspects concerning the optimisation of a hydrogen fueled engine. Int. J. Hydr. Energy 26:981. Verhelst, S., and Sierens, R. (2001b). Hydrogen engine-specific properties. Int. J. Hydr. Energy 26:987. Zhang, Y., and Boehman, A. L. (2007). Impact of biodiesel on NOX emissions in a common rail direct injection diesel engine. Energy Fuels 21:2003.

Oxide Fuel 11 Solid Cells Using Syngas Robert J. Kee, Huayang Zhu, and Gregory S. Jackson Contents 11.1 Introduction................................................................................................... 330 11.2 Review of Solid Oxide Fuel Cells (SOFCs)................................................... 331 11.2.1 Membrane-Electrode Assemblies...................................................... 332 11.2.1.1 Cathode............................................................................... 333 11.2.1.2 Electrolyte Membrane......................................................... 334 11.2.1.3 Anode.................................................................................. 335 11.2.2 Electrochemistry................................................................................ 337 11.2.2.1 Cell Potential and Overpotentials....................................... 337 11.2.2.2 Ohmic Overpotential.......................................................... 338 11.2.2.3 Activation Overpotentials................................................... 339 11.2.2.4 Concentration Overpotentials............................................. 341 11.2.2.5 Charge Transfer Pathways.................................................. 342 11.2.3 Thermal and Heterogeneous Catalytic Chemistry............................ 343 11.2.4 Porous-Media Transport.................................................................... 345 11.2.5 Effects of Syngas Impurities..............................................................346 11.3 SOFC Materials............................................................................................. 347 11.3.1 Electrolyte Materials......................................................................... 347 11.3.2 Cathode Materials..............................................................................348 11.3.3 Anode Materials................................................................................ 349 11.4 SOFC Stacks and Systems............................................................................. 350 11.4.1 Planar Stacks..................................................................................... 350 11.4.2 Tubular Cells and Stacks................................................................... 351 11.4.3 Systems Integration........................................................................... 352 11.4.4 Co-Generation of Syngas................................................................... 353 11.5 SOFC Modeling............................................................................................. 354 11.5.1 Modeling Approach........................................................................... 355 11.5.2 Performance with Syngas from CPOx of Hydrocarbons.................. 356 11.5.3 Performance with Syngas from Steam Reforming of Hydrocarbons.....361 11.5.4 Performance with Syngas from Coal and Biomass Gasification....... 365 11.6 Conclusions.................................................................................................... 367 Acknowledgments................................................................................................... 369 References............................................................................................................... 369 329

330

Synthesis Gas Combustion: Fundamentals and Applications

11.1 Introduction As the pressure rises to implement strategies to minimize emissions of CO2 and other gases contributing to global warming, fossil fuel power plants that operate with high thermal efficiencies and offer the potential for effective CO2 sequestration can be expected to gain in importance. CO2 sequestration (or carbon capture) will be particularly of great importance for coal-based power plants, and numerous power plant designs based on converting coal to syngas are at various stages of development with the potential for implementing effective carbon capture. To this end, solid oxide fuel cells (SOFCs) offer a promising approach to convert coal-derived syngas directly to electrical power. They are ideally suited for implementing carbon capture because they pull O2 out of the airflow across a solid oxide ion (O2–) conducting membrane. The fuel oxidation products (CO2 and H2O) are thus separated from dilution by the N2 in the airflow, and the concentrated CO2 and H2O flow is more amenable to effective CO2 sequestration. The potential of SOFCs has led to simulation studies on integrating SOFCs with coal gasification and carbon sequestration for high-efficiency, nearzero-emissions power plants (Dijkstra and Jansen, 2004; Moller et al., 2004; Verma et al., 2006; Araki et al., 2007a; Trembly et al., 2007c). While such centralized SOFCbased power plants operating on coal (or even biomass)-derived syngas still require many years of development effort, assessing SOFCs operation in such future plants with syngas feeds is benefiting from shorter-time-scale implementation of SOFCs operating on syngas from hydrocarbon reformers for small-scale portable and backup power. This chapter will review how SOFCs perform on syngas fuel feeds. Unlike low-temperature polymer-electrolyte fuel cells (PEFCs), which require high-purity H2 with less than 100 ppm CO, SOFCs with their high temperature operation (above 600°C) are ideally suited for operation on syngas, derived from coal gasification (Gemmen and Trembly, 2006; Verma et al., 2006; Trembly et al., 2007c), biomass gasification (Aravind et al., 2005; Ouweltjes et al., 2006; Aloui and Halouani, 2007), or reforming hydrocarbons (Finnerty et al., 2000; Gupta et al., 2006c; Tanaka et al., 2006). Furthermore, with specialized cell architectures, SOFCs have shown the capability to operate on direct hydrocarbon feeds (McIntosh and Gorte, 2004; Lin et al., 2005), including liquid fuels (Murray et al., 2006). However, under such conditions, the hydrocarbon fuel is likely converted to syngas by reforming reactions with either added steam or product H2O before reaching the region where electrochemical oxidation takes place (Zhu et al., 2005; Zhan and Barnett, 2006). As such, even with direct hydrocarbon feeds, SOFCs are likely electrochemically oxidizing syngas, and thus understanding the role of syngas is critical for SOFC design and optimization for high-efficiency performance. SOFCs are the subject of significant research and development throughout the world. Several recent reviews provide comprehensive guides to the current state of the art. These include papers that discuss basic principles (Weber and Ivers-Tiffee, 2004; Kee et al., 2005), electrolyte and electrode materials (Fleig, 2003; Adler, 2004; Atkinson et al., 2004; Jiang and Chan, 2004), and systems development (Ormerod 2003; McIntosh and Gorte, 2004). While SOFCs have begun to break into niche markets for small-scale (less than 10 kW) portable and stationary power applications, their potential for large-scale central power applications (with future CO2

Solid Oxide Fuel Cells Using Syngas

331

sequestration) has not yet been realized primarily due to the need for improved longterm stability and reliability, particularly when operating on fuels that introduce detrimental impurities into the SOFC. These challenges motivate much of the materials and systems research in SOFCs ongoing today throughout the world. Syngas may be produced by alternative processes from a variety of feedstocks, delivering significantly different compositions to the fuel cell. In this chapter, syngas is defined broadly to include at least H2 and CO, but may also include CH4, H2O, CO2, N2, and some trace impurities such as H2S. Syngas produced by the steam reforming of hydrocarbons (outside of the SOFC) usually contains significant amounts of excess steam. Although the steam may be condensed prior to entering the fuel cell, the extra energy required to reheat the dry syngas to SOFC temperatures makes the H2O removal less attractive from a system perspective. Depending upon the reformer temperature, significant levels of CH4 may be incorporated into the syngas. If the syngas is prepared by catalytic partial oxidation of a hydrocarbon using air, the syngas contains high levels of N2. Depending on the parent fuel, the H2/CO ratios can vary significantly. Gasified coal usually contains high CO levels depending on the steamto-oxygen ratio of the gasifier. Gasification processes typically use excess steam, which remains in the syngas. Gasified biomass usually contains high levels of N2. Syngas produced from coal or biomass typically requires significant cleanup before entering the SOFC to remove impurities associated with metals, sulfur, and other elements. Significant research is ongoing to assess how various impurities impact the SOFC performance (Cordiner et al., 2007; Marquez et al., 2007; Trembly et al., 2007a, 2007c; Zha et al., 2007) and how materials and designs can be implemented to minimize the impact of these trace impurities (Cheng et al., 2006a; Gong et al., 2007; Xu et al., 2007; Zha et al., 2007). To explore how syngas impacts SOFC performance, the basic physical and chemical principles responsible for SOFC operation are reviewed first. This chapter first presents a discussion of SOFC chemistry and electrochemistry at the individual cell level. This is followed by a brief discussion of SOFC fuel cell stack (of individual cells) and system architectures. Fuels generally, and syngas particularly, affect the selection and development of SOFC anode materials because the fuel is in direct contact with the anode. Additionally, there can be relevant synergies with electrolyte and cathode materials. Material issues and the complications of impurities in the fuel flow are discussed subsequently in this chapter. Although SOFCs can operate with wide variations in syngas composition, the resulting fuel cell performance depends on the particular syngas composition. This chapter is completed with a presentation of detailed numerical model results that explore how syngas composition impacts SOFC performance.

11.2 Review of Solid Oxide Fuel Cells (SOFCs) Fuel cells convert chemical potential energy (in the fuel and oxidizer species) directly into electrical energy along with some heat—unlike a combustion process that converts chemical energy exclusively into heat. Fuel cells, like combustion, oxidize the fuel species into reaction products, principally H2O and CO2. An individual cell is comprised of a membrane-electrode assembly (MEA) and a current-

332

Synthesis Gas Combustion: Fundamentals and Applications

collection system. The practical operating potential for each MEA is approximately in the range of 0.6 < Ecell < 0.8 V. Thus, individual cells are usually connected electrically in series as a stack such that the electrons required by one cathode are supplied from the anode of the neighboring cell. The stack voltage is the sum of the individual cell voltages. The fuel cell functions as an electrical power supply that can drive an external load. The electrons produced at the most electrically negative anode flow through the load and back to the fuel cell at the most electrically positive cathode. Thus, the electrochemical oxidation of the fuel delivers electrical power to the external load.

11.2.1 Membrane-Electrode Assemblies The heart of a fuel cell, the membrane-electrode assembly (MEA), is illustrated in Figure 11.1 for an SOFC. An SOFC MEA includes the following: (1) a dense electrolyte membrane that conducts O2– ions, (2) an anode (negative electrode) where fuel diffuses through a porous support layer to an electrochemically active functional layer where the fuel is oxidized, and (3) a cathode (positive electrode) where O2 diffuses through a porous layer to the electrochemically active region for reduction. Figure  11.1 illustrates a single planar anode–supported SOFC MEA section, and the annotations show some of the underlying physical and chemical processes. The electrolyte membrane is a dense ceramic that conducts O2– ions, but is impervious to gas transport and electronic current. The electrolyte separates the two electrodes, Cathode interconnect Air

e–

i

O2–

2–

CH 4

H2O, CO2

H 2, CO

H2O

H2

H2

Fuel

e–

e– H2, CO

Anode interconnect

H2O, CO2

Fuel channel

H2O

Anode support

O

Electrolyte Anode fuctional layer

e– O2

Cathode

Load

Air channel

+

e–

e–

–

Figure 11.1  Section of a planar membrane-electrode assembly (MEA), illustrating essential fluid, chemical, and electrochemical processes.

Solid Oxide Fuel Cells Using Syngas

333

the cathode with the oxidizer flow (usually air) and the anode with the fuel flow. The MEA is usually on the order of a millimeter thick, with one of the electrodes or the electrolyte providing the structural support and the other two components being very thin (tens of micrometers). The MEA is sandwiched between metallic interconnect structures that also form fuel and air flow channels. The ribs that separate channels provide electrical contact to the electrode. Such electrical contact allows for the individual cells to be placed on top of each other into what is generally referred to as a fuel cell stack. 11.2.1.1 Cathode The operation of an MEA can be in large part explained by following the flow of charge, beginning with the electron flow from the load, through the interconnect and into a porous cathode structure. The cathode’s purpose is to facilitate the electrochemical reduction of O2 to form O2– ions that can cross the electrolyte membrane. The cathode reaction may be stated globally as

O2(g,c) + 4e –(c) ↔ 2O2–(e)

(R1)

The nomenclature O2(g,c) implies oxygen in the gas phase (g) on the cathode side (c) of the electrolyte membrane. The electrons e –(c) come from a solid electronconducting phase in the cathode, which usually serves also as the cathode electrocatalyst to facilitate O2 adsorption and electrochemical reduction. The oxide ions O2–(e) are at crystal lattice sites within the electrolyte phase (e). R1 is referred to as a charge transfer reaction because electrical charge passes from one phase, the cathode electrocatalyst (c), to another phase, the electrolyte (e). If these phases are at different electric potentials, as they are during SOFC operation, then the flow of charge across the voltage difference provides the ability to do work. Figure  11.2 shows a microscopic view of the MEA structure surrounding the dense electrolyte. A porous matrix, often composed of two ceramic materials, forms the cathode. One material serves as the electrocatalyst and as an electron conductor (e.g., strontium-doped lanthanum manganate [LSM]), and the other functions as an O2– ion conductor, often the same material as the electrolyte (e.g., yttria-stabilized zirconia [YSZ]). Significant research has been undertaken to develop cathode materials that have mixed ionic and electronic conductivity (MIEC) and can serve the role of both materials in the more conventional composite cathodes. Both two-material composites and MIEC materials for cathodes have been reviewed thoroughly (Adler, 2004). For the porous matrix of the cathode, characteristic primary particle sizes are usually less than 1 µm, with typical pore dimensions less than 1 µm. For the more conventional composite cathodes, charge transfer chemistry (e.g., R1) proceeds at the so-called three-phase boundary (TPB). The TPB is formed at the intersections of the electron-conducting electrode material, the O2–-conducting electrolyte material, and the gas phase. As illustrated in Figure 11.2, there can be parallel flow of electrons and ions throughout the entire porous cathode structure, with the electron flux being higher near the interconnect structure and the O2– flux being higher near the electrolyte-membrane interface. Depending on material properties, the depth of the cathode participating in charge transfer is confined to a few tens of

334

Synthesis Gas Combustion: Fundamentals and Applications

e–

O2–

e

Air

e–

O2–

Cathode

O2–

e– Ni

Reform

e–

e–

–

Fuel

Fuel

–

e

e–

Fuel

Fuel O2–

Charge Transfer

e– –

H2O H2 H2O CO H2 CO2 CO WGS CH4

Ni Anode

–

H2

–

Air

e

H2O

H

OH

O2–

e–

–

H

O2– Air

Air

e

YSZ

e–

O2–

OH

e–

YSZ Electrolyte – + – ++ Ni –– + + –– + Anode – –

OH–

+

O2

H

Oxygen reduction

O– – ++ O + + O– O– O– e– O–+ +++ + – + O O– ++ e– + + + O2– O2– + Mixed Ionic-Electronic Conductor (MIEC) O2

e–

Fuel e–

Anode functional layer Electrolyte

e–

Anode support layer e–

Load

Figure 11.2  Microscopic representation of an MEA in the vicinity of the dense electrolyte.

micrometers adjacent to the dense electrolyte (Zhao and Virkar, 2005; DeCaluwe et al., 2008; Zhu and Kee, 2008). Outside this electrochemically active region, charge transport is primarily electron flow. Because the dense SOFC electrolyte is electrically insulating, only O2– ions can enter the electrolyte membrane. 11.2.1.2 Electrolyte Membrane SOFC electrolytes are O2– ion conductors, which ideally are electrically insulating and impervious to gas flow. This impermeability to gases keeps the cathode airflow (and the significant N2 dilution) separated from the anode fuel and the resulting oxidation product flow on the anode side. This makes SOFCs very attractive for future power plant designs that implement CO2 sequestration (Verma et al., 2006; Araki et al., 2007b). Any gas leakage through pinholes or other defects in the membrane allows mixing of fuel and oxidizer, and possibly combustion. Combustion deleteriously competes with the charge transfer reactions that convert chemical energy to electrical work, and thus reduces the efficiency of the cell. The excess heat release of combustion can also cause significant damage to the MEA architecture. Thus, it is important that fabrication processes minimize pinhole leaks through the electrolyte membrane. SOFC-electrolyte materials are generally a doped crystal lattice, with doping causing oxygen vacancies. The most common electrolyte material, YSZ, which is used almost exclusively at this point by commercial developers, provides conduction by having the zirconia with its Zr4+ doped with yttria Y3+ to provide stable oxide vacancies. These vacancies provide pathways for O2– transport at adequately high temperatures (above 700°C) via a lattice-hopping mechanism, where the oxygen

Solid Oxide Fuel Cells Using Syngas

335

moves on to vacancies. However, the relatively high resistance for this O2– conduction requires that the electrolytes be kept as thin as possible, and trade-offs in electrolyte fabrication must balance the need for pinhole-free, structurally sound membranes (by increasing thickness) with the desire for low resistance (by decreasing thickness). This trade-off for YSZ electrolytes leads to working SOFC designs with electrolyte thicknesses ranging from 8 to 20 µm. To provide better O2– transport at temperatures less than 700°C, other electrolyte materials have been investigated extensively, including gadolinia-doped ceria (GDC) (Hu et al., 2004) or perovskites such as LaGaO3 (Ishihara et al., 2006). Longterm durability issues have not yet been resolved for these alternative electrolyte materials. In any electrolyte material, an electrochemical potential gradient drives O2– transport from cathode toward anode, and the electrolyte material must be stable over the range of electrochemical potentials. 11.2.1.3 Anode For most SOFC MEA architectures, the anode serves as the support structure and therefore is typically much thicker than the electrolyte and the cathode. With its increased thickness (up to 1 mm), the anode is often composed of both a thin functional layer adjacent to the electrolyte and a thicker support layer, which interfaces with the anode fuel flow. Both layers are porous to allow for reactant gas transport into and product gas transport out of the anode structure. The thin functional layer microarchitecture is formulated to enhance electrochemical activity, whereas the thick support layer microarchitecture is designed for structural compatibility, high electronic conductivity, and if necessary, high activity for reforming direct hydrocarbon feeds. Like the cathode, the anode functional layer is typically a porous composite of an electrically conducting, electrocatalyst material and an O2–-ion-conducting material. For many commercially developed SOFCs, the electrocatalyst is nickel (Ni), and the O2–-ion-conducting material is YSZ, which provides a chemically stable interface with the electrolyte membrane. Many other materials are being considered for anode electrocatalyst, including conductive ceramics (Madsen and Barnett, 2005; Haag et al., 2008) and metal/ceria composites (McIntosh and Gorte, 2004; CostaNunes et al., 2005). As stated above, the functional layer microarchitecture should be designed for high electrochemical activity. As with composite cathodes, the anode functional layer promotes charge transfer reaction at three-phase regions formed by the intersection of electrode (electron-conducting), electrolyte (ion-conducting), and gas phases. Electrons are transferred from the O2– ions into the electrode phase. As in the cathode structure, the depth of the region where charge transfer reactions occur is usually confined to a few tens of microns of the porous material near the dense electrolyte (Zhao and Virkar, 2005; Schneider et al., 2006; DeCaluwe et al., 2008; Zhu and Kee, 2008). Charge transfer in the anode, for operation with syngas or internally reformed hydrocarbons, is almost certainly dominated by the electrochemical oxidation of H2, stated globally as

H2(g,a) + O2–(e) ↔ H2O(g,a) + 2e –(a)

(R2)

336

Synthesis Gas Combustion: Fundamentals and Applications

However, CO can be directly oxidized electrochemically, and there is some evidence that small hydrocarbons can be as well. Again, stated globally,

CO(g,a) + O2–(e) ↔ CO2(g,a) + 2e –(a)

(R3)



CH4(g,a) + 4O2–(e) ↔ CO2(g,a) + H2O(g,a) + 8e –(a)

(R4)

The anode gases (g,a) are all in the pore spaces, and the O2– ions and electrons are in the electrolyte phase (e) and the anode electrocatalyst phase (a), respectively. Outside of the region where significant charge transfer takes place, the porous anode-support-layer microarchitecture must provide good electron conductivity since all available charge has transferred from the ion-conducting phase to the electron-conducting anode phase. Thus, far from the dense electrolyte, there is no electrochemical need for the ion-conducting phase in the porous matrix. However, to maintain thermochemical and material compatibility, it is usually desirable to fabricate the anode support layer with a material combination similar to that of the functional layer. In addition to providing mechanical support for the thin cathode, electrolyte, and functional layer, the anode support layer must play other important roles. It must provide good electrical conductivity, minimizing Ohmic losses between the functional layer and the current-collecting interconnect. It must also offer little resistance to diffusive gas transport for reactants from the fuel channel toward the functional layer and for product species transported back toward the flow channel. These design objectives are usually accomplished by using relatively larger particle sizes and pore spaces than those in catalyst support layers in nonelectrochemical systems. The anode support layer also plays an important role in promoting heterogeneous catalytic chemistry. The nickel in a Ni/YSZ composite serves as a catalyst that promotes steam-reforming and water-gas-shift reactions. These reactions can be represented globally as

CH4 + H2O ↔ CO + 3H2   ΔHreac ≈ 206 kJ/mol

(R5)



CO + H2O ↔ CO + H2

(R6)

ΔHreac ≈ –41 kJ/mol

As the electrochemically produced steam from R2 diffuses out toward the flow channel, it encounters small hydrocarbons and CO diffusing toward the functional layer. At high temperature and in the presence of an appropriate catalyst (e.g., Ni), the heterogeneous chemistry produces H2. This benefits SOFC performance because H2 is much more electrochemically active than the original fuel species. The highly endothermic steam-reforming process requires high temperature (above 800°C for CH4 but less for higher hydrocarbons), but the water-gas-shift can proceed at relatively lower temperatures (below 600°C) on a Ni catalyst. When small hydrocarbons such as methane are in the syngas, the endothermic reforming can be used to control MEA temperature.

Solid Oxide Fuel Cells Using Syngas

337

11.2.2 Electrochemistry All the processes discussed qualitatively in the previous section can be evaluated quantitatively. Indeed, such quantitative understanding is important to system design and development. Each of the physical and chemical processes that are responsible for fuel cell functioning comes at the cost of expending some available chemical energy. As illustrated in Figures 11.1 and 11.2, the negatively charged ions and electrons flow from the positive terminal (cathode) toward the negative terminal (anode). Chemical energy is required to drive negatively charged electrons toward a negative terminal. The needed energy that is extracted from the fuel and oxidizer manifests itself as an efficiency loss in the sense that this energy is not available to the external load. From the viewpoint of the load, the fuel cell appears as a voltage source with electrons flowing from negative terminal, through the load, and toward the positive terminal. 11.2.2.1 Cell Potential and Overpotentials Under open-circuit conditions, the MEA cell potential Ecell is a maximum at the thermodynamic reversible potential Erev (also known as the Nernst potential). Erev can be evaluated as

Erev = −

0 ∆Greac RT − ln ∏ pkνk ne F ne F

(11.1)

0 is the change in standard-stated free energy associated In this expression, ΔGreac with the global oxidation reaction, ne is the moles of electrons transferred per mole of global reaction, F = 96485 (Coulombs/gmol of electrons) is Faraday’s constant, R is the universal gas constant, T is the temperature, pk is a species partial pressure (measured in atmospheres), and νk are stoichiometric coefficients in the global reaction (for reactants νk < 0). To make this concrete, the simplest case of H2 oxidation is shown here.­



H2(g,a) + 1/2O2(g,c) ↔ H2O(g,a)

(R6)

In this case,­­

0 ∆GR6 = µ H0 2O,a − µ 0H2 ,a − 1 2 µ O0 2 ,c

(11.2)

where μ0k are species standard-state chemical potentials at T. The reversible potential is given as

Erev = −

0 ∆GR6 RT  pH2O,a  − ln   ne F 2 F  pH2 ,a pO1/22,c 

(11.3)

The partial pressures in Equation 11.3 must be evaluated in the appropriate flow channel. Under typical SOFC operating conditions, Erev ranges from 1.0 to 1.2 V, 0 0 0 = ∆H R6 − T ∆SR6 depending on the fuel and product partial pressures. Since ∆GR6

338

Synthesis Gas Combustion: Fundamentals and Applications

0 and ∆SR6 ≈ –55.1 J/gmol·K at typical operating temperatures between 700 and 900°C, Erev decreases about 0.029 V for every 100°C increase in temperature. However, as discussed below, this small decrease in Erev is more than offset by decreases in voltage losses (i.e., overpotentials) with increasing temperature when reasonable current is drawn through the SOFC. While delivering power, an SOFC cannot achieve its reversible potential. Electric potential loss required to overcome various internal barriers to the global reaction is referred to as overpotential, η, which is typically calculated or measured as a function of the current density i in terms of amps per unit geometric area of active membrane. There are three types of overpotential: Ohmic, activation, and concentration. Thus, the actual operating cell voltage Ecell can be expressed as the difference of Erev and the sum of overpotentials as



Ecell = Erev − ηOhm (i ) − ηact,a (i ) − ηact,c (i ) − ηconc,a (i ) − ηconc,c (i )

(11.4)

All η(i) increase in magnitude with increasing i. The absolute sign around ηact,c is discussed below in Section 11.2.2.3. The overpotentials represent losses of free chemical energy in the fuel that cannot be delivered as electrical work to the external load. However, because the losses manifest themselves as heat, they can serve to maintain necessary operating temperatures in the range of 800°C, particularly for smaller stacks with higher external surface-to-volume ratios. SOFC systems are capable of conversion efficiencies over 50%, meaning that more than half of the chemical energy can be converted to electricity. The remaining energy (i.e., the inefficiencies) heats the system. In small systems (<1 kW), the challenge is usually to design sufficient insulation to retain the heat. In large systems the challenge is usually to remove the heat to avoid overheating interior sections of the stack. 11.2.2.2 Ohmic Overpotential The Ohmic overpotential ηOhm (i) is equal to iRbulk,tot, where Rbulk,tot is the area­specific resistance to driving O2– ions through the dense electrolyte and the electrochemically active regions of the anode and cathode. There are also contributions to the Ohmic overpotential associated with the electronic resistance of interconnect materials and materials interfaces, but in well-designed SOFCs, the resistance to O2– flow dominates the overall Ohmic resistance. The overall Ohmic overpotential can thus be written as

ηOhm (i ) = i ( Rbulk,e + Rbulk,a + Rbulk,c ) = iRbulk,tot

(11.5)

Rbulk,e typically dominates Rbulk,tot, and it depends on the electrolyte thickness δe and the effective thickness of the O2– conduction into the anode and cathode functional layers. It is inversely proportional to the O2– ion conductivity σO, which depends strongly on temperature and is usually expressed as

σ O = σ O0 exp (− EO RT )

(11.6)

Solid Oxide Fuel Cells Using Syngas

339

For the most common electrolyte YSZ, σO0 ≈ 3.6 × 105 S·K/cm and EO ≈ 8.0 × 10 J/mol (Sasaki and Maier, 2000). To achieve low ηOhm, YSZ-based electrolytes of thicknesses approximately 10 microns or more require SOFC operating temperatures > 750°C to achieve acceptably low Rbulk,e for high power densities (above 0.5 W/cm2 of electrolyte membrane). Much of the alternative electrolyte material research is directed at providing lower Rbulk,e such that lower-temperature (below 700°C) operation provides reasonable current densities. 4

11.2.2.3 Activation Overpotentials Activation overpotentials ηact,a (i) and ηact,c (i) are associated with the energy needed to drive charge transfer reactions (R1, R2, and R3). The relationship between i and ηact is often described by the Butler-Volmer relationship, which gives charge transfer reaction rates in terms of ηact.

  α Fη   α F η  i = i0 exp  a act  − exp − c act   RT   RT  

(11.7)

Unlike thermal reactions, where rates depend only on temperature and species activities, charge transfer reaction rates also depend on electric potential differences between phases in which the participating species reside. The Butler-Volmer equation expresses a charge transfer reaction’s rate of progress as the difference in forward and reverse rates. The exchange current density i0 depends on temperature (generally in an Arrhenius form with a thermal activation energy barrier) and the chemical activities (concentrations) of the participating species. i0 is the rate at which the charge transfer reaction proceeds in both the anodic and cathodic direction at the equilibrium electric potential. The first term on the right-hand side represents the rate of progress in the reaction’s anodic direction (i.e., the direction that produces electrons). The second term represents the rate of progress in the reaction’s cathodic direction (i.e., the direction that consumes electrons). The potential energy diagram in Figure 11.3 illustrates a global charge transfer reaction. The potential energy surface on the left represents the reactants, H2 in the gas phase and an O2– ion within the electrolyte at the anode interface. The gas phase is presumed to be electrically neutral, carrying no charge. The potential energy surface on the right represents the reaction products, which include electrons within the anode. The two potential energy surfaces are shown on the right representing two different anode voltages. The dashed curve represents the product surface at the equilibrium electrical potential difference, defined as Eaeq = (ϕa – ϕe,a)eq, where ϕa is the electric potential of the anode electrocatalyst material and ϕe,a is the electric potential of the electrolyte material at the electrolyte-anode interface. At Eaeq the reaction is proceeding in equal and opposite rates i0 = ia = ic in the anodic and cathodic directions, but delivering no net current i = 0. When the electric potential of the anode (the negative electrode) is increased (i.e., making it less negative) the potential energy surface is lowered, which reduces the barrier for current to flow in the anodic direction. Thus, increasing the anode electric potential makes it easier for negatively charged electrons to be produced at the negative electrode. Increasing the

340

Synthesis Gas Combustion: Fundamentals and Applications

Free Energy, G

Anodic ia

Cathodic ic

Eaeq

αaFηact

Increasing φa Ea > Eaeq Electrolyte H2 + O2–

Fηact

Ea

Andoe H2O + 2e– Reaction Coordinate

Figure 11.3  Potential energy surface to assist visualizing aspects of the Butler-Volmer equation.

anode electric potential at the electrolyte interface reduces the overall cell potential, and thus provides the basis for evaluating the anode activation overpotential, ηact,a = Ea – Eaeq. The anodic symmetry parameter αa, which appears in Equation 11.7, measures the change in the anodic barrier height relative to the activation overpotential. The cathodic symmetry parameter αc (not shown in Figure 11.3 but in Equation 11.7) relates to the barrier height for the reverse cathodic direction reaction. For an elementary charge transfer reaction αa + αc =1 and both symmetry parameters are generally close to 1/2. For global reactions, however, the symmetry parameters can be much different and their sum is not constrained to unity. The Butler-Volmer equation (Equation 11.7) provides a quantitative relationship between the ηact and the net charge transfer rate i. When ηact,a > 0 the reaction proceeds in the anodic direction, producing electrons in the anode. When ηact,a < 0 the reaction proceeds in the cathodic direction, consuming electrons and producing O2– ions in the electrolyte. At the cathode, where electrons are consumed, the net charge transfer reaction must be proceeding in the cathodic direction. This is the reason for the absolute value around ηact,c in Equation 11.4. For typical SOFC operating conditions and materials, Ni/YSZ composite anodes and LSM/YSZ composite cathodes |ηact,c| > ηact,a. This is because the exchange current density i0 for the cathode is smaller than that of the anode. i0 in Equation 11.7 increases strongly with temperature for H2 electrochemical oxidation (R2) and even more so for O2 reduction (R1). Thus, both ηact,a and ηact,c decrease with increasing temperature. The decrease particularly in ηact,c contributes to higher Ecell and thus higher power densities Ecell · i with increasing temperature. Similarly i0, which are related to forward and reverse reaction rates (Zhu et al., 2005), tends to increase with reactant pk, and thus higher-pressure operation can be another means of reducing both ηact,a and ηact,c. However, the impact of increasing pk tends to be less significant than increasing temperature for H2 oxidation and O2 reduction. Much more can be said about charge transfer chemistry, but the intent here is simply to explain the basic principles. Using the Butler-Volmer equation quantitatively requires further information about i0, including functional dependence on species activities and materials electrocatalytic activity, and available three-phase boundary

341

Solid Oxide Fuel Cells Using Syngas

available for charge transfer. The functional forms of these relationships are guided by theory and experiment, with specific parameters established empirically using experimental data. More information and quantitative functional relationships can be found in the literature (Kee et al., 2005; Zhu et al., 2005; Bessler et al., 2007; Zhu and Kee, 2008). 11.2.2.4 Concentration Overpotentials Concentration overpotentials ηconc are associated with gas-phase transport within the porous electrode structures. For example, H2 concentrations are lower at the anodeelectrolyte interface than they are in the fuel channel. This is because the flux of H2 consumed electrochemically at the interface is driven by molecular diffusion and concentration gradients between the channel and the electrolyte. Similarly, product species concentrations (e.g., H2O) are higher near the dense electrolyte and lower in the fuel channel. Consider evaluating the reversible cell voltage Erev from Equation 11.1 using the species concentration (or partial pressures pk) in the fuel channel and at the electrolyte interface. Erev is lower at the electrolyte interface, where fuel pk are lower and product pk higher. The difference in Erev at the channel flow conditions and the electrolyte interface is ηconc,a for the anode and ηconc,c for the cathode. This can be expressed as follows for H2 electrochemical oxidation in the anode and O2 reduction in the cathode. RT  pH2 ,a pH2O,a,e  ηconc,a = (11.8) ln   2 F  pH2 ,a,e pH2O,a  ηconc,c =



RT  pO2 ,c  ln   4 F  pO2 ,c,e 

(11.9)

In the above expressions, pk,a and pk,c refer to partial pressures in the anode and cathode flow channels and pk,a,e and pk,c,e represent the partial pressures in the electrochemically active regions of the anode and cathode, respectively. As i increases, more fuel is consumed and products formed, which increases the concentration or pk gradients in the porous electrodes and hence ηconc. The rate at which the ηconc increase with i depends on the effective diffusion coefficients through the porous electrode matrices. The diffusive fluxes are functions of electrode thickness, porosity, and tortuosity, and to a first approximation are proportional to pk,a – pk,a,e and pk,c – pk,c,e for the anode and cathode, respectively. Thus, with an effective mixture diffusive coefficient Dk,eff for the relevant species in each porous electrode, ηconc can be written as follows:



ηconc,a = −

   RT δai RT  RT δai ln 1 −  1 +  2 F  2 FDH2 ,eff,a pH2 ,a   2 FDH2O,eff,a pH2O,a 

(11.10)

RT  RT δci (1 − XO2 ,c )  ln 1 −  4 F  4 FDO2 ,eff,c pO2 ,c 

(11.11)

ηconc,c = −

342

Synthesis Gas Combustion: Fundamentals and Applications

where XO2,c is the O2 mole fraction in the cathode flow, and δa and δc are the thickness of the anode and cathode, respectively. For typical SOFC-MEA structures with the thick anode support layers, ηconc,a > ηconc,c for expected values of i. Thus in general, the dominant overpotentials in typical SOFC operation are ηOhm, ηact,c, and ηconc,a. Although T appears in the numerators of Equations 11.10 and 11.11, the strong dependence of Dk,eff on Tn (where typically n > 1.5; Kee et al., 2003) results in decreases in ηconc,a with increasing temperature. Thus, like the other dominant overpotentials, ηact,c and ηOhm, ηconc,a also decreases with temperature, resulting in a general higher Ecell and thus higher power density Ecell·i with increasing temperature. The challenge, however, with operating at higher temperature involves material stability and detrimental solid-state reactions that can increase SOFC degradation and decrease performance over time. These trade-offs and the limitations in particular on the stability of interfaces for sealing and for current collection tend to set limits on operating temperature, ranging up to 1000°C, depending on the particular SOFCMEA architecture. 11.2.2.5 Charge Transfer Pathways Within SOFC anode-functional-layer pores, a mixture of H2, CO, and possibly CH4 is available to participate in charge transfer electrochemistry. It has been assumed in modeling studies, however, that H2 dominates the electrochemistry and R5 and R6 convert the CO and CH4 into H2 for more rapid electrochemistry (Zhu et al., 2005; Gupta et al., 2006c). In Ni/YSZ anodes, the H2 charge transfer rate is double that of CO at 750°C, and it is more than triple that of CO at 1000°C (Holtappels et al., 1999; Matsuzaki and Yasuda, 2000). The significantly lower CO charge transfer rate is attributed to slow CO surface diffusion on Ni to the three-phase boundary. Multiple studies have shown that fuel mixtures of H2 and CO, with as high as 75% CO, yield comparable SOFC polarization characterizations to pure H2 (Weber et al., 2002; Jiang and Virkar, 2003; Sukeshini et al., 2006). This result is attributed to the catalytic water-gas-shift reaction that rapidly converts CO to H2, using H2O produced from the H2 electrochemical oxidation. Direct CH4 charge transfer has been shown to be quite slow in Ni/YSZ systems (Sukeshini et al., 2005), and the high ηact,a associated with the direct CH4 electrochemical oxidation suggest the unimportance of this reaction in the presence of H2 and CO. However, with Ni/YSZ-anode-supported cells, CH4 is largely reformed by H2O to CO and H2 before reaching the three-phase boundaries in the anode functional layer (Weber et al., 2002; Zhu et al., 2005; Zhu and Kee, 2006a). There are good reasons to seek alternative materials to replace Ni/YSZ anodes. Ni promotes carbon deposits from hydrocarbon fuels, its catalytic activity is reduced by sulfur, and it can be oxidized to NiO at high overpotentials. Thus, it is important to understand charge transfer pathways on alternative anode materials. One alternative system involves simply replacing YSZ with gadolium-doped ceria (GDC) in both the anode and electrolyte (Baron et al., 2004). Results from this and similar studies show that CO fuel fractions above 10% reduce performance compared to pure H2. As little as 5% CH4 mixed in the H2 causes deleterious carbon deposits. However, other recent results with this anode using hydrocarbon feeds suggest that

Solid Oxide Fuel Cells Using Syngas

343

detrimental performance can be avoided with a ceria-based electrolyte and careful anode design (Liu et al., 2002; Zhan et al., 2004). The search for completely Ni-free anodes has led to the development of ceriabased anodes with presumably nonactive metals such as Cu or Au. These anodes, wherein ceria provides the active surface, show improved performance with carbonaceous fuels and avoid the tendency of Ni-based anodes to promote the carbon deposit formation (Kim et al., 2001; He et al., 2003; Krishnan et al., 2004). The active ceria component has mixed ionic and electronic conductivity (MIEC) above 700°C. The charge transfer reactions occur through a redox cycle involving the oxidation of the fuel species with the removal of oxygen ions from the ceria, which itself is converted from Ce4+ to Ce3+. The associated oxygen vacancy is refilled by ions supplied from the electrolyte, but a clear description of the charge transfer processes in MIEC materials remains uncertain (Fleig, 2005). Exploring alternative anode material systems to optimized charge transfer reactions for syngas as well as other carbonaceous fuels is an area of active research, and the following section provides further discussion.

11.2.3 Thermal and Heterogeneous Catalytic Chemistry Thermal chemistry plays several important roles in SOFC technology. The most important role concerns catalytic chemistry, primarily in the anode support layer. Gas-phase chemistry usually plays a relatively small role with syngas fuels, but a potentially more important role with hydrocarbon fuels. Particularly with higher hydrocarbons, deleterious carbon deposits can be formed via homogeneous processes leading to large polyaromatic compounds or via heterogeneous routes leading to ­graphitic-like coke formation on catalyst surfaces. This encourages the use of upstream hydrocarbon reforming or partial oxidation to syngas for many SOFC applications. As discussed earlier, catalytic steam reforming and water-gas-shift processes are important in converting fuel species to H2. These processes reduce the need for less effective direct electrochemical oxidation of CO and CH4 (or other hydrocarbons). Electrochemical oxidation of CO and CH4 requires high activation overpotentials (Sukeshini et al., 2005), which reduces cell efficiency. R5 and R6 express the global processes of reacting CH4 and CO with H2O to form H2. These global reactions actually represent the catalytic chemistry, which contains a large number of elementary steps as expressed with detailed reaction mechanisms (Hecht et al., 2005; Zhu et al., 2005; Janardhanan and Deutschmann, 2006; Bessler et al., 2007). Detailed reaction mechanisms capture the functional behavior of the reforming chemistry with respect to operating conditions much better than the global reactions. For example, detailed mechanisms can predict the observed benefits to adding small amounts of air to the fuel stream to promote on-anode catalytic partial oxidation (Zhan et al., 2004). In such a case, the detailed surface mechanism can represent the competition on the catalytic surface between partial oxidation and steam-reforming reactions. To date, detailed Ni/YSZ surface mechanisms have only been developed for chemistry for CH4 and syngas, and mechanisms for

344

Synthesis Gas Combustion: Fundamentals and Applications

higher hydrocarbons on any electrode material remain a challenge for the SOFC and catalytic reactor modeling communities. For syngas, with or without some CH4, gas-phase chemistry can usually be neglected. At temperatures below 900°C the reaction rates to form major species are very small (Gupta et al., 2006b). However, if higher hydrocarbons are included, homogeneous chemistry can become more important, possibly leading to carbon deposits (Gupta et al., 2006c). Within the SOFC, catalytic surface chemistry dominates, but the relatively long residence times at high temperature in noncatalytic inlet fuel manifolds and flow distribution systems can lead to some fuel pyrolysis, molecular weight growth, and deposit formation (Gupta et al., 2006a, 2006b). Higher hydrocarbons can undergo dehydrogenation, either homogeneously in the gas phase (Sheng and Dean, 2004; Gupta et al., 2006a, 2006b; Randolph and Dean, 2007; Pomfret et al., 2008) or heterogeneously on catalyst surfaces (McIntosh and Gorte, 2004; McIntosh et al., 2004), resulting in carbon deposition on anode surfaces. Detailed homogeneous chemistry, including molecular weight growth and the formation of polyaromatic hydrocarbons, can involve several thousand elementary reaction steps (Sheng and Dean, 2004; Randolph and Dean, 2007). Detailed heterogeneous chemistry is even more difficult to quantify. Although not entirely quantitative, equilibrium predictions for stable graphite can provide a qualitative guide to where deposits may be problematic (Sasaki and Teraoka, 2003). The predictions shown in Figure  11.4 are based on the assumption that graphitic carbon is the only possible condensed phase. At high temperature (above 900°C), a straight line between the H vertex and the midpoint of the C-O axis divides the deposit and the no-deposit 0.0

C 1.0

0.1

0.9

0.2

CO

1000°C 800°C 600°C

0.5

0.8

0.3

CH4

0.2

400°C

0.1 0.9

0.8

0.6

0.5

0.4

0.3

0.2

0.1

0.7

No deposits

1.0 0.0

0.6

0.4

0.9

H

0.7

1.0

0.7

Do

0.6

de ca

ne

0.5

CH4 S/C = 3

Deposits

0.4

BioDry BioWet

0.8

0.3

CoalDry CoalMov

0.0 O

Figure 11.4  Ternary diagram showing regions of equilibrium carbon formation as functions of elemental composition and temperature. Several particular compositions are identified with symbols. The labels refer to syngas compositions shown in Table 11.5 and discussed in later sections.

345

Solid Oxide Fuel Cells Using Syngas

regions. At lower temperature, the dividing line shows significant curvature and an enlarged deposit region. As illustrated by symbols in Figure 11.4, practical syngas fuel streams can be close to deposit limits. However, within an operating fuel cell, oxygen is crossing the electrolyte membrane and H2O and is being generated via electrochemical oxidation. Thus, deposit propensity decreases along the SOFC anode channel as elemental oxygen in the anode gases increases in the form of H2O as well as CO2.

11.2.4 Porous-Media Transport Figures 11.1 and 11.2 illustrate how gases must be transported through the porous electrodes, between the flow channels and the interface with the dense electrolyte. The composite electrodes are characterized by small, tortuous pores. The steadystate mass conservation for gas-phase species within the pore spaces may be represented as ∇ ⋅ jk = acatWk sk



(11.12)

where acat is the specific catalyst area (effective area per unit volume) and Wk are the species molecular weights. Molar production rates by heterogeneous chemistry are represented by s· k. The mass fluxes jk depend on ordinary and Knudsen molecular diffusion as well as pressure-driven convective fluid flow (Darcy flow). Because the pore size within an electrode can be comparable to the molecular mean-free-path length, there is little probability for gas-gas collisions. Thus, gas-phase chemistry within the pore spaces is usually negligible. The dusty-gas model (DGM) may be used to represent the relationship among the molar concentrations, molar fluxes, concentrations gradients, and the pressure gradient in the SOFC electrodes (Mason and Malinauskas, 1983; Zhu et al., 2005). This implicit constitutive relationship is written as





∑  [ X ][JX −] D[ X ] J  + DJ l ≠k

l

k

T

k e kl

l

k e k , Kn

= −∇ [ X k ] −

[ Xk ] e k,Kn

D

Bg ∇p µ

(11.13)

In this expression, Jk is the molar flux of gas-phase species k, [Xk] are the molar concentrations, [XT] = p/RT is the total molar concentration, and Bg is the permeability. The mass fluxes jk = W k Jk. The mixture viscosity is given as μ, and Dkle and e e Dk,Kn are the effective ordinary and Knudsen diffusion coefficients. Dkle and Dk,Kn can be derived from kinetic theory and geometric properties of the porous media (Zhu et al., 2005; DeCaluwe et al., 2008). The transport equation (Equation 11.13) in conjunction with Equation 11.3 reveals the advantages of increasing pk for increasing E rev and decreasing ηconc, both of which increase E cell for a given i (Equation 11.4). Increasing temperature, on the other hand, can increase diffusion rates and thus reduce ηconc, but it also decreases E rev. The effect of increasing temperature on ηconc is not as strong as the effect of increasing pk , and generally, SOFC operating temperatures are chosen based on electrolyte ion conductivity, catalyst activity, and material stability.

346

Synthesis Gas Combustion: Fundamentals and Applications

11.2.5 Effects of Syngas Impurities Even low levels of fuel impurities can significantly affect SOFC anode performance. For typical Ni-based anodes, this is particularly true for species that can react with the Ni to form new phases with poor electrochemical activity (Trembly et al., 2007a). Syngas formed from natural gas and liquid fuels can contain relatively high levels (several 100 ppm) of sulfur compounds (primarily H2S). Other trace impurities can also appear, particularly when the syngas is formed from liquid fuels. Coal and biomass gasification can contain numerous impurities, including tars and particulates, H2S, alkali and other metals, NH3, and halides. Many impurities arising from the metallic and halide compounds form condensable species that can be removed by low-temperature or hot-gas cleanup processes in central power plant facilities. However, significant efforts for hot cleanup processes are being developed to improve efficiencies for power plants with coal and biomass gasification by avoiding the syngas cooling and reheating expense. Review of these development efforts is outside the scope of this chapter and can be found elsewhere (Higman and van der Burgt, 2003; Trembly et al., 2007c). Some species formed in coal gasification from the metal impurities have high vapor pressures and are difficult to remove by cold or hot cleanup processes. For a specific gas cleanup process, it is important to know how various species for a given impurity element are partitioned between gas and solid phases. Such partitioning depends strongly on the cleanup operating temperature. A recent study explored the partitioning for a range of trace elements in coal gasification products and indicated that three elements, Sb, As, and P, can have significant gas-phase species breakthrough even from hot-gas cleanup, and each of these elements can form secondary phases with Ni that degrade anode performance (Trembly et al., 2007c). These studies are motivated by the possible implementation of large-scale SOFC applications with coal gasification for so-called zero-emissions plants with carbon sequestration (Verma et al., 2006), and these studies are spawning new research on the effects of phosphorous and other elements on the stability of common SOFC anode materials. AsH3 has received attention because of the high As content in some coals and the high volatility of AsH3, and therefore its likelihood of reaching the SOFC anode. It has received attention with typical Ni/YSZ anodes because As can form a Ni5As2 phase that has limited catalytic activity in nonelectrochemical catalytic systems, and recent studies suggest that AsH3 has a long-term impact on Ni/YSZ anode performance due to the formation of secondary Ni arsenide phases (Trembly et al., 2007a). Most alkali metals can be removed with gasification cleanup processes, but if not adequately removed, alkali-metal-containing and halide species from coal- or biomass-derived syngas can also cause irreversible damage to fuel cell anodes. While limited research has been done on the effects of alkali metals on conventional anode performance, it is understood that the presence of these metals (coming from gas-phase impurities from the feed or volatile compounds from solid materials) in the electrochemically active regions of Ni/YSZ anodes can lead to rapid degradation of anode performance (Jensen et al., 2003; Liu et al., 2003). Halide species that can contaminate coal-derived syngas such as HCl cause similarly rapid degradation in Ni/YSZ SOFC anodes (Buchinger et al., 2006; Trembly et al., 2007b). Efforts to

Solid Oxide Fuel Cells Using Syngas

347

develop anode materials and architectures that can minimize degradation due to metallic and halide impurities will be dependent upon how cleanup processes for coal (and biomass) gasification develop. While relatively little research has been published on the effects of many impurities, significant work has been pursued to understand the effects of H2S on various anode materials. Additionally, efforts have been ongoing to develop new anode materials with higher sulfur tolerance (Aguilar et al., 2004; Cheng et al., 2006b). Sulfur concentration as low as a few parts per million can deactivate Ni catalysts, which significantly reduces conventional Ni/YSZ anode performance. This has led to considering alternative electrolyte-phase materials, notably GDC, which decreases the rate of Ni degradation (Ouweltjes et al., 2006). CeO2 nanoparticle addition to Ni/YSZ anodes has been shown to provide dramatically improved sulfur tolerance, and this remains a promising approach for syngas applications with low H2S concentrations (Kurokawa et al., 2007a). Alternatively, electronically conducting perovskite materials can replace Ni/YSZ cermets altogether, and surprisingly good performance has been achieved with perovskite anode electrocatalysts such as strontium-doped lanthanum vanadate (Cheng et al., 2006a) and yttrium-doped strontium titanate (Kurokawa et al., 2007b) when using syngas with relatively high concentrations of H2S. Nonetheless, questions remain as to the long-term durability of these anodes for commercial applications. Thus, despite the practical importance, there is still significant progress yet to be made on both understanding H2S chemistry and developing sulfur-tolerant anode materials and architectures.

11.3 SOFC Materials A great deal of current research and development concentrates on improving the performance and durability of MEA structures, either through improved electrolyte and electrocatalyst materials and material combinations or through the design and fabrication of improved microstructure architecture in the porous electrodes. There are a number of comprehensive materials reviews for the electrolyte (Huang et al., 2001; Brandon et al., 2003; Fergus, 2006), the cathode electrocatalyst (Adler, 2004; Serra et al., 2006), and the anode electrocatalyst (Atkinson et al., 2004; Jiang and Chan, 2004). While this article does not discuss all of the details of materials research involved in developing the numerous electrolyte and electrode materials, the following subsections provide a useful summary for SOFC materials development in the context of syngas fuels.

11.3.1 Electrolyte Materials As stated earlier, the most common electrolyte material is yttrium-stabilized zirconia, or YSZ (i.e., zirconia ZrO2 doped with 8% yttrium). The Y atoms with a +3 valence are substituted in the crystal lattice for Zr atoms with a +4 valence. Such doping in ceramic oxides enhances O2– ion conductivity by creating oxygen ion vacancies in the lattice. To compensate the charge, one oxygen ion must be removed for every two Y atoms added. Therefore, 8% YSZ has a vacancy concentration of 4% in the oxygen sublattice. Since the addition of Y dopants stabilizes the relatively

348

Synthesis Gas Combustion: Fundamentals and Applications

open fluorite crystal structure (which without Y doping is only stable at higher temperatures), the vacancies can move relatively easily through the crystal via hopping. To reduce ion conduction losses, very thin YSZ electrolyte membranes are being developed with thicknesses less than 10 microns (Zhao and Virkar, 2005). Thinner membranes enable lower operating temperatures and higher performance, but they are difficult to fabricate without defects or pinholes that permit gas leakage from the anode to cathode. Alternative electrolyte materials with higher σ are being implemented—notably gadolinia-doped ceria (GDC) (Hu et al., 2004) and samaria-doped ceria (SDC) (Zhang et al., 2006). Ceria-based electrolytes provide higher σO. However, at temperatures above 700°C, doped ceria begins to show appreciable electronic conductivity σe, which causes unacceptable electronic leakage current across the membrane. Good fuel cell performance requires negligibly small electronic leakage current through the electrolyte, forcing all electronic current to flow through the external load. To minimize electron conduction, ceria-based electrolytes are usually designed to operate below 600°C (Zhang et al., 2006). Such reduced operating temperatures influence the choice of anode catalyst materials for carbonaceous fuels such as syngas. Further research has explored alternative doped perovskites for electrolyte materials (Huang et al., 2001), but as of yet, this work has not shown the necessary performance or long-term stability under reactive flow conditions.

11.3.2 Cathode Materials At SOFC operating temperatures, several perovskite materials are adequately active to facilitate electrocatalytic reduction of O2 to O2– (R1). The most commonly used material is strontium-doped lanthanum manganate (LSM), which provides good O2 reduction activity and adequate σe. However, LSM particles must be mixed with YSZ particles to provide adequate three-phase-boundary lengths in the porous cathode because LSM does not provide adequate σO for O2– transport to the dense electrolyte membrane. With LSM/YSZ composite cathodes, the cathode activation overpotentials ηact,c are one of the dominant voltage losses with increasing current density i. This has spurred significant research in other cathode catalyst perovskites, such as lanthanum strontium cobalt ferrite (LSCF), that are mixed ionic-electronic conductors (MIECs). These MIEC perovskites conduct both O2– and electrons. Thus, in principle, they can be used as single-component cathodes. Unfortunately, these materials are subject to solid-state reaction with YSZ electrolytes (Adler, 2004). To impede these deleterious solid-state reactions, a thin barrier layer (e.g., very thin strontium-doped ceria) is required between the YSZ electrolyte and the LSCF cathode. Other cathode materials are being researched to lower ηact,c(i), particularly for temperatures less than 700°C. One promising material is barium strontium cobalt iron oxide (BSCF), which has been shown to deliver high power densities up to 1 W/cm2 at 600°C when the anode has minimal losses with humidified H2 as the fuel (Shao and Haile, 2004). However, like LSCF, long-term chemical stability of BSCF over the range of operating conditions in SOFC applications remains to be proven,

Solid Oxide Fuel Cells Using Syngas

349

and the critical requirements for long-term stability of such cathode materials (with any necessary protective interlayers) have led to slow introduction of these cathode materials into commercial development efforts.

11.3.3 Anode Materials Because fuel contacts the anode, anode materials are most impacted by the fuel composition. The most common SOFC anode is constructed as a porous Ni/YSZ cermet composite. Ni serves as the fuel oxidation catalyst and the electronic conductor, while YSZ provides the O2– conduction. These cermet anodes are fabricated with roughly equal parts (on a volume basis) of Ni, YSZ, and open pores. Very near the YSZ electrolyte-membrane interface in the electrochemically active anode functional layer, a reduction in anode porosity is designed via small primary Ni and YSZ particle sizes less than 1 micron (Jiang and Virkar, 2003; Zhao and Virkar, 2005). The functional layer provides increased TPB area and thus increased electrochemical activity. Away from the electrolyte membrane in the anode support layer, primary particle sizes are typically on the order of 1 to 5 microns, with pore spaces less than 1 micron. The propensity for carbon deposition on Ni-based anodes usually requires the full or partial reformation of hydrocarbon parent fuels to syngas prior to entering the SOFC. However, considering the system complexity of additional reactors and the potential benefits of integrated on-anode reforming, there is considerable research devoted to anode architectures and alternative materials that enable the direct use of hydrocarbon fuels. For example, it may be possible to develop barrier layers that maintain steam-carbon ratios above the coking threshold in the presence of Ni (Lin et al., 2006; Zhu et al., 2006). Other strategies, like recycling a fraction of the anode exhaust, may enable on-anode reforming of hydrocarbons. Adding a bit of O2 (or air) to the fuel stream can promote on-anode partial oxidation and eliminate deposit formation (Zhan et al., 2004; Zhan and Barnett, 2005). In addition to carbon deposits, H2S poisoning, as discussed earlier, and redox stability are significant issues with Ni. Detrimental formation of NiO depends on cell potential as well as local pH2 and pH2O (Steele, 1999). Significant mechanical swelling associated with Ni oxidation causes very significant, and often irreversible, damage to the cell. To avoid electrochemical oxidation of Ni, SOFC systems with Ni/YSZ anodes are usually operated with Ecell > 0.6 V. There are significant efforts to develop anodes that do not use Ni. For example, using combinations of ceria together with metals such as Cu or Au that have minimal C–H bond-breaking activity can be effective (Lu et al., 2003; McIntosh and Gorte, 2004; Gross et al., 2007). The ceria promotes reforming chemistry with minimal hydrocarbon deposits and the metals provide a good electronic conductor. These systems exhibit stable performance with CH4 and higher hydrocarbons, and thus likely with syngas as well. Anticipating a variety of trace fuel species in some syngas streams, alternative fuel-tolerant anodes are likely to play a significant role in future systems.

350

Synthesis Gas Combustion: Fundamentals and Applications

11.4 SOFC Stacks and Systems Just as there are numerous materials combinations, SOFC systems take alternative geometric configurations. Most current systems use either planar or tubular layouts. The advantages of each system configuration depend on the size of the application as well as the composition of the fuel stream.

11.4.1 Planar Stacks Figure 11.5 illustrates a planar counterflow configuration that is modeled after recent development work in Germany (Gubner et al., 2006). The figure is an exploded view showing two cells, with the layers separated for visualization. As assembled, all layers are pressed together and sealed around the edges. In this system, the anodesupported MEA is brazed into a metallic support frame. A thin YSZ electrolyte and LSM/YSZ cathode are on the upper side of the MEA, and the porous Ni/YSZ anode is on the lower side. The MEA assembly is sandwiched between metallic interconnect structures that also form flow channels with current-collection ribs that contact the electrodes, as illustrated in Figure 11.5. Fuel and air flows are introduced through a manifold of circular passages from below and then flow through the anode and cathode side channels, respectively. An exhaust manifold is placed at the opposite end of the channels with the exhaust flow directed downward via similar circular passages. The ribs in the metallic interconnect contact the cathode of the cell below and the anode of the cell above. Thus, the cells are arranged electrically in series to provide a useful voltage from the stack. The electrochemically generated electrical current flows through an external load.

+

Cathode interconnect ribs and channels

i

Load

Anode interconnect ribs and channel Cathode MEA support frame Anode interconnect ribs and channels

–

e–

Fuel exhaust Air exhaust

Fuel feed

Figure 11.5  Exploded view showing two cells of a planar counterflow solid oxide fuel stack modeled after a system developed in Germany at Forschungszentrum Jülich.

351

Solid Oxide Fuel Cells Using Syngas

By exchanging the fuel inlet and exhaust lines, it is relatively straightforward to convert the counterflow layout to a co-flow configuration. Planar stacks are often designed in a cross-flow arrangement. In this case, the interconnects are designed to have the air channels oriented orthogonal to the fuel channels. Many planar systems use a cross-flow arrangement to reduce the design and fabrication challenges in the inlet and exhaust flow manifolding.

11.4.2 Tubular Cells and Stacks Figure 11.6 illustrates an anode-supported tubular cell, a design developed by several research and development groups for both small-scale (Sammes et al., 2005) and large-scale (Lundberg et al., 2003) applications. In anode-supported designs, the thin, dense electrolyte and porous cathode are applied to the outside of the anode tube. However, cathode support is also possible, such as in the systems developed by Siemens (Lundberg et al., 2003). In this case, the thin electrolyte and anode layers are applied to the outside of the cathode tube. Although configured as a tube, the underlying physics and chemistry is the same as in planar systems (such as Figure 11.1). As illustrated in Figure 11.6, current-collection wires are attached to the anode (e.g., Ni) and cathode (e.g., Ag) of the tube (Zhu and Kee, 2007). Other designs with relatively short tubes may rely on the conduction along the length of the anode support layer to collect current from inside the tube.

2–

2

H CO

H2

HC

H2 2

2

CO, H

O H2

CO

O H2

i

O H2

O H2

Load

CO

Shifting + CO2 CO + H2O = H2

Anode support Functional layer Dense electrolye Cathode Current collection

– +

O

O2

2–

O

O2

2–

O

O2

Reforming Fuel + H2O = H2 + CO

i

e–

Figure 11.6  Cutaway view of an anode-supported tubular cell. Fuel flows inside the tube (exposed to the anode) and air surrounds the outside of the tube (exposed to the cathode).

352

Synthesis Gas Combustion: Fundamentals and Applications

Recuperator Catalytic tail-gas combustor Cathode air exhaust Cathode hot-air feed Anode-supported SOFC tube strack System exhaust Hot-zone insulation

Cold-air feed Partially reformed anode feed gas Catalytic partialoxidation reactor (CPOx) Fuel-vapor feed CPOx air feed

Figure 11.7  Illustration of the hot zone in a low-power, tube-stack SOFC system.

Individual tubes, or cells, are usually connected electrically in series to form a fuel cell stack, although the geometry does not resemble a stack as in planar systems (e.g., Figure 11.5). Tubular cells are wired externally, connecting the anode of one tube to the cathode of the next. As illustrated in Figure 11.7, the fuel is usually fed in parallel into a set of tubes through a flow distribution manifold.

11.4.3 Systems Integration Figure 11.7 illustrates a possible hot-zone layout for a small tubular SOFC system. Especially for small systems (less than 1 kW), close thermal integration is needed to maintain temperatures in the range of 750 to 800°C. In the system in Figure 11.7, fuel and air first enter a catalytic partial oxidation (CPOx) reactor, which partially oxidizes a hydrocarbon fuel to syngas using air as the oxidizer. CPOx reactors are usually implemented with a Rh-based catalyst supported on a porous alumina foam (Hickman and Schmidt, 1993; Schmidt et al., 2003). The resulting syngas mixture (also containing some H2O, CO2, and N2) flows to a manifold that feeds the SOFC anode tubes. Unspent fuel (primarily H2 and CO) and reaction products (H2O and CO2) combine with air from the cathode side of the tubes in a tail-gas catalytic combustor. The tail-gas catalyst is used to fully oxidize any unspent fuel. Hot combustion products from the tail-gas combustor are directed to a recuperator. Figure 11.7 also shows a recuperator for exhaust heat recovery for preheating the cathode air supply. Heated air leaving the recuperator enters the cathode chamber

Solid Oxide Fuel Cells Using Syngas

353

that is formed between the outsides of the tubes and a containment vessel. Gases leaving the hot side of the recuperator form the exhaust from the system. The tubes are usually connected electrically in series in order to get adequate voltages for power transmission, because Ecell for each cell is nominally maintained in the range 0.6 < Ecell < 0.8 V, depending on the load demand. As the load and thus i increase, Ecell and thus the total SOFC voltage decrease due to increases in the overpotentials η(i) for each cell. Usually, power electronics are used to regulate the final supply voltage for the external load circuit. Smaller portable systems (below 1 kW) are often integrated with rechargeable batteries or supercapacitors to meet transient load profiles. There are many alternatives to the configuration illustrated in Figure  11.7, and with larger central plants based on coal or biomass gasification, approaches to SOFC system integration are highly variable, and no clear system design has been implemented to date. However, for larger-scale hydrocarbon (such as natural gas) SOFC applications, the CPOx reactor, as in Figure  11.7, can be replaced with a steam reformer. The endothermic steam-reforming process (i.e., R5) requires additional heat, and one approach is to burn a fraction of the fuel to provide reforming heat. The syngas mixture leaving the reformer has higher pH2 than the syngas from the CPOx reactor, and thus it can provide better anode performance in the SOFC stack. A great deal of research and development has focused on anode recycle strategies (Peters et al., 2002) where a fraction of the hot anode exhaust (containing high steam levels) is recycled directly back to the anode inlet. Recycle strategies are valuable because the electrochemical oxidation reactions and anode material stability usually require residual fuel to maintain long-term cell performance. A significant challenge in this technology is to develop recycle pumps that can handle gases at temperatures around 800°C. If the exhaust stream is cooled before recycle, system efficiency suffers because the gases must be reheated within the cell. The steam (and CO2) in the recycle stream serves to promote internal reforming within the anode channels. The intimate thermal coupling between fuel cell heat release and internal reforming improves thermal management and system efficiency for applications using hydrocarbon feeds.

11.4.4 Co-Generation of Syngas Although SOFCs are usually designed to produce electricity from fuels such as syngas, they may also be operated as a chemical reactor to produce syngas via electrochemical partial oxidation (EPOx) (Marnellos and Stoukides, 2004; Sund­macher et al., 2005; Alcaide et al., 2006; Athanassiou et al., 2007). The electrocatalytic oxidation of CH4 to produce syngas using SOFCs is the subject of current research (Ishihara et al., 1999; Semin et al., 1999; Sobyanin and Belyaev, 2000; Zhan et al., 2006). When electricity is the only desired output, the stoichiometry of the fuel–air flow should be controlled to prefer full oxidation to CO2 and H2O. However, if syngas is the desired output, the operating stoichiometry should be controlled as follows using methane as the example fuel:

CH4 + 1/2O2 ↔ CO + 2H2

(R7)

354

Synthesis Gas Combustion: Fundamentals and Applications

This global reaction can be split into two half-cell reactions at the cathode and the anode, with the cathode reaction being the O2 electrochemical reduction reaction (R1) and the anode reaction being

CH4(g,a) + O2–(e) ↔ CO(g,a) + 2H2(g,a) + e –(a)

(R8)

R8 is a global reaction, which is realized in a multistep process that can be summarized by steam reforming (R5), water–gas-shift (R6), and H2 and CO electrochemical oxidation (R2 and R3). To achieve a global reaction (R8), CH4-air stoichiometry should be increased fourfold over conditions for full oxidation. The increase in fuel flow does not significantly change the electrical power production if ηconc,a and ηact,a are not significantly impacted by the increase of fuel fraction in the anode gas composition. Compared to the conventional CH4 partial oxidation to syngas, the electrocatalytic oxidation in an SOFC simultaneously produces electricity and syngas. Furthermore, because CH4 and O2 are not mixed, the potential for hazardous explosion is reduced compared to a conventional partial oxidation process, and the SOFC process eliminates any need for N2 separation from the syngas. Finally, the selectivity between electricity and syngas can be controlled to meet varying load demands. When operating with the high hydrocarbon concentrations needed for selectivity to syngas, deposit formation is a practical concern. This is especially the case when low-cost Ni-based anodes are used. Anode barrier layers have been developed for these conditions to trap H2O in the anode support layer for improved reforming (Lin et al., 2006; Zhu et al., 2006), and operating conditions, such as low cell voltage to increase steam production, can also provide stable CPOx operation (Zhan and Barnett, 2005).

11.5 SOFC Modeling There is a great deal of current research devoted to modeling SOFC systems, and numerous alternative approaches have been reported. These models can be of great value in designing and optimizing cell architecture and operating conditions. They can also provide insight in interpreting experimental observations. One recent study has specifically investigated the behavior of coal-derived syngas in planar Ni/YSZ SOFC systems (Gemmen and Trembly, 2006). Among other operating conditions, they consider the effects of pressure up to 15 atm. The model reveals a critical pressure (about 8 atm) under which H2 is produced through the methane steam-reforming reaction and above which methane is the principal product. The study also discusses the important thermal consequences associated with internal steam-reforming and water-gas-shift reactions. Another recent investigation used a two-dimensional isothermal model to represent an SOFC button cell operating on syngas fuel (Suwanwarangkul et al., 2006). This investigation included experimental validation. Deleterious carbon deposition was observed at an operating temperature of 800°C. At 900°C, however, no carbon formation was observed. Thermodynamic analysis is used to interpret the effects of pressure, temperature, and syngas composition on the carbon formation process.

Solid Oxide Fuel Cells Using Syngas

355

11.5.1 Modeling Approach The physical models used in the presentation here are previously documented (Zhu et al., 2005; Zhu and Kee, 2007), and only a brief summary is given here. Gas flow within a tube or flow channel is treated as plug-flow, neglecting radial spatial variations. Gas-phase chemistry is neglected owing to very small reaction rates for methane and syngas at temperatures below around 900°C (Gupta et al., 2006b). Reactive porous-media transport is modeled using a dusty-gas model (DGM), which represents pressure-driven convective fluid flow as well as ordinary and Knudsen molecular diffusion (Mason and Malinauskas, 1983). The porous-media problem is solved through the thickness of the electrodes, but axial transport is neglected due to the very high length-to-thickness ratios of the electrodes. Reforming chemistry within the Ni/YSZ anode is modeled with an elementary reaction mechanism that incorporates steam and dry reforming as well as partial oxidation (Hecht et al., 2005). This detailed heterogeneous mechanism considers forty-two reactions among six gasphase species and twelve surface-adsorbed species. The mechanism does not specifically account for coke formation reactions. Charge transfer is assumed to proceed at the interface between the porous electrode structures and the dense electrolyte. The available three-phase-boundary length and catalyst area are taken as empirical parameters (incorporated into an exchange current density prefactor iO* ) that are adjusted to represent measured MEA performance. The model assumes that charge transfer in the anode proceeds only through H2 produced as a result of reforming chemistry and water–gas-shift chemistry (Kee et al., 2005; Zhu et al., 2005). This is based on the observation that CO charge transfer is slow compared to that of H2 and to the rates of CO oxidation via the heterogeneous water-gas-shift reaction. The models include thermal transport in the gas phase and in the MEA structure. Heat is generated within the MEA via inefficiencies in the charge transfer chemistry and ion transport. Some of this heat may be consumed by internal reforming. An anode-­supported cell using Ni/YSZ has reasonably large thermal conductivity. Thus, axial conduction is included within the tube wall. Heat is transferred to (or from) the inner fuel flow by convection at the tube wall. Heat is also transferred to the outer airflow by convection. Table  11.1 summarizes the physical parameters used for the modeling results discussed in the following section. The parameters were set to represent a particular MEA structure operating in a button cell configuration (Lin et al., 2006). The measured polarization characteristics for the cell operating on humidified CH4 are compared to the model predictions in Figure 11.8. Model predictions in the following section use the same MEA structure, but implemented in a tubular cell with operation on various syngas compositions derived from a range of carbonaceous fuels. A model for a particular tubular, anode-supported SOFC (e.g., Figure  11.6) is used as a means to compare SOFC performance with alternative syngas mixtures. The tube has an inner diameter (Da) of 0.8 cm and is 25 cm long. The porous Ni/YSZ anode tube wall is 900 µm thick. The effective thermal conductivity of the Ni/YSZ tube wall is 10.5 W/m·K. The YSZ electrolyte is 20 microns thick, and the cathode is 50 microns of porous LSM/YSZ. Other physical and chemical parameters are shown in Table 11.1.

356

Synthesis Gas Combustion: Fundamentals and Applications

Table 11.1 Parameters for the Tubular MEA Structure Used in Simulations Material Thickness (µm) Porosity Tortuosity Average pore radius (µm) Average particle diameter (µm) Exchange current factor, ik* (A·cm–2) Anodic symmetry factor, αa Cathodic symmetry factor, αc Specific catalyst area, acat (cm–1) Electrolyte conductivity prefactor, σO (S·cm–1) Electrolyte conductivity activation, Eact (J·mol–1)

Anode

Cathode

Electrolyte

Ni/YSZ 900 0.35 4.8 0.20 1.0 8.5 1.5 0.5 1080 3.6E5 2.0E4

LSM/YSZ 50 0.35 4.0 0.25 1.25 2.4 1.5 0.5

YSZ 20

In all cases, the anode inlet flow conditions are fixed at 60 cm/s, 750°C, and atmospheric pressure in order to focus in on the effects of syngas composition on simulated SOFC performance. The operating voltage is also fixed at Ecell = 0.75 V. The cathode air temperature is maintained at 750°C, and heat is transferred from the exterior of the tube via a heat transfer coefficient. Changes in temperature can impact the simulated performance since increasing (decreasing) cell inlet temperature can decrease (increase) dominant overpotentials (ηOhm, ηact,c, and ηconc,a). However, predicted trends associated with changes in syngas compositions are similar at different operating temperatures. The syngas mixtures in this simulation study represent a variety of sources and processing techniques, including catalytic partial oxidation (CPOx) and steam reforming (SR) of methane and dodecane. These processes are assumed to provide equilibrium product distributions at a given reactor temperature. Especially for steam reforming, the reactor temperature significantly affects syngas composition. In the cases of coal and biomass gasification, there are many possible processes to produce the syngas, usually involving O2 or air with H2O in a gasifier. As illustrative examples, the results presented here use measured compositions for a few processes that are available in the literature.

11.5.2 Performance with Syngas from CPOx of Hydrocarbons CPOx reactors are known to operate near chemical equilibrium at high temperature with active catalysts. Table 11.2 shows predicted equilibrium composition of a CPOx reactor operating on either CH4-air or C12H26-air mixtures at outlet temperatures of 800 and 1000°C. In all cases, the inlet mixture is stoichiometric to syngas (i.e., O2:CH4 = 0.5 and O2:C12H26 = 6). Because air is used, the resulting syngas mixtures

357

Solid Oxide Fuel Cells Using Syngas Functional layer

Cathode

Anode suport

Electrolyte

Load

Fuel

1.4 p = 1 atm T = 800°C

Cell Potential (Volts)

1.0

1.2 1.0

0.8

0.8 0.6 0.6 0.4

0.4 97% CH4 3% H2O

0.2 0.0

0

1.0 2.0 3.0 Current Density (A/cm2)

Power Density (W/cm2)

1.2

0.2 0.0

Figure 11.8  The button cell configuration used to develop the MEA model and results of button cell performance (Ecell vs. i) with internal methane steam reforming on the Ni/YSZ anode.

358

Synthesis Gas Combustion: Fundamentals and Applications

Table 11.2 Equilibrium Mole Fractions from CH4-Air and C12H26 -Air CPOX Reactors Operating at Two Temperatures with All Inlet Mixtures Stoichiometric for H2 and CO Formation Methane (CH4) Case T (°C) Inlet C12H26 CH4 O2 N2 H2 CO H2O CO2 ΔH (kJ/mol) ΔG (kJ/mol) ΔG/ΔH

Dodecane (C12H26)

CPOxM800

CPOxM1000

800 Outlet

1000 Outlet

0.010

0.001

Inlet

CPOxD800

CPOxD1000

800 Outlet

1000 Outlet

0.007

0.001

0.481 0.259 0.245 0.004 0.004 139.3 93.38 0.670

0.475 0.271 0.252 0.001 0.000 139.1 92.2 0.663

0.034 0.296 0.148 0.556

0.393 0.391 0.195 0.007 0.004 160.5 110.0 0.686

0.386 0.407 0.204 0.001 0.000 159.7 108.0 0.676

0.203 0.763

contain significant levels of N2. For the most part, the compositions are insensitive to temperature. As should be expected because of higher carbon fraction in the fuel, the dodecane reformate contains higher CO and lower H2 levels. The heating values ΔH are greater for the CH4 reformate than for the C12H26, but ΔG/ΔH, which is an indicator of the reversible efficiency (Zhu and Kee, 2006b), is quite similar for all the cases. Figures 11.9 and 11.10 compare solution profiles with the syngas input coming from CPOx of methane and dodecane. The lower panels of these figures show temperature profiles for the MEA (i.e., tube wall) and the fuel product flow (i.e., gas flowing inside the tube). The velocity profile is for the gas flow inside the anode tube. The middle panel shows profiles of the species mole fractions Xk and i. The upper panels show gas-phase Xk profiles through the thickness of the porous anode. These profiles are shown for three axial locations along the length of the tube. Within the porous anode the gradients of fuel species (H2 and CO) are toward the dense electrolyte interface (top of the panel), with gradients of product species (H2O and CO2) toward the flow channel. There is very little CH4 in these syngas streams, and thus very little reforming within the anode structure. However, the water-gasshift process is active, which accounts for much of the curvature in the species profiles through the anode structure. A purely diffusive process would result in nearly linear profiles. The syngas enters the tube at 750°C. As a result of internal heat generation within the MEA, the tube wall temperature increases. Because of axial conduction within the tube wall, the wall temperature exceeds the flow temperature in the entry regions

359

Solid Oxide Fuel Cells Using Syngas

H2 0.4 0.2

H2

0.6

CO2 0.4 0.2

CO

CO 0

0

0.1 0.2 0.3 0.4 0.5 Mole Fraction

Mole Fraction

0.4

H2O

0

0

0.1 0.2 0.3 0.4 0.5

H2

0.2

N2

H2

0.4

CO 0.2 0

0

0.1 0.2 0.3 0.4 0.5 Mole Fraction 1.2 1.0

H2O

0.8 0.6

CO2

0.1

800

MEA

790

Flow temperature

0.4

H2

CO

0 810 Temperature (°C)

CO2

0.6

Mole Fraction

N2

0.3

H2O

0.2 65

Methane–Air CPOx

64

• T = 1000°C • 29.6% CH4 • 14.8% O2, 55.6% N2

780 770

63 62 61

760 750

0

5

10 15 Distance along Tube (cm)

20

Current (A/cm2)

0.6

0.8

N2

25

Velocity (cm/s)

CO2

0.8 H2O

MEA Position (mm)

N2

MEA Position (mm)

MEA Position (mm)

0.8

60

Figure 11.9  Solution profiles for case CPOxM1000.

of the tube. In these models, the tube wall is assumed to be insulated axially at both ends. Convection between the wall and the flow increases the flow temperature. In the downstream sections of the tube, the flow temperature is slightly greater than the wall temperature. As the local i decreases due to fuel depletion, internal heat generation within the MEA decreases. The flow carries energy from the upstream hightemperature regions downstream. The differences in peak temperatures between the methane and dodecane cases are only around 10°C, which suggests that changes in syngas composition for CPOx-processed hydrocarbons will likely not change thermal management strategies for SOFC stacks. The slightly higher temperatures for the CH4 case result from higher H2 content, and consequently higher i causing both increased power production and internal heat generation. Cell performance can be measured in terms of conversion efficiency, fuel utilization, and power density (Zhu and Kee, 2006b). The cell efficiency εcell is defined by the electrical power produced divided by the heat that would be released upon full oxidation of the fuel:

360

Synthesis Gas Combustion: Fundamentals and Applications

0.8

CO 0.4 H2

0

0

0.6

Mole Fraction

N2

CO

0.4 0.2 0

0.1 0.2 0.3 0.4 0.5

CO2

0

Mole Fraction

0.4

CO H2

0.2 0

N2

H2O

0.6

0.1 0.2 0.3 0.4 0.5

0

Mole Fraction

0.1 0.2 0.3 0.4 0.5 Mole Fraction

0.5

1.0

N2

0.4

0.8

H2

0.3 0.2

CO

0.1

0.6 H2O

CO2

0 800 Temperature (°C)

CO2

790

MEA

780

0.4

H2

CO

0.2 65

Dodecame–Air CPOx • T = 1000°C • 3.4% C12H26 • 20.3% O2, 76.3% N2

Flow temperature

770

64 63 62 61

760 750

0

5

10 15 Distance along Tube (cm)

20

25

Velocity (cm/s)

0.2

N2

H2

Current (A/cm2)

H2O

0.6

0.8 H2O

MEA Position (mm)

CO2 MEA Position (mm)

MEA Position (mm)

0.8

60

Figure 11.10  Solution profiles for case CPOxD1000. L



εcell =

∫ iE 0

cell

dA

m f,in ∆hf,in



(11.14)

In this expression, m· f,inΔhf,in is inlet fuel mass flow rate times the specific enthalpy associated with completely oxidizing the inlet fuel stream (after CPOx). The electrical work is the product of iEcell integrated over the active membrane-electrode assembly (MEA) area. Fuel utilization εU can be written as

εU = 1 −

m f,out ∆hf,out m f,in ∆hf,in

(11.15)

361

Solid Oxide Fuel Cells Using Syngas

Table 11.3 Predicted Overall Performance for Alternative Syngas Mixtures in a Tubular SOFC CPOxM1000 CPOxD1000 SRM500 SRM700 SRD500 SRD700 CoalDry CoalMov BioDry BioWet

Efficiency, εcell (%)

Utilization, εU (%)

Power (W)

52.0 49.6 47.8 53.3 48.2 53.0 44.0 45.4 48.4 50.7

92.6 90.6 71.4 92.1 71.8 92.1 82.2 80.8 80.4 84.2

29.8 24.8 35.2 35.6 32.2 32.8 39.3 37.8 22.7 20.2

where the “in” and “out” refer to the inlet and outlet of the fuel cell. This definition accounts for the energy content of any remaining fuels (or fuel by-products) that leave in the anode exhaust. εcell and εU consider only performance of the SOFC and not overall system performance. Table 11.3 shows computed performance from all cases considered in this study. Comparing performance of the two CPOx cases shows that despite the different syngas compositions, overall efficiency and fuel utilization are close. The methane-based mixture, with higher H2 content, shows slightly higher efficiency and utilization. However, the average power density (Ecell · imean) for the methane CPOx effluent is 0.386 W/cm2 electrolyte membrane, whereas for the dodecane CPOx, it is 0.321 W/cm2. The initially higher energy content of the methane-based syngas delivers significantly higher power from the tube.

11.5.3 Performance with Syngas from Steam Reforming of Hydrocarbons Table  11.4 shows syngas mixtures coming from the steam reforming of methane and dodecane. In all cases the steam-carbon ratio is 2.5, which is a typical operating point for catalytic steam reformers. The steam-carbon ratio is usually chosen to prevent coking within the reformer. Steam reformation is a strongly endothermic process, requiring external heat input to maintain the desired reforming temperature. Comparing Tables 11.2 and 11.4 reveals that the syngas produced by steam reforming has a significantly higher energy content (i.e., ΔH) than that from the CPOx process. This is the result of energy added to maintain the reforming temperature. Unlike the CPOx reactor, the reformer outlet temperature has a significant influence on the syngas composition, with lower-temperature operation producing higher levels of CH4 in the reformate. This CH4 is ultimately reformed within the SOFC anode, which can be beneficial (Gupta et al., 2006c). The endothermic internal

362

Synthesis Gas Combustion: Fundamentals and Applications

Table 11.4 Equilibrium Mole Fractions from Methane and Dodecane Steam Reforming at Two Temperatures (Inlet Steam-to-Carbon Ratios Are S/C=2.5) Methane (CH4) Case T (°C) Inlet C12H26 CH4 H2 CO H2O CO2 ΔH (kJ/mol) ΔG (kJ/mol) ΔG/ΔH

SRM500

SRM700

500 Outlet

700 Outlet

0.142 0.353 0.013 0.413 0.079 205.2 167.6 0.817

0.008 0.595 0.111 0.220 0.066 185.9 124.9 0.672

Dodecane (C12H26)

Inlet

SRD500

SRD700

500 Outlet

700 Outlet

0.130 0.309 0.018 0.416 0.126 186.3 151.6 0.813

0.006 0.533 0.123 0.246 0.092 172.2 114.0 0.662

0.0323 0.286

0.714

0.9677

reforming can improve efficiency and assist in thermal management within the cell. Furthermore, as can be seen from Table 11.4, the low-temperature steam reforming produces syngas with greater ΔH and greater reversible efficiency ΔG/ΔH. The examples compare SOFC performance using syngas from low- and hightemperature steam reforming. Figure 11.11 illustrates solution profiles for the hightemperature CH4 reformer case, where there is very little CH4 slip into the SOFC anode. The MEA and fuel temperatures rise considerably above the inlet temperature due to the rapid heat release within the MEA at the front end of the cell. Both the MEA and fuel temperatures increase to a peak around 3 cm, with the MEA temperature significantly higher than the fuel flow. The tube wall with its high thermal conductivity transfers heat along the axis of the tube and spreads this heat through the SOFC structure. In the downstream sections the flow temperature exceeds the MEA temperature. In this region heat transfer to the relatively cool external air exceeds the MEA heat release, serving to reduce the MEA temperature. The internal heat generation, which scales approximately as i2, decreases along the length of the tube as the current density i decreases. The fuel flow temperature is also decreasing via heat transfer with the wall. The fuel velocity varies very little in this example. The velocity depends on the molar flow rate and the density. With a fuel stream that is primarily H2 and CO, the charge transfer chemistry does not cause a mole change. Although atomic oxygen is crossing the electrolyte membrane and entering the fuel channel, the electrochemical oxidation of one mole of H2 produces one mole of H2O (R2). Thus, H2 electrochemistry does not change the molar flow rate within the tube. The small velocity variations are simply the result of the density variations associated with the

363

Solid Oxide Fuel Cells Using Syngas

H2

0.4

CO2

0.2

0.6 0.4 CO 0.2

H2

CO

Temperature (°C)

Mole Fraction

0

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 820 810 800 790 780 770 760 750

0

0.2 0.4 0.6 Mole Fraction

0.8

0

H2O

0

0.2 0.4 0.6 Mole Fraction

0.8

CO2 0.6 H2 0.4

H2O

0.2 0

CO 0

0.2 0.4 0.6 Mole Fraction

0.8

1.2

H2O

1.0 0.8 0.6

H2 CO2 MEA

0.4 0.2

CO

0 65

Flow temperature

64 63

Methane steam reforming • T = 700°C • 28.6% CH4 • 71.4% H2O, (S/C = 2.5) 0

5

62 61 10 15 Distance along Tube (cm)

Current (A/cm2)

0.6

MEA Position (mm)

H2O

0.8

CO2

20

25

Velocity (cm/s)

0.8 MEA Position (mm)

MEA Position (mm)

0.8

60

Figure 11.11  Solution profiles for case SRM700.

temperature variations. As discussed in the subsequent example, the situation is different when methane is in the fuel. The middle panel in Figure 11.11 shows Xk and i profiles. The inlet fuel is dominated by H2 at nearly 60%, with 11% CO and 22% H2O. As the H2 and CO fuel is depleted, i also decreases along the length of the channel. The operating condition for this example leads to high utilization (Table 11.3). At the channel exit there is only about 5% H2 remaining and the CO is entirely depleted. The higher H2 content in the steam-reforming syngas also leads to higher power output than in the CPOx syngas, and power densities for the four steam-reforming cases shown in Table 11.3 range from 0.417 to 0.466 W/cm2 of electrolyte membrane. The upper panels of Figure 11.11 show the gas-phase Xk in the pore spaces through the thickness of the anode. The concentration profiles indicate the concentration driving forces for H2 and CO diffusion into the anode and for H2O and CO2 out from the MEA into the fuel channel. Only H2 is consumed via charge transfer chemistry in the three-phase region near the dense electrolyte interface (top of the upper panels).

364

Synthesis Gas Combustion: Fundamentals and Applications

CO

0.6

0.8 H2O

H2

0.4 CH4 0.2

0.8

CO2

0.6 H2O 0.4 CO 0.2

CH4

MEA Position (mm)

CO2

MEA Position (mm)

CO2

0.6 H2

0.4

CH4 CO

0.2

H2 0

0.2

0.4

0.6

0

0.8

Temperature (°C)

Mole Fraction

Mole Fraction 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 780

0.2

0.4

0.6

0.8

0

0

Mole Fraction

0.2

0.4

0.6

0.8

Mole Fraction 1.0

H2O

0.8 0.6

CO

CH4

H2

0.4

CO2

0.2 0 75

775

MEA

770

Flow temperature

70

Methane steam reforming • T = 500°C • 28.6% CH4 • 71.4% H2O, (S/C = 2.5)

765 760 755 750

0

Current (A/cm2)

0

H2O

0

5

10 15 Distance along Tube (cm)

65

20

25

Velocity (cm/s)

MEA Position (mm)

0.8

60

Figure 11.12  Solution profiles for case SRM500.

CO and H2O participate in the water–gas-shift process via catalytic chemistry on the Ni surfaces within the anode structure. The small amount of CH4 in this fuel stream is heterogeneously reformed within the anode. In downstream sections, as the fuel is depleted and i decreases, the species gradients within the anode also decrease. Figure  11.12 illustrates the performance that results from the low-temperature steam reforming, which provides approximately 14% CH4 into the fuel cell inlet. The cell temperatures are approximately 30°C lower than those in the previous case without methane (Figure 11.11). The decrease in temperature is due to the endothermic CH4 reforming in the anode structure and to the lower i, resulting in less waste heat production. In addition to lower temperatures overall, the temperature variation along the length of the cell is only about 12°C, compared with nearly 50°C in the previous case. Reducing temperature variations is a generally desirable goal on SOFC design and operation to avoid high thermal stresses due to variations in coefficients of thermal expansion of the different MEA materials.

365

Solid Oxide Fuel Cells Using Syngas

As seen in the lower panel of Figure 11.12, the velocity increase is much larger than that in the previous case (Figure 11.11). The steam reforming of methane (i.e., R5) produces a net mole increase, which serves to accelerate the flow. The middle panel of Figure 11.12 shows that fuel utilization is much lower than that in the previous case. The flow at the tube exit still contains 10% H2, 2% CO, and 4% CH4. In practice, such a circumstance would likely call for a reduced inlet flow velocity. However, to assist making comparisons, the same operating conditions are used for the two examples. The high CH4 in the inlet flow for the case in Figure 11.12 provides additional H2 due to steam reforming inside the porous anode. This compensates to some extent for the low initial H2 fractions, but also lowers the cell temperature, which decreases the average current density and thus overall syngas fuel utilization.

11.5.4 Performance with Syngas from Coal and Biomass Gasification Table 11.5 shows syngas composition for selected coal (Gemmen and Trembly, 2006) and biomass (Omosun et al., 2004) gasification processes. These cases are chosen to be somewhat representative, but there are many alternative compositions that can be found in the literature. Furthermore, for the purposes of this study, clean syngas is assumed with all higher hydrocarbons, tars, and other impurities removed. Table 11.5 reveals that the selected coal gasification processes deliver syngas with relatively high energy content ΔH, while the biomass-derived syngas has relatively low energy content due to the relatively high N2 concentration. The reversible efficiencies ΔG/ΔH for the coal syngas are slightly lower than those for the biomassderived syngas. However, because of lower energy content, larger cell areas would be needed to realize the higher efficiency from the biomass feeds. Table 11.5 Equilibrium Mole Fractions of Syngas from Selected Coal-and-Biomass Gasification Processes Coal Case CH4 H2 CO H2O CO2 N2 ΔH (kJ/mol) ΔG (kJ/mol) ΔG/ΔH

Biomass

CoalDry Dry Feed

CoalMov Moving Bed

BioDry Dry Feed

BioWet Wet Feed

0.000 0.300 0.603 0.020 0.016 0.047 248.1 160.1 0.645

0.042 0.264 0.460 0.163 0.029 0.028 232.3 157.8 0.679

0.047 0.200 0.153 0.000 0.129 0.471 130.5 95.78 0.734

0.040 0.170 0.130 0.150 0.110 0.400 110.9 78.93 0.712

366

Synthesis Gas Combustion: Fundamentals and Applications

0.6 0.4

H2O

0.2

H2

0

CO

0.2

0.4

0.6

0.6

CO2 H2O

0.4 N2 0.2 0

H2

0

CO

0.2

Mole Fraction

0.4

0.6

CO2

0.6 0.4

N2 H2

0

H2O

CO

0.2

0

Mole Fraction

0.2

0.4

0.6

Mole Fraction

Temperature (°C)

1.2

0.5 0.4

CO

0.3

0.8

H2

0.2

0.6

CO2

0.4

N2

0.1 0 820 810 800 790 780 770 760 750

1.0

H2O

0.2

Current (A/cm2)

Mole Fraction

0.6

0 65 Flow temperature

MEA

64 63

Coal syngas • Dry feed

0

5

62 61 10 15 Distance along Tube (cm)

20

25

Velocity (cm/s)

0

CO2

N2

0.8 MEA Position (mm)

0.8 MEA Position (mm)

MEA Position (mm)

0.8

60

Figure 11.13  Solution profiles for case CoalDry.

Consider first the two coal-derived fuels. The dry feed has high CO content and nearly no steam. The moving-bed syngas has less H2 and CO, and significantly more steam than the dry-feed case. Figures  11.13 and 11.14 show the considerable differences in the solution profiles. Nevertheless, as seen from Table 11.3, the overall performance is quite similar. In both cases the predicted efficiency is about 45%, the utilization is about 80%, and the net power is about 38 W. Both achieve peak MEA temperatures around 810°C. Because the dry-feed case has relatively high CO and no steam at the inlet, it could be more susceptible to forming deposits. In the dry-feed case, water-gas-shift reactions are active later in the tube because the H2O is solely the product of the charge transfer chemistry. Figure  11.15 shows solution profiles for case BioWet (solution for BioDry not shown). The syngas for the biomass cases are dominated by N2, which significantly reduces the available energy content. As seen from Table 11.3, the net power produced is only about one half of that produced by the coal-derived syngas. However,

367

Solid Oxide Fuel Cells Using Syngas

CO2 CH4

0.2

H2 0

CO

0.4 H2O 0.2

CO

0.6

N2 CO

0.4

H2 CH4

0.2

CH4

0.1 0.2 0.3 0.4 0.5 Mole Fraction

0

0

H2O

0.1 0.2 0.3 0.4 0.5

0

0

Mole Fraction

0.1 0.2 0.3 0.4 0.5 Mole Fraction

Mole Fraction

0.5 0.4

CO2

CO

0.3

1.0 H2O

0.8 0.6

H2

0.2

0.4

0.1

CH4

0.2

N2

0 65

Temperature (°C)

0 820 810 800 790

63 62

Coal syngas

780

• Moving bed

770 760

64

Flow temperature

MEA

0

5

61 10 15 Distance along Tube (cm)

20

25

Velocity (cm/s)

0

0.6

CO2

Current (A/cm2)

N2

0.8

CO2

H2

MEA Position (mm)

0.6 0.4

N2

0.8

H2O MEA Position (mm)

MEA Position (mm)

0.8

60

Figure 11.14  Solution profiles for case CoalMov.

the conversion efficiency is somewhat higher than with the coal-based syngas. As with the coal cases, the two biomass cases show remarkably similar overall performance (Table  11.3), despite quite different syngas mixtures and local solution profiles. The biomass cases achieve somewhat lower MEA temperatures, primarily owing to lower energy content in the syngas feed.

11.6 Conclusions SOFC systems have the potential to convert syngas derived from a variety of sources with very high conversion efficiencies. The implementation of such systems will depend largely upon the development of economic materials and systems that provide long-term durability with operation at reasonable power densities. The development of such systems is currently being driven both by small-scale power applications based on hydrocarbon reforming (or partial oxidation) for near-term applications and

368

Synthesis Gas Combustion: Fundamentals and Applications

0.4

H2

0.2 0

H2O N2

CO

0

CO

0

0.1 0.2 0.3 0.4 0.5

H2O 0

0.4

H2 CO

0.2 0

0.1 0.2 0.3 0.4 0.5

N2

0.3

N2 H2O

0

0.1 0.2 0.3 0.4 0.5

Mole Fraction

0.4 Mole Fraction

CO2

0.2

Mole Fraction

Mole Fraction 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 65

H2O CO2

0.2 0.1

H2

CH4

0 780 Temperature (°C)

CH4

0.4

CO2

0.6

775

CO

64

MEA

770

63

Flow temperature

765 760

62

Biomass wet

61

755 750

0

5

10 15 Distance along Tube (cm)

20

25

Current (A/cm2)

CO2

H2

0.6

MEA Position (mm)

0.6

0.8

N2

Velocity (cm/s)

0.8

CH4 MEA Position (mm)

MEA Position (mm)

0.8

60

Figure 11.15  Solution profiles for case BioWet.

by large-scale applications based on coal- or biomass-derived syngas in the longer term. Development of these longer-term central power plant systems will likely be spurred by the desire to implement CO2 sequestration because SOFC-based plants offer key advantages for efficient CO2 capture technology. To understand SOFC performance on the different range of syngas studies, modeling results presented here compare fuel cell performance as a function of syngas compositions that are derived from hydrocarbon preprocessing as well as coal or biomass gasification. The conversion efficiencies are found to depend only weakly on the syngas composition. However, the power densities depend strongly on composition, especially the H2 levels and net heating values of the fuel. H2 likely dominates the charge transfer process. CO and CH4 in the fuel stream are converted to H2 by internal water-gas-shift and steam-reforming processes, respectively. The steam needed to support these processes can be supplied with the syngas and is produced by the electrochemical charge transfer. These catalytic reactions within the porous anode structure enable fuel flexibility within wide ranges of syngas composition.

Solid Oxide Fuel Cells Using Syngas

369

The suitability for SOFC systems to incorporate carbon capture improves as full fuel utilization is approached, at which point the anode exhaust composition is entirely CO2 and steam. The steam is easily condensed, thus easily separating CO2. Although further sequestration of CO2 is needed, the separation technology is straightforward. Current SOFC technology, based on porous Ni/YSZ anodes, has the potential to provide very good performance using a wide range of syngas mixtures. However, there are significant limitations and restrictions that make sustaining such performance a significant practical challenge. With conventional Ni/YSZ anodes, the fuel must be purified of potentially harmful components, including sulfur compounds, halides, and higher hydrocarbons. Both sulfur and carbon can cover the Ni and deactivate its thermal and electrochemical catalytic activity. These and other operational limitations of conventional Ni-based anodes have led to research on new materials and cell architectures. Alternative anode architectures, such as anode barrier layers, may enable Ni-based anodes to operate with significant hydrocarbon content. New anode materials may enable improved fuel flexibility, improved resistance to impurities, and broader operating ranges in temperature and voltages. Advances in materials and cell architectures will be important in the development of cost-effective applications using syngas fuels derived from hydrocarbon preprocessors as well as large-scale coal and biomass gasification plants.

Acknowledgments This work was supported by the DoD Multidisciplinary University Research Initiative (MURI) program administered by the Office of Naval Research under Grant N00014-02-1-0665.

References Adler, S. B. (2004). Factors governing oxygen reduction in solid oxide fuel cell cathodes. Chem. Rev. 104:4791. Aguilar, L., Zha, S. W., Li, S. W., Winnick, J., and Liu, M. (2004). Sulfur-tolerant materials for the hydrogen sulfide SOFC. Electrochem. Solid-State Lett. 7:A324. Alcaide, F., Cabot, P. L., and Brillas, E. (2006). Fuel cells for chemicals and energy cogeneration. J. Pow. Sour. 153:47. Aloui, T., and Halouani, K. (2007). Analytical modeling of polarizations in a solid oxide fuel cell using biomass syngas product as fuel. Appl. Therm. Eng. 27:731. Araki, T., Taniuchi, T., Sunakawa, D., Nagahama, M., Onda, K., and Kato, T. (2007a). Cycle analysis of low and high H-2 utilization SOFC/gas turbine combined cycle for CO2 recovery. J. Pow. Sour. 171:464. Araki, T., Taniuchi, T., Sunakawa, D., Nagahama, M., Onda, K., and Kato, T. (2007b). Cycle analysis of low and high H-2 utilization SOFC/gas turbine combined cycle for CO2 recovery. J. Pow. Sour. 171:464. Aravind, P. V., Ouweltjes, J. P., de Heer, E., Woudstra, N., and Rietveld, G. (2005). Impact of biosyngas and its components on SOFC anodes. Paper presented at the SOFC-IX Symposium, Quebec City, Canada. Athanassiou, C., Pekridis, G., Kaklidis, N., Kalimeri, K., Vartzoka, S., and Marnellos, G. (2007). Hydrogen production in solid electrolyte membrane reactors (SEMRs). Int. J. Hydrogen Energy 32:38.

370

Synthesis Gas Combustion: Fundamentals and Applications

Atkinson, A., Barnett, S., Gorte, R. J., Irvine, J. T. S., McEvoy, A. J., Mogensen, M., Singhal, S. C., and Vohs, J. (2004). Advanced anodes for high-temperature fuel cells. Nature Mater. 3:17. Baron, S., Brandon, N., Atkinson, A., Steele, B., and Rudkin, R. (2004). The impact of woodderived gasification gases on Ni-CGO anodes in intermediate temperature solid oxide fuel cells. J. Pow. Sour. 126:58. Bessler, W. G., Warnatz, J., and Goodwin, D. G. (2007). The influence of equilibrium potential on the hydrogen oxidation kinetics of SOFC anodes. Solid State Ionics 177:3371. Brandon, N. P., Skinner, S., and Steele, B. C. H. (2003). Recent advances in materials for fuel cells. Annu. Rev. Mater. Res. 33:83. Buchinger, G., Hinterreiter, P., Raab, T., Griesser, S., Claassen, R., Claassen, D. P., Sitte, W., and Meissner, D. (2006). Operating microtubular SOFCs with hydrogen chloride and hydrogen sulfide containing fuels and synthetic wood gas. J. Fuel Cell Sci. Technol. 3:280. Cheng, Z., Zha, S. W., Aguilar, L., Wang, D., Winnick, J., and Liu, M. L. (2006a). A solid oxide fuel cell running on H2S/CH4 fuel mixtures. Electrochem. Solid-State Lett. 9:A31. Cheng, Z., Zha, S. W., and Liu, M. L. (2006b). Stability of materials as candidates for sulfurresistant anodes of solid oxide fuel cells. J. Electrochem. Soc. 153:A1302. Cordiner, S., Feola, M., Mulone, V., and Romanelli, F. (2007). Analysis of a SOFC energy generation system fuelled with biomass reformate. Appl. Therm. Eng. 27:738. Costa-Nunes, O., Gorte, R. J., and Vohs, J. M. (2005). Comparison of the performance of Cu-CeO2-YSZ and Ni-YSZ composite SOFC anodes with H-2, CO, and syngas. J. Pow. Sour. 141:241. DeCaluwe, S. C., Zhu, H., Kee, R. J., and Jackson, G. S. (2008). Importance of anode microstructure in modeling solid oxide fuel cells. J. Electrochem. Soc. 155:B538. Dijkstra, J. W., and Jansen, D. (2004). Novel concepts for CO2 capture. Energy 29:1249. Fergus, J. W. (2006). Electrolytes for solid oxide fuel cells. J. Pow. Sour. 162:30. Finnerty, C., Tompsett, G. A., Kendall, K., and Ormerod, R. M. (2000). SOFC system with integrated catalytic fuel processing. J. Pow. Sour. 86:459. Fleig, J. (2003). Solid oxide fuel cell cathodes: Polarization mechanisms and modeling of the electrochemical performance. Annu. Rev. Mater. Res. 33:361. Fleig, J. (2005). On the current-voltage characteristics of charge transfer reactions at mixed conducting electrodes on solid electrolytes. Phys. Chem. Chem. Phys. 7:2027. Gemmen, R. S., and Trembly, J. (2006). On the mechanisms and behavior of coal syngas transport and reaction within the anode of a solid oxide fuel cell. J. Pow. Sour. 161:1084. Gong, M. Y., Liu, X. B., Trembly, J., and Johnson, C. (2007). Sulfur-tolerant anode materials for solid oxide fuel cell application. J. Pow. Sour. 168:289. Gross, M. D., Vohs, J. M., and Gorte, R. J. (2007). Recent progress in SOFC anodes for direct utilization of hydrocarbons. J. Mater. Chem. 17:3071. Gubner, A., Nguyen-Xuan, T., Bram, M., Remmel, J., and de Haart, L. (2006). Lightweight cassette type SOFC stacks for automotive applications. Luzern, Switzerland: European Fuel Cell Forum. Gupta, G. K., Dean, A. M., Ahn, K., and Gorte, R. J. (2006a). Comparison of conversion and deposit formation of ethanol and butane under SOFC conditions. J. Pow. Sour. 158:497. Gupta, G. K., Hecht, E. S., Zhu, H. Y., Dean, A. M., and Kee, R. J. (2006b). Gas-phase reactions of methane and natural-gas with air and steam in non-catalytic regions of a solidoxide fuel cell. J. Pow. Sour. 156:434. Gupta, G. K., Marda, J. R., Dean, A. M., Colclasure, A. M., Zhu, H. Y., and Kee, R. J. (2006c). Performance predictions of a tubular SOFC operating on a partially reformed JP-8 surrogate. J. Pow. Sour. 162:553. Haag, J. M., Madsen, B. D., Barnett, S. A., and Poeppelmeier, K. R. (2008). Application of LaSr2Fe2CrO9-delta in solid oxide fuel cell anodes. Electrochem. Solid-State Lett. 11:B51.

Solid Oxide Fuel Cells Using Syngas

371

He, H. P., Vohs, J. M., and Gorte, R. J. (2003). Effect of synthesis conditions on the performance of Cu-CeO2-YSZ anodes in SOFCs. J. Electrochem. Soc. 150:A1470. Hecht, E. S., Gupta, G. K., Zhu, H. Y., Dean, A. M., Kee, R. J., Maier, L., and Deutschmann, O. (2005). Methane reforming kinetics within a Ni-YSZ SOFC anode support. Appl. Catal. A Gen. 295:40. Hickman, D. A., and Schmidt, L. D. (1993). Production of syngas by direct catalytic-oxidation of methane. Science 259:343. Higman, C., and van der Burgt, M. (2003). Gasification. New York: Elsevier Science. Holtappels, P., De Haart, L. G. J., Stimming, U., Vinke, I. C., and Mogensen, M. (1999). Reaction of CO/CO2 gas mixtures on Ni-YSZ cermet electrodes. J. Appl. Electrochem. 29:561. Hu, J. M., Zhang, J. Q., Cao, C. N., and Hsing, I. M. (2004). Kinetics investigation of H-2/CO electrooxidation in PEFCs by the combined use of equivalent circuit fitting and mathematical modeling of the faradaic impedance. Electrochim. Acta 49:5227. Huang, K., Wan, J., and Goodenough, J. B. (2001). Oxide-ion conducting ceramics for solid oxide fuel cells. J. Mater. Sci. 36:1093. Ishihara, T., Tabuchi, J., Ishikawa, S., Yan, J., Enoki, M., and Matsumoto, H. (2006). Recent progress in LaGaO3-based solid electrolyte for intermediate temperature SOFCs. Solid State Ionics 177:1949. Ishihara, T., Yamada, T., Akbay, T., and Takita, Y. (1999). Partial oxidation of methane over fuel cell type reactor for simultaneous generation of synthesis gas and electric power. Chem. Eng. Sci. 54:1535. Janardhanan, V. M., and Deutschmann, O. (2006). CFD analysis of a solid oxide fuel cell with internal reforming: Coupled interactions of transport, heterogeneous catalysis and electrochemical processes. J. Pow. Sour. 162:1192. Jensen, K. V., Wallenberg, R., Chorkendorff, I., and Mogensen, M. (2003). Effect of impurities on structural and electrochemical properties of the Ni-YSZ interface. Solid State Ionics 160:27. Jiang, S. P., and Chan, S. H. (2004). A review of anode materials development in solid oxide fuel cells. J. Mater. Sci. 39:4405. Jiang, Y., and Virkar, A. V. (2003). Fuel composition and diluent effect on gas transport and performance of anode-supported SOFCs. J. Electrochem. Soc. 150:A942. Kee, R. J., Coltrin, M. E., and Glarborg, P. (2003). Chemically reacting flow: Theory and practice. Hoboken, NJ: John Wiley & Sons. Kee, R. J., Zhu, H. Y., and Goodwin, D. G. (2005). Solid-oxide fuel cells with hydrocarbon fuels. Proc. Comb. Inst. 30:2379. Kim, H., Park, S., Vohs, J. M., and Gorte, R. J. (2001). Direct oxidation of liquid fuels in a solid oxide fuel cell. J. Electrochem. Soc. 148:A693. Krishnan, V. V., McIntosh, S., Gorte, R. J., and Vohs, J. M. (2004). Measurement of electrode overpotentials for direct hydrocarbon conversion fuel cells. Solid State Ionics 166:191. Kurokawa, H., Sholklapper, T. Z., Jacobson, C. P., De Jonghe, L. C., and Visco, S. J. (2007a). Ceria nanocoating for sulfur tolerant Ni-based anodes of solid oxide fuel cells. Electrochem. Solid-State Lett. 10:B135. Kurokawa, H., Yang, L. M., Jacobson, C. P., De Jonghe, L. C., and Visco, S. J. (2007b). Y-doped SrTiO3-based sulfur tolerant anode for solid oxide fuel cells. J. Pow. Sour. 164:510. Lin, Y. B., Zhan, Z. L., and Barnett, S. A. (2006). Improving the stability of direct-methane solid oxide fuel cells using anode barrier layers. J. Pow. Sour. 158:1313. Lin, Y. B., Zhan, Z. L., Liu, J., and Barnett, S. A. (2005). Direct operation of solid oxide fuel cells with methane fuel. Solid State Ionics 176:1827. Liu, J., Madsen, B. D., Ji, Z. Q., and Barnett, S. A. (2002). A fuel-flexible ceramic-based anode for solid oxide fuel cells. Electrochem. Solid-State Lett. 5:A122. Liu, Y. L., Primdahl, S., and Mogensen, M. (2003). Effects of impurities on microstructure in Ni/YSZ-YSZ half-cells for SOFC. Solid State Ionics 161:1.

372

Synthesis Gas Combustion: Fundamentals and Applications

Lu, C., Worrell, W. L., Gorte, R. J., and Vohs, J. M. (2003). SOFCs for direct oxidation of hydrocarbon fuels with samaria-doped ceria electrolyte. J. Electrochem. Soc. 150:A354. Lundberg, W. L., Veyo, S. E., and Moeckel, M. D. (2003). A high-efficiency solid oxide fuel cell hybrid power system using the mercury 50 advanced turbine systems gas turbine. J. Eng. Gas Turb. Power 125:51. Madsen, B. D., and Barnett, S. A. (2005). Effect of fuel composition on the performance of ceramic-based solid oxide fuel cell anodes. Solid State Ionics 176:2545. Marnellos, G., and Stoukides, M. (2004). Catalytic studies in electrochemical membrane reactors. Solid State Ionics 175:597. Marquez, A. I., Ohrn, T. R., Trembly, J. P., Ingram, D. C., and Bayless, D. J. (2007). Effects of coal syngas and H2S on the performance of solid oxide fuel cells. Part 2. Stack tests. J. Pow. Sour. 164:659. Mason, E., and Malinauskas, A. (1983). Gas transport in porous media: The dusty-gas model. New York: American Elsevier. Matsuzaki, Y., and Yasuda, I. (2000). Electrochemical oxidation of H-2 and CO in a H-2-H2OCO-CO2 system at the interface of a Ni-YSZ cermet electrode and YSZ electrolyte. J. Electrochem. Soc. 147:1630. McIntosh, S., and Gorte, R. J. (2004). Direct hydrocarbon solid oxide fuel cells. Chem. Rev. 104 :4845. McIntosh, S., He, H. P., Lee, S. I., Costa-Nunes, O., Krishnan, V. V., Vohs, J. M., and Gorte, R. J. (2004). An examination of carbonaceous deposits in direct-utilization SOFC anodes. J. Electrochem. Soc. 151:A604. Moller, B. F., Arriagada, J., Assadi, M., and Potts, I. (2004). Optimisation of an SOFC/GT system with CO2-capture. J. Pow. Sour. 131:320. Murray, E. P., Harris, S. J., Liu, J., and Barnett, S. A. (2006). Direct solid oxide fuel cell operation using isooctane. Electrochem. Solid-State Lett. 9:A292. Omosun, A. O., Bauen, A., Brandon, N. P., Adjiman, C. S., and Hart, D. (2004). Modelling system efficiencies and costs of two biomass-fuelled SOFC systems. J. Pow. Sour. 131:96. Ormerod, R. M. (2003). Solid oxide fuel cells. Chem. Soc. Rev. 32:17. Ouweltjes, J. P., Aravind, P. V., Woudstra, N., and Rietveld, G. (2006). Biosyngas utilization in solid oxide fuel cells with Ni/GDC anodes. J. Fuel Cell Sci. Technol. 3:495. Peters, R., Dahl, R., Kluttgen, U., Palm, C., and Stolten, D. (2002). Internal reforming of methane in solid oxide fuel cell systems. J. Pow. Sour. 106:238. Pomfret, M. B., Marda, J., Jackson, G. S., Eichhorn, B. W., Dean, A. M., and Walker, R. A. (2008). Hydrocarbon fuels in solid oxide fuel cells: In situ Raman studies of graphite formation and oxidation. J. Phys. Chem. C 112:232. Randolph, K. L., and Dean, A. M. (2007). Hydrocarbon fuel effects in solid-oxide fuel cell operation: An experimental and modeling study of n-hexane pyrolysis. Phys. Chem. Chem. Phys. 9:4245. Sammes, N. M., Du, Y., and Bove, R. (2005). Design and fabrication of a 100 W anode supported micro-tubular SOFC stack. J. Pow. Sour. 145:428. Sasaki, K., and Maier, J. (2000). Re-analysis of defect equilibria and transport parameters in Y2O3-stabilized ZrO2 using EPR and optical relaxation. Solid State Ionics 134:303. Sasaki, K., and Teraoka, Y. (2003). Equilibria in fuel cell gases. I. Equilibrium compositions and reforming conditions. J. Electrochem. Soc. 150:A878. Schmidt, L. D., Klein, E. J., Leclerc, C. A., Krummenacher, J. J., and West, K. N. (2003). Syngas in millisecond reactors: Higher alkanes and fast lightoff. Chem. Eng. Sci. 58:1037. Schneider, L. C. R., Martin, C. L., Bultel, Y., Bouvard, D., and Siebert, E. (2006). Discrete modelling of the electrochemical performance of SOFC electrodes. Electrochim. Acta 2:14. Semin, G. L., Belyaev, V. D., Demin, A. K., and Sobyanin, V. A. (1999). Methane conversion to syngas over Pt-based electrode in a solid oxide fuel cell reactor. Appl. Catal. A Gen. 181:131.

Solid Oxide Fuel Cells Using Syngas

373

Serra, J. M., Uhlenbruck, S., Meulenberg, W. A., Buchkremer, H. P., and Stover, D. (2006). Nano-structuring of solid oxide fuel cells cathodes. Top. Catal. 40:123. Shao, Z. P., and Haile, S. M. (2004). A high-performance cathode for the next generation of solid-oxide fuel cells. Nature 431:170. Sheng, C. Y., and Dean, A. M. (2004). Importance of gas-phase kinetics within the anode channel of a solid-oxide fuel cell. J. Phys. Chem. A 108:3772. Sobyanin, V. A., and Belyaev, V. D. (2000). Gas-phase electrocatalysis: Methane oxidation to syngas in a solid oxide fuel cell reactor. Solid State Ionics 136:747. Steele, B. C. H. (1999). Fuel-cell technology—Running on natural gas. Nature 400:619. Sukeshini, A. M., Habibzadeh, B., Becker, B. P., Pomfret, M. B., Demircan, O., Walker, R. A., Eichhorn, B. W., and Jackson, G. S. (2005). CO and CH4 electrochemical oxidation on Ni patterned anodes. Paper presented at the SOFC-IX Symposium, Quebec City, Canada. Sukeshini, A. M., Habibzadeh, B., Becker, B. P., Stoltz, C. A., Eichhorn, B. W., and Jackson, G. S. (2006). Electrochemical oxidation of H-2, CO, and CO/H-2 mixtures on patterned Ni anodes on YSZ electrolytes. J. Electrochem. Soc. 153:A705. Sundmacher, K., Rihko-Struckmann, L. K., and Galvita, V. (2005). Solid electrolyte membrane reactors: Status and trends. Catal. Today 104:185. Suwanwarangkul, R., Croiset, E., Entchev, E., Charojrochkul, S., Pritzker, M. D., Fowler, M. W., Douglas, P. L., Chewathanakup, S., and Mahaudom, H. (2006). Experimental and modeling study of solid oxide fuel cell operating with syngas fuel. J. Pow. Sour. 161:308. Tanaka, Y., Nguyen, T. L., Kato, T., Shimada, T., Sugano, K., Negishi, A., Kato, K., and Nozaki, K. (2006). Performance of a tubular solid oxide fuel cell with model kerosene reformate gas. Solid State Ionics 177:3323. Trembly, J. P., Gemmen, R. S., and Bayless, D. J. (2007a). The effect of coal syngas containing AsH3 on the performance of SOFCs: Investigations into the effect of operational temperature, current density and AsH3 concentration. J. Pow. Sour. 171:818. Trembly, J. P., Gemmen, R. S., and Bayless, D. J. (2007b). The effect of coal syngas containing HCl on the performance of solid oxide fuel cells: Investigations into the effect of operational temperature and HCl concentration. J. Pow. Sour. 169:347. Trembly, J. P., Gemmen, R. S., and Bayless, D. J. (2007c). The effect of IGFC warm gas cleanup system conditions on the gas-solid partitioning and form of trace species in coal syngas and their interactions with SOFC anodes. J. Pow. Sour. 163:986. Verma, A., Rao, A. D., and Samuelsen, G. S. (2006). Sensitivity analysis of a Vision 21 coal based zero emission power plant. J. Pow. Sour. 158:417. Weber, A., and Ivers-Tiffee, E. (2004). Materials and concepts for solid oxide fuel cells (SOFCs) in stationary and mobile applications. J. Pow. Sour. 127:273. Weber, A., Sauer, B., Muller, A. C., Herbstritt, D., and Ivers-Tiffee, E. (2002). Oxidation of H-2, CO and methane in SOFCs with Ni/YSZ-cermet anodes. Solid State Ionics 152:543. Xu, Z. R., Luo, J. L., and Chuang, K. T. (2007). CoS-promoted MoS2 catalysts for SOFC using H2S-containing hydrogen or syngas as fuel. J. Electrochem. Soc. 154:B523. Zha, S. W., Cheng, Z., and Liu, M. L. (2007). Sulfur poisoning and regeneration of Ni-based anodes in solid oxide fuel cells. J. Electrochem. Soc. 154:B201. Zhan, Z. L., and Barnett, S. A. (2005). Use of a catalyst layer for propane partial oxidation in solid oxide fuel cells. Solid State Ionics 176:871. Zhan, Z. L., and Barnett, S. A. (2006). Solid oxide fuel cells operated by internal partial oxidation reforming of iso-octane. J. Pow. Sour. 155:353. Zhan, Z. L., Lin, Y. B., Pillai, M., Kim, I., and Barnett, S. A. (2006). High-rate electrochemical partial oxidation of methane in solid oxide fuel cells. J. Pow. Sour. 161:460. Zhan, Z. L., Liu, J., and Barnett, S. A. (2004). Operation of anode-supported solid oxide fuel cells on propane-air fuel mixtures. Appl. Catal. A Gen. 262:255.

374

Synthesis Gas Combustion: Fundamentals and Applications

Zhang, X. G., Robertson, M., Yick, S., Deces-Petit, C., Styles, E., Qu, W., Xie, Y. S., Hui, R., Roller, J., Kesler, O., Maric, R., and Ghosh, D. (2006). Sm0.5Sr0.5CoO3+Sm0.2Ce0.8O1.9 composite cathode for cermet supported thin Sm0.2Ce0.8O1.9 electrolyte SOFC operating below 600 degrees C. J. Pow. Sour. 160:1211. Zhao, F., and Virkar, A. V. (2005). Dependence of polarization in anode-supported solid oxide fuel cells on various cell parameters. J. Pow. Sour. 141:79. Zhu, H. Y., Colclasure, A. M., Kee, R. J., Lin, Y. B., and Barnett, S. A. (2006). Anode barrier layers for tubular solid-oxide fuel cells with methane fuel streams. J. Pow. Sour. 161:413. Zhu, H. Y., and Kee, R. J. (2006a). Modeling electrochemical impedance spectra in SOFC button cells with internal methane reforming. J. Electrochem. Soc. 153:A1765. Zhu, H. Y., and Kee, R. J. (2006b). Thermodynamics of SOFC efficiency and fuel utilization as functions of fuel mixtures and operating conditions. J. Pow. Sour. 161:957. Zhu, H. Y., and Kee, R. J. (2007). The influence of current collection on the performance of tubular anode-supported SOFC cells. J. Pow. Sour. 169:315. Zhu, H. Y., and Kee, R. J. (2008). Modeling distributed charge-transfer processes in SOFC membrane electrode assemblies. J. Electrochem. Soc. 155:B715. Zhu, H. Y., Kee, R. J., Janardhanan, V. M., Deutschmann, O., and Goodwin, D. G. (2005). Modeling elementary heterogeneous chemistry and electrochemistry in solid-oxide fuel cells. J. Electrochem. Soc. 152:A2427.

Index A Acid gas scrubbing cold gas cleanup, 20, 21, 22–23 pollutant production and control, 186 Acoustic measurements, blowout, 265 Acoustic oscillations, combustion instability, 273, 274, 275, 278 Activated carbon, mercury sorbents, 186 Activated complex formation, explosion characteristics, 41 Activation energy catalytic combustion, 224 diluent effects, 89 explosion characteristics, 35, 36 Activation overpotentials, SOFCs, 339–341, 343 Adiabatic calculations, stoichiometric hydrogen– oxygen system, 33, 34 Adiabatic flame temperature combustion in premixed systems diluent effects, 107, 109, 110, 111–113 H2/CO composition and, 106, 107 pressure and temperature effects, 114, 115–116 internal combustion engines, dual fuel with syngas and diesel, 320 laminar flame properties, 72–74, 88 Adiabatic temperature profiles, conventional hydrocarbon fuels, 106 Adsorption reactions, catalytic combustion of syngas, 236, 237 Aerodynamic effects on flame properties, 142 AFC (alkaline fuel cell), 210 Air, gasification technology, 2, 3 Air-blown gasifiers, diffusion flame combustors, 201 Air intake SOFC, heating with recovered exhaust, 352–353 staged fuel injection, 177 Air intake fumigation, 320, 324 Air separation unit (ASU), diluent options, 203 Air staging, NO control, 176 Air velocity, in premixer, 198 Airflow iron deposits on nozzle and, 55 SOFCs, 330 Alkaline fuel cell (AFC), 210 Alkalis cold gas cleanup, 20, 21 SOFC effects, 346–347

syngas cleanup requirements, 19 trace elements, 185 Alumina catalysts, 234, 235, 352 American Society for Testing and Materials (ATSM) test methods, 4 Ammonia impurities biogas, catalytic combustion, 226, 230 cleanup/purification, 18, 20, 21, 197 and NOx emissions, 197 combustion conditions and, 173 emissions controls, 199, 204 formation mechanisms, 171, 172 in-furnace control, 176 postcombustion pollutant control methods, 176 Ammonia injection, selective catalytic reduction, 204 Ammonia production chemical synthesis with syngas, 216 NOx reduction, 178 Anchoring, flame blowout, 262 flame speed and, 154, 198 turbulent combustion, 145, 150, 154, 198 Anemometry, 158 Anode flow, fuel cells, 214 Anodes catalytic partial oxidation (CPO), 343 impurity effects, 346–347 materials, 349 membrane-electrode assemblies, 332, 333, 335–336 planar stacks, 350 Antimony, 185, 346 Argon/argon mixtures activation energy determination, 89 premixed flame propagation in high-pressure media, 57, 58, 60 Arrhenius equation, ignition delay models, 299–300 Arsenic, 185, 346 As received (as) coal, 4 Ash; See Slag/ash Ash fusion temperatures, 4 Atmospheric conditions explosion characteristics of H2-containing systems, 41 premixed flame propagation in high-pressure media, 52–53

375

376 Autocorrelation function, turbulent combustion, 134 Autoignition combustion characteristics, 100–101 defined, 262 fuel properties and, 261 internal combustion engines dual-fuel compression ignition (CI) combustion, 305 hydrogen-assisted, 291, 294 operability issues, 278–284 premixed syngas combustion, 202 Axial flow velocity combustion instability, 278 flashback, 268–269 turbulent combustion, 148 Axial profiles, catalytic combustion, 233 carbon monoxide/air mixtures, 246, 248, 250–251, 254 hydrogen/air mixtures, 240, 241

B Back mixing, explosion characteristics of H2-containing systems, 39–40 Backward reaction rates, diluent effects, 107 Barium strontium cobalt iron oxide (BSCF), 348–349 Barrier layers catalytic combustion, 232 SOFCs, 348, 349, 354, 369 Batch-operated fuel processor, engine, 205 Bending effect, turbulent combustion, 151 Best-available control technology (BACT), 198 Biodiesel, 316 Biogas catalytically stabilized thermal (CST) combustion, 226 heat release rates, 319, 320 reciprocating internal combustion engines, 205 Biomass ash properties, 7 coal-biomass systems, 8 Biomass fuels catalytic combustion catalysts, 234 circulating fluid bed systems, 13 feedstock preparation, 8 feedstock properties, 5 fluid bed gasifiers, 12–13 SOFCs, 330 modeling, 365, 366–367, 368 performance with syngas from, 365–367, 368 production of, 331 trace elements, 185 Bituminous coal, 5, 6, 13

Index Blast furnace gas, 2 Blending, and PSR ignition/extinction, 103–104 Blow-off, 80, 202 Blowout, 30; See also Extinction catalytic partial oxidation (CPO), 228 extinction strain rate and, 92, 263 operability issues, 261, 262–268 operating conditions, 202 syngas composition and, 100, 228 Boilers carbon monoxide emissions, 182 circulating fluid bed systems, 13 sulfur pollutants, 180 VOC emissions, 182 Bomb apparatus, combustion characteristics, 99–100 Bond breaking, diluent decomposition, 107 Boron, 185 Bosch smoke unit (BSU), 301 Bottoming cycle, 212, 217 Boundary conditions/regions explosion characteristics of H2-containing systems, 38 flashback in, 270–271 turbulent combustion, 137, 145, 151, 153, 154 Brake-specific NOx emissions, 316 Brake thermal efficiency, 301 Branching reactions diluent effects, 90 laminar flame properties, 84–85, 90 reaction mechanisms and kinetics explosion characteristics of H2-containing systems, 31–32, 33–34, 35, 38 premixed flame propagation in high-pressure media, 57, 59, 61 Brayton cycle, 213 Briquettes, 10 Brush, flame; See Turbulent combustion properties, premixed syngas Bubble, vortex breakdown, 277 Bunsen flames, 156, 157, 159–160, 162 Burning rates/speed/velocity blowout modeling, 262 combustion characteristics, 99, 100–101 laminar flame properties flame propagation, pressure and, 82–83, 86 flame speed studies, 80 preheat temperature effects, 88 pollutant production, NO control, 177 premixed flame propagation in high-pressure media, 52, 56–58 contaminants and, 62–63 iron deposits and, 55 model predictions, 53–54, 59, 63 and turbulent combustion, 149

Index Butler-Volmer equation, 339, 340–341 Button-cell configuration, SOFC modeling, 355, 357

C Carbon feedstock properties, 5 gasification technology, 2 Carbon bed, cold gas cleanup, 20, 22 Carbon conversion rate, 4, 6 Carbon deposits fuel cell anode, 212 SOFCs, 342–343, 349 modeling, 354 syngas cogeneration, 354 thermal and heterogeneous catalytic chemistry, 343, 344–345 Carbon dioxide biogas composition, 205 cold gas cleanup, 20, 21 explosion characteristics of H2-containing systems, 39 natural gas composition, 195 nonpremixed flame properties, 95 syngas cleanup requirements, 19 syngas composition, 194 warm gas cleanup, 24 water–gas shift; See Water–gas shift Carbon dioxide diluents; See also Dilution/ diluents diffusion flame combustors, 201 flame propagation, 88 flame properties, 89, 90 nonpremixed flame properties, 95–96 oxy-fuel, 206, 207 Carbon dioxide production chemical looping systems, 208–209, 210 combustion characteristics; See Combustion characteristics dual-fuel compression ignition (CI) combustion, 301 flame chemical structures nonpremixed flames, 95–96 premixed flames, 75, 76 gasification technology, 2, 3 hydrogen-assisted SI methane combustion, 311 mass fraction in burned gases, 107, 108, 109, 111 oxy-combustion, 206 pollutant formation and control, 187–188 SOFCs, 214, 330, 331 Carbon dioxide separation, 203 Carbon dioxide sequestration/carbon capture, 2, 20, 25, 215, 216, 290, 330 chemical looping system products, 209, 216

377 oxycombustion products, 206, 208 SOFCs, 330, 331, 334, 346, 368, 369 Carbon monoxide infrared spectra, 55, 56 inhibitor effect in H2/CO autoignition, 100 pure, adiabatic flame temperature, 72, 73 SOFCs charge transfer pathways, 342 hydrogen formation, 343 modeling, 355 syngas composition, 331 syngas variability, 194 waste gas composition, 205 Carbon monoxide combustion catalytic combustion, 238, 246–247 combustor design and development, 194 flame temperature comparisons, 196 laminar flame properties diluent effects, 89–90 flame structure, 75, 76–77 oxy-combustion, 207 pollutant production and control carbon dioxide, 187 VOCs, 182, 183, 184 premixed systems, 100, 104–106, 107 reaction mechanisms and kinetics, 40, 42 chemical reaction R13: CO + HO2 = CO2 + OH, 40, 41, 43, 58, 63 contaminants affecting burning rate, 55, 62–63 explosion characteristics of H2-containing systems, 31 theory–experiment disparities, 43–44 reburning, 178, 179 Carbon monoxide production decomposition of lubricating oil, 185 emissions requirements, 199–200 engine emissions dual-fuel compression ignition (CI) combustion, 301 hydrogen-assisted SI methane combustion, 308, 310, 311 entrained flow gasifiers, 14 exhaust aftertreatment, 204 pollutant formation and control, 180–182, 188 Carbonyls cleanup process, 23 cold gas cleanup, 21, 23 and ignition process, 284 syngas cleanup requirements, 19 Carriers, chemical looping systems, 210 Cascading hydrogenation, VOC emissions, 183 Catalysts catalytic combustion, 234–235 Fischer-Tropsch process, 215 reactivity, 6

378 selective catalytic reduction; See Selective catalytic reduction SOFCs, 335 electrochemistry, 336 membrane-electrode assemblies, 333 tail gas, 352 thermal and heterogeneous catalytic chemistry, 343 Catalytic aftertreatment, 194 Catalytic chemistry emissions controls from catalytic reactor, 182 engine exhaust, 204, 215 syngas combustor design, 194 pollutant control, 176, 194 SCRs, 176, 197, 199; See also Selective catalytic reduction SOFCs, 343–345; See also Catalytic partial oxidation, SOFCs sulfur pollutants, 180 surface processes combustion characteristics studies, 104, 125–126 induction chemistry sensitivity to, 62 syngas cleanup/purification, 18, 19, 194, 197, 199 Catalytic combustion, 223–256 catalysts, high-temperature applications, 234–235 emissions, 229–230 physicochemical processes, 223–226 power generation systems, 226–229 SOFC systems integration, 352–353 syngas, 235–256 carbon monoxide/air mixtures, 246–247 hydrogen/air mixtures, 239–245 hydrogen/carbon monoxide mixtures, 247–256 numerical results, 236–239 thermal management, 230–233 Catalytic converters, 176 Catalytic partial oxidation (CPO) reaction mechanisms, 227–230 SOFCs, 331, 343–344 modeling, 356 performance with syngas from, 356, 358–361 systems integration, 352–353 thermal management, 232–233 Catalytic steam reforming, SOFCs, 343 Catalytically stabilized thermal (CST) combustion, 226–227, 230, 231, 242 Cathodes membrane-electrode assemblies, 332 planar stacks, 350 SOFCs

Index materials, 348–349 membrane-electrode assemblies, 333–334 Cell potential and overpotentials, SOFCs, 337–338 Center of mass of flame, 275, 276, 277 Center of mass of flame front, 274 Ceramic reactors, catalytic combustion, 223, 235 Ceria-based electrode materials, gadoliniumdoped ceria (GDC), 335, 342, 343 Ceria-based electrolytes, 348 Cermets, 347, 349 Cetane number, 299, 318 Chain branching; See Branching reactions Chain-breaking reactions, turbulent combustion, 136 Chain propagation, 31, 33, 35 Chain termination catalytic combustion of syngas, 236 explosion characteristics of H2-containing systems, 31, 33, 35 flame propagation, 84 Channel flow catalytic combustion, 231, 232, 246 solid oxide fuel cells, 341 Channel geometry, catalytic combustion simulations, 236 Charge transfer, SOFCs, 339, 340–341, 342–343, 355 Chemical induction, thermal chain explosive regime, 51 Chemical kinetics; See Combustion characteristics; Reaction mechanisms and kinetics Chemical looping combustion (CLC) systems, 208–210, 214 Chemical production cleanup requirements, 19 utilization of syngas, 214–216 Chemical solvents, acid gas scrubbing, 21, 22 Chemical structure combustion in counterflow diffusion flames, 118–119 complex mixtures, 75 flame properties, 74–77 laminar flame propagation, 74–76 Chemical time, and combustion instability, 274 Chemical zones, 119 Chemiluminescence measurements, 265, 276 CHEMKIN, 50, 101, 114, 178 CHEMKIN II, 49, 50, 104 CHEMKIN OPPDIF, 90, 95 CHEMKIN PREMIX, 74 CHEMSHOCK, 49 Chlorine products gasification technology, 3 pollutant production and control, 186

Index CHR (closed homogeneous reactors), 101–104 Chromium carbonyls, 54 Chromium oxide, 215 Circulating fluidized bed (CFB) boilers, 169, 170 CIVB-induced flashback, 272, 273 CLC; See Chemical looping combustion systems Cleanup/purification, 18–24, 195 cold gas cleanup, 20–23 acid gas scrubbing, 22–23 tars and carbonyls, 23 fuel cell systems, 212 fuel purity, 197 for SOFCs, 346 syngas, 188, 194 tar removal, 186 warm gas cleanup, 23–24 Closed homogeneous reactors (CHR), ignition and extinction in, 101–104 Coal-biomass systems, 8 Coal-derived fuels, 2 gasification chemical looping systems, 208–209 feedstock preparation, 7 feedstock properties, 4, 5 reactivity, 6 high-pressure coal–gas fed diesel engines, 205 pollutant production and control chlorine removal, 186 emissions controls, 199 mercury sorbents, 186 NOx, 174–175, 179 reburning effects simulation, 179 SOFCs, 330, 331 impurity effects, 346–347 modeling, 365–366, 367 performance with syngas from, 365–367, 368 Coal-fired power plants, carbon monoxide production, 180 Coal tar, 10 Cobalt catalysts, Fischer-Tropsch process, 215 Co-current (downdraft) moving bed gasifiers, 9, 10 Cogeneration of syngas, SOFCs, 353–354 Coke formation, SOFCs, 343, 349 Cold cleanup, 346 Cold gas cleanup, 19, 20–23 Cold gas efficiencies, 15, 16 Collision parameters/phenomena catalytic combustion of syngas, 236 explosion characteristics, 32, 42 flame propagation, pressure and, 81 premixed flame propagation in high-pressure media, 59

379 Collisional efficiencies, 39 Combined cycle, chemical looping, 209–210 Combustion circulating fluid bed systems, 12–13 gasification reactions, 3 Combustion chamber combustion instability, 274 iron deposits on nozzle and, 55 turbulent combustion, 150–151 Combustion characteristics, 99–126; See also Reaction mechanisms and kinetics catalytic combustion; See Catalytic combustion counterflow diffusion flames, 114, 116, 117–124 diluent effects, 117–120, 121, 122 H2/CO mixtures, 117 pressure effects, 120–121, 123, 124 diffusion flame combustors, 200–202 flame properties; See Laminar flame properties and pollutant production; See Pollutant formation and control premixed CHR and PSRs, ignition and extinction in, 101–104 premixed syngas combustion, 202–203 premixed systems, 104–114, 115–116 CO/H2 mixtures, 104–106, 107 diluents and, 107–112, 113, 114 pressure and temperature effects, 112–114, 115–116 strained counterflow premixed flames, 124, 125 turbulence; See Turbulent combustion properties, premixed syngas Combustion-induced vortex breakdown (CIVB)induced flashback, 272, 273 Combustion instability defined, 262 fuel properties and, 261 operability issues, 273–278 Combustion pulsation-induced flashback, 270 Combustor configuration design and development considerations, 194 gas turbines, 200–203 Combustor pressure drop, 198 Competing reactions explosion characteristics of H2-containing systems, 38 flame structure, 75 Complex potential energy surface, kinetic models, 40 Compressible flow combustion characteristics studies, 104 induction chemistry sensitivity to, 62

380 Compression ignition (CI) fuels, hydrogenassisted and dual-fuel combustion, 296–302 diesel with hydrogen experimental configuration, 304, 305 results and discussion, 314–317 diesel with syngas experimental configuration, 304–306 results and discussion, 317–324 ignition delay, 298–300 performance and emissions, 300–302 Compressive stretch, stabilizing mechanisms, 140 Compressor surge, flashback, 270 Computational singular perturbation (CSP) analysis, 41, 42 Concave flame curvature, 140 Concentration overpotentials, SOFC, 341–342 Condensation metals, 186 particulate classification, 186 Conical flame, 146, 151 Constant enthalpy and pressure (CONP), shock tube modeling, 49 Constant internal energy and volume (CONV), shock tube modeling, 49 Constant pressure dual-chambered bomb, 52 Constant volume adiabatic calculations, stoichiometric hydrogen–oxygen system, 33, 34 Consumption speed, 147, 151, 152, 153 Contaminants biogas, 205 carbon monoxide, 54, 62–63 syngas gasification technology, 36 and ignition, 284 induction chemistry sensitivity to, 62–63 and NOx emissions, 173, 175 removal before combustion, 188, 194; See also Cleanup/purification Convective heat transfer, oxy-combustion, 206 Convective time, 278 Convective time delay, 273, 274, 275 Conventional fuels combustion characteristics, 100, 101, 106 emissions, 188, 200 CO, 180, 181 NOx, 172, 177 trace elements, 185, 186 firing temperature, 201 flame speed and adiabatic temperature profiles, 106 internal combustion engines, 205, 290, 291, 297, 302, 314, 324 laminar flame properties, 73, 77 power generation systems, 226 SOFC electrocatalytic oxidation versus, 354

Index Convex flame curvature, 140, 163 Conveyance gas, 8 Cooling catalytic combustion, 242 fuel cells, 212–213, 214 thermal management, 17 Copper anodes, SOFCs, 343 Copper/zinc oxide/chromium oxide catalysts, 215 Core flow, flashback turbulent flame propagation in, 268–270 vortex breakdown-driven flame propagation in swirling flow core, 271–273 Correlation functions, turbulent combustion, 133, 134, 143, 149, 150 Corrosion, oxygen in exhaust stream and, 206 COS, 20, 21, 23, 197 Countercurrent (updraft) moving bed gasifiers, 9, 10 Counterflow, flame speed studies, 80 Counterflow diffusion flame combustion characteristics diluent effects, 117–120, 121, 122 extinction limit, 126 H2/CO mixtures, 117 pollutants, NOx, 173 pressure effects, 120–121, 123, 124 Counterflow premixed flames, strained, 124, 125 CPO; See Catalytic partial oxidation Cracking tars, 186 transport reactors, 13 Critical branching factor, 31 Critical geometries, R13 kinetic models, 40 Critical residence time, combustion characteristics studies, 125 Critical velocity gradient, 270–271 Critical wall gradient, 271 Cross-correlations, turbulent combustion components, 131, 132, 133, 134, 162 Cryogenic oxygen plant, 15 CST; See Catalytically stabilized thermal combustion Curvature, flame, 90, 139, 142, 162, 163 probability density functions, 141–142 stabilizing mechanisms, 140 stretched flame model, 139, 141–142 wrinkled flamelet insensitivity to, 139 wrinkle scale quantification, 141 Cycle efficiency, chemical looping, 209–210 Cyclones, 20, 21 Cylindrical bomb, 52

D Damköhler number, 263, 264, 268 Damköhler scaling laws, 143 Darcy flow, 345

Index DDC turbodiesel engine, 304 Decomposition diluents, 107 engine oils carbon monoxide emissions, 182 carbon monoxide production, 181 VOC emissions, 185 Degenerate diffusion reaction sheet, catalytic combustion, 224 Density, energy, 48–49 Density, mixture flame propagation, pressure and, 81, 83 and flame thickness, 77 preheat temperature effects, 88 Desorption reactions, catalytic combustion of syngas, 236, 237 Detached flames, 157, 267 Detailed modeling of counterflow flame (DMCF) code, 114, 124 Detonation, 45, 51 Diesel duel-fuel engines, 205, 296–302 with hydrogen experimental configuration, 304, 305 results and discussion, 314–317 ignition delay, 298–300 performance and emissions, 300–302 with syngas experimental configuration, 304–306 results and discussion, 317–324 Diffuser, high-velocity premixer flow slowing, 198 Diffuser burner, 152–153 Diffusion blowout modeling, 265 catalytic combustion, 233 of diluents, 107 laminar flame properties, 74 flame speed, H2/CO ratio and, 81 flame stretch, 90 flame thickness, 77, 80 strain and, 92 nonpremixed flame properties, 96 thermal/diffusive effect; See Lewis number turbulent combustion, 131, 140 Diffusion combustion dual-fuel compression ignition, 319, 321, 323 nitrogen diluent and, 197 Diffusion flame combustors counterflow diffusion flame combustion characteristics diluent effects, 117–120, 121, 122, 126 extinction limit, 126 H2/CO mixtures, 117 pollutants, 173 pressure effects, 120–121, 123, 124 emissions controls, 199 gas turbines, 194, 200–202

381 Diffusion time scales, explosion characteristics of H2-containing systems, 31, 33 Dilute-phase solids transport, feedstock preparation, 8 Dilution/diluents combustor design and development, 194 counterflow diffusion flames, 117–120, 121, 122 diffusion flame combustors, 200, 201–202 flame propagation in high-pressure media, 56 flame temperature comparisons, 196 fuel composition, 196, 197 gas turbines, 203–204 and laminar flame properties, 88–90 adiabatic flame temperature, 73 extinction strain rates, 95 flame propagation, 80 nonpremixed flames, 95, 96 and NOx emissions, 56, 173, 199 oxy-fuel, 206, 207 premixed system combustion characteristics, 102, 107–112, 113, 114, 126 strained counterflow premixed flames, 124 Direct hydrocarbon feeds, SOFCs, 330 Direct methanol fuel cell (DMFC), 210 Direct numerical simulations (DNS), turbulent combustion, 151, 161 Displacement speed, 147–148, 149, 150, 151–152, 155 Dissipation combustion instability prediction, 277 turbulent combustion, 130, 131, 133, 137 eddy cascade hypothesis, 132 superadiabaticity, 163 Dissipation rate, 163 Dissociation-recombination, catalytic combustion of H2/CO, 238 Distributed energy, reciprocating internal combustion engines, 204–205 DLN (dry-low-NOx) gas turbines, 129, 130, 142 DMCF (detailed modeling of counterflow flame) code, 114, 124 DMFC (direct methanol fuel cell), 210 Dodecane, 356, 358–361, 362 Downdraft (co-current) moving bed gasifiers, 9, 10 Downflow, entrained flow gasifiers, 14 Dry ash free (daf) coal, 4 Dry-low-NOx (DLN) gas turbines, 129, 130, 142 Dry oxidation, 100 Drying, feedstock preparation, 7 Dual-fuel diesel applications; See Diesel duelfuel engines Dual-fuel engines; See Internal combustion engines, hydrogen-assisted combustion Dusty-gas model, 345, 355 Dynamic stability, combustion characteristics studies, 100

382 E Eddies, turbulent combustion, 132, 133, 134, 136, 138, 139 Eddy cascade hypothesis, 132 Eddy viscosity, 131 Efficiency fuel cells, 214 process; See Process efficiency/energy inputs Electric power output, SI engine generator, 306, 307, 308 Electric power production; See Integrated gasification combined-cycle systems Electrochemical partial oxidation (EPOx), 353–354 Electrochemistry, SOFCs, 330, 337–343 activation overpotentials, 339–341 cell potential and overpotentials, 337–338 charge transfer pathways, 342–343 concentration overpotentials, 341–342 ohmic overpotential, 338–339 Electrolyte membrane, SOFCs, 334–335 Electrolytes impurity effects, 347 membrane-electrode assemblies, 332–333 planar stacks, 350 SOFCs, 347–348 Emissions catalytic combustion, 223, 229–230 dual-fuel compression ignition (CI) engines, 300–302 hydrogen and diesel, 314–317 syngas and diesel, 320 exhaust aftertreatment, 194, 204, 206 gas turbines, 194, 199–200 hydrogen-assisted spark ignition (SI) engines, 290, 291, 307–308, 309–310, 311 SI engine generator, 309 Emissions controls; See Pollutant formation and control Endothermic gasification, 4, 14 Endothermic reforming, SOFCs, 336 Energy chemical looping systems, 209 diluent decomposition, 107 diluent effects, 90 flame propagation, pressure and, 81 ignition studies, 48–49 turbulent combustion, 133–134 Energy generation; See Power generation Energy inputs; See Process efficiency/energy inputs Engines; See also Internal combustion engines carbon monoxide production, 180, 181 performance

Index NOx in exhaust gas recirculation stream and, 36 premixed flame propagation in highpressure media, 56 VOC emissions, 182 Engine sizes, reciprocating internal combustion engines, 205 Enthalpy, shock tube modeling, 49 Entrained flow, slagging gasifiers, 13–14, 16 Entrained flow gasifiers high-ash coal and, 4 thermal management, 16, 17, 18 Entrained solids, fluid bed gasifiers, 11 Envelop flames, 151, 153 Ethane, 195 Exergy, 19 chemical looping systems, 209 fuel cells, hybrid, 213 Exhaust aftertreatment combustor design and development, 194 gas turbines, 204 oxy-combustion, 206 Exhaust gas recirculation (EGR); See Recirculation Exhaust heat recovery, SOFCs, 352–353 Exhaust recycling, SOFC anode protection, 349 Exit temperature, cold gas cleanup, 21 Experimental perturbations and nonidealities, 45 Explosion characteristics, H2-containing systems, 30–37 Explosion limits; See also Extended second limit autoignition, 280–281, 283 hydrogen–oxygen mixtures, 31, 32 NOx effects, 36–37 premixed flame propagation in high-pressure media, 57 Explosive regimes, thermal chain, 51 Extended second limit explosion characteristics, 35, 38–39, 41 ignition studies, 45 premixed flame propagation in high-pressure media, 57, 60 shock tube pressure histories, 46 Extinction blowout modeling, 262, 263, 265, 268 catalysts, 234 combustion characteristics studies, 101–104, 126 hydrogen-assisted SI methane combustion, 308 laminar flame properties, 72, 92–95 Extinction limits, 126 Extinction strain rate and blowout, 92, 263 laminar flame properties, 92, 93, 94–95

Index F Feedstocks gasification technology, 1, 2, 3–8 ash composition, 6–7 preparation of feedstock, 7–8 reactivity, 4, 6 gasifier types and comparison of gasifier types and approaches, 15, 16 feed systems, 8 Fenimore (prompt) NOx, 172, 229 Fiber biomass, 5, 8 Filters cold gas cleanup, 21 particulate removal, 185, 186 purification of syngas, 20 Fischer-Tropsch (FT) process, 214–215, 216 moving bed gasifiers, 10 purification of syngas, 18 Flame acceleration in spinning tubes, 272 Flame anchoring; See Anchoring, flame Flame center of mass, 274–275, 276, 277 Flame confinement, catalytic combustion, 240 Flame crossing length, 144 Flame detachment, blowout, 267 Flame expansion, operability issues, 261 Flame extinction; See Extinction Flame front, 162 center of mass, 274 gas dilatation along, 272 heat release rate and turbulent flame speed, 143 turbulent combustion, 138–139 Flame front surface area, 142, 143 Flame geometry; See also Flame structure combustion instability, 277 diffusion flame combustors, 201 laminar flame properties, 77–80 stretch effects, 91 turbulent combustion, 151, 153–154 Flame kernel, turbulent combustion, 148, 149–150, 156, 157 Flame lift-off position, 160 Flame propagation combustion characteristics studies, 100, 101 flashback, 268–269 turbulent flame propagation in, 268–270 vortex breakdown-driven flame propaga­ tion in swirling flow core, 271–273 in high-pressure media, 52–61, 62 laminar flame properties adiabatic flame temperature and, 72 lean mixtures, 75 operability issues, 261

383 Flame propagation speed in high-pressure media, 52 laminar flame properties diluent effects, 89 hydrogen levels and, 75 pressure and, 84 Flame properties; See also Laminar flame properties catalytic combustion; See Catalytic combustion combustion characteristics studies, 100 combustion instability, 277–278 diluent effects, 89–90 fundamental combustion characteristics; See Combustion characteristics hydrogen-assisted internal combustion, 291 premixed syngas combustion, 203 Flame residence time, diffusion flame combustors, 202 Flame speeds combustion in premixed systems, 99, 100, 101 diluent effects, 107, 108, 109, 111–113 equivalence ratio and, 104, 105 H2/CO composition and, 106, 107 pressure and temperature effects, 112–113, 114, 115–116 combustion instability, 277 flashback, 268–269 in boundary layer, 270–271 in premixer, 198 laminar flame properties diluent effects, 89–90 and flame propagation, 80 flame propagation, pressure and, 82, 83 flame stretch and, 90 H2/CO ratio and, 81 preheat temperature effects, 86, 87, 88 strain and, 92 stretch effects, 90, 91 premixed flame propagation in high-pressure media, 52 turbulent combustion, 134–135, 142–154 correlation of, 149, 150 diffuser burner, 152–153 experimental configurations of laboratory burners, 145, 146, 147, 148 factors affecting, 126 flame kernel, 149–150 flame surface area ratio, 142, 144 flame surface density and stretch factor measurement, 144–145 global displacement speed, 149 heat-release rate and, 142–154 hydrogen and, 156–157

384 idealized one-dimensional double infinite planar turbulent flame, 143–145 scaling laws, 143–144 stagnation and low swirl barrier studies, 150–151, 152 syngas and hydrogen effects on, 154–157 Flame stability; See Stability, flame Flame structure combustion in counterflow diffusion flames, 117, 119–120, 126 combustion in premixed systems diluent effects, 107, 126 flame propagation in high pressure media, 61 pressure and temperature effects, 115–116 spatial variations of major and minor species for different equivalence ratios, 105, 106 laminar flame properties in nonpremixed systems, 96 laminar flame properties in premixed systems pressure effect on flame propagation, 83–84 reaction mechanisms, 74–77 stretched flame model; See Stretch, flame turbulent combustion, 135–136 reaction to curvature and strain, 139–140 surface area, flame speed calculations, 152 surface area ratio, 142, 144 surface density measurement, 144–145, 147–148, 152 wrinkle structure, 134–141 Flame surface integral, 144–145 Flame temperature combustion in premixed systems diluent effects, 107, 109, 111–113 factors affecting, 126 combustor design and development, 194 diffusion flame combustors, 200 diluent effects, 88 flame extinction, 89, 92, 93 flame propagation, pressure and, 81–82 gasoline, hydrogen, and methane, 294 strain and, 90, 91, 92 Flame volume, dilution and, 201 Flamelet description blowout modeling, 262, 263 stretched flamelet, 139, 144, 163, 164 wrinkled flamelet, 135–141, 144, 163, 164 Flammability limits carbon monoxide production, 181 combustion characteristics studies, 101 extinction, 93–94 flashback in boundary layer, 271 Flashback, 30, 37, 52

Index catalytic partial oxidation (CPO), 229 combustion characteristics studies, 100 defined, 262 determinants of, 80 flame speed and, 198 fuel properties and, 261 hydrogen effects on, 160, 229 lean combustion and, 202, 229 operability issues, 268–273 premixed syngas combustion, 202 in premixer, 198 turbulent combustion, 153–154, 160 Flashback limit, 160 Flow fields flashback in boundary layer, 271 stationary premixed turbulent flames, 146, 147 turbulent combustion, 155 Flow properties SOFCs, porous media transport, 345 turbulent combustion, 133, 152 Flow rate diffusion flame combustors, 201–202 heating value of fuels and, 197 premixed syngas combustion, 202–203 Flow reactors combustion characteristics studies, 102 ignition studies, 43 theory-experiment disparities, 62 Flow recirculation, turbulent combustion, 154 Flow velocity blowout modeling, 266 combustion pulsation-induced flashback, 270 turbulent combustion, 158 Flue gas purification of syngas, 18 trace element removal, 186 Flue gas recirculation (FGR); See Recirculation Fluid bed gasifiers comparison of gasifier types and approaches, 16 particulates, 186 reactivity, 6 types of, 10–13 bubbling bed, 11–12 circulating, 12–13 transport reactor, 13 Fluid dynamics, shock tube, 45 Fluid transport, catalytic combustion, 242 Fly ash, 21 Forced flame, combustion instability, 276–277 Formaldehyde, 183 Fouling; See also Carbon deposits; Sulfur impurities catalysts, 204 heat exchangers, 18

385

Index Fourier transform infrared spectroscopy, 55, 56 Frequency combustion instability, 278 combustor, 275 turbulent combustion, 134 Fuel conventional; See Conventional fuels diluent options, 203 gasification technology, 3 Fuel/air ratio oscillations, combustion instability, 273–274 Fuel cells carbon monoxide emissions, 182 SOFCs; See Solid oxide fuel cells syngas cleanup requirements, 19 utilization of syngas, 210–214 Fuel composition and combustion instability, 274, 275 premixed syngas combustion, 203 Fuel heating value; See Heating value Fuel injection, staged (reburning), 177–179 Fuel jets, flashback, 270 Fuel NO, 171, 176 Fuel pressure drop and mixing, gas turbines, 197–198 Fuel production, utilization of syngas, 214–216 Fumigation, diesel engine intake air, 320, 324

G Gadolinium-doped ceria (GDC), 335, 342, 347, 348 Gas hourly spatial velocities (GHSVs), 234 syngas purification, 18–24 cold gas cleanup, 20–23 warm gas cleanup, 23–24 Gas-liquid transformation temperature, trace element condensation, 186 Gas-phase chemistry catalytic combustion, 225, 237 carbon monoxide/air mixtures, 251 chain terminating reactions, 236 hydrogen/air mixtures, 239–242, 243, 244 ignition, 224, 225 explosion characteristics of H2-containing systems, 31–32 ignition studies, 43–44 Gas-phase impurities, SOFC effects, 346 Gas quench approach cold gas cleanup, 21 thermal management, 16, 17, 18 Gas turbine conditions, 61, 157, 164 Gas turbine cycle, fuel cells, 213, 214 Gas turbines carbon monoxide production, 180 catalytic combustion, 224 ignition delay, 283

iron deposits, 54–55 pollutant production and control mercury, 185–186 sulfur, 180 VOC emissions, 182 turbulent combustion, 142, 154 utilization of syngas, 193–204 Gaseous feedstocks, 3 Gases SOFCs, porous media transport, 345 transport reactors, 13 Gasification technology, 1–25, 214–216 feedstock properties, 3–8 ash composition, 6–7 preparation of feedstock, 7–8 reactivity, 4, 6 gasifier types, 8–13 comparison of gasifier types and approaches, 15–16 entrained flow, slagging gasifiers, 13–14, 16 fluid bed gasification, 10–13, 16 moving bed gasification, 9–10, 16 thermal management, 16–18 general process and major reactions, 2–3 history, 2 hybrid chemical looping combustiongasification system, 209 pollutant production and control, particulates, 186–187 Gasifiers, diluent options, 203 Gasoline combustion hydrogen and methane impacts on, 291, 294–295 physical and chemical properties, 294 Generation of syngas, SOFCs, 353–354 Generator, SI engine electrical output, 306, 307, 308 emissions, 309 Geometry, flame; See Flame geometry Geometry, reactor, 241, 242 Global consumption speed, 147, 151 Global displacement speed, 148, 149 Glow-plug-assisted diesel engine, 205 Gold anodes, SOFCs, 343 Gradients, autoignition, 278 Graphitic-like coke formation, SOFCs, 343 GRI-Mech 3.0, 101, 263, 279 Grid turbulence, 161 Grinding, feedstock preparation, 8

H Halide impurities cold gas cleanup, 20, 21 gasification technology, 3

386 pollutant production and control, 186 SOFC effects, 346, 347 syngas cleanup requirements, 19 HCO reactions catalytic combustion of H2/CO, 238–239 reaction mechanisms and kinetics, 40, 58 Heat exchanger cold gas cleanup, 21 deposition of metals on, 186 syngas reheating after purification, 195 thermal management, 18 Heating rates, transport reactors, 13 Heating value, 197–198 and adiabatic flame temperature, 73 carbon dioxide emissions calculation, 187–188 chemical looping systems, 209 combustor design and development, 194 feedstock properties, 5 and flow rate, 197 fuel composition, 194–195 gasoline, hydrogen, and methane, 294 thermal management, 16, 17, 18 variability of syngas, 194 Heat of vaporization, syngas cooling and, 19 Heat recovery; See also Recirculation exhaust aftertreatment, 204 preheating reactants with waste heat, 157 SOFCs, 352–353 steam generation, 204 Heat release combustion characteristics, 126 combustion characteristics in counterflow diffusion flames, 117, 118 diluent effects, 119, 120 pressure and, 123 combustion characteristics in premixed systems ignition delay in perfectly stirred reactors, 101 strained counterflow premixed flames, 125 dual-fuel compression ignition (CI) combustion, 298 experimental curves, 306 peak values, 319–320, 321 syngas, 306 fuel cells, 212–213, 214 ignition studies, 45 laminar flame properties adiabatic flame temperature, 73 flame extinction, 92 flame propagation, pressure and, 83 flame stretch and, 90 flame structure, hydrogen levels and, 74, 75, 76

Index nonpremixed flame properties, 95 three body radical termination reaction, 75 operability issues and combustion instability, 262, 273–278 flashback, gas dilatation along flame front, 272 premixed syngas combustion, 203 shock tube, 45 syngas conversion to other fuels, 216 and turbulent flame speed, 142–154 diffuser burner, 152–153 experimental configurations of laboratory burners, 145, 146, 147, 148 flame kernel, 149–150 flame surface area ratio, 142, 144 flame surface density and stretch factor measurement, 144–145 global displacement speed, 149 idealized one-dimensional double infinite planar turbulent flame, 143–145 scaling laws, 143–144 stagnation and low swirl barrier studies, 150–151, 152 Heat transfer catalytic combustion, 225 diluent effects, 88 oxy-combustion, 206 turbulent combustion, 136 Helium mixtures, 57, 58, 60 Heterogeneous fuel depletion, catalytic combustion of syngas, 236 Hetero-/homogeneous chemistry coupling, 236, 247–250 Hexaaluminates, 234 High-ash coal, 4 High-energy density mixtures, ignition studies, 48–49, 62 High-pressure coal-gas fed diesel systems, 205 High-pressure conditions ignition, 40, 41, 43 premixed flame propagation, 52–61, 62 turbulent combustion, 157–161 High-pressure/low-temperature ignition, 43–52 modeling approaches, 49–51 other systems, 51–52 shock tube pressure histories, 46–49 High-temperature conditions catalytic combustion, 234–235 turbulent combustion, 157–161 Higher heating value (HHV), 198 feedstock properties, 5 natural gas composition, 195 Homogeneous charge compression ignition (HCCI), 290, 296 Homogeneous chemical reaction mechanism, 238–239

Index Homogeneous (gas-phase) ignition, 224 Homogeneous reactors (HR), 101–104 Honeycomb reactors, catalytic combustion, 223–224 catalysts, 235 hydrogen/air mixtures, 242 syngas, 236 Hot cleanup, 346 Hot spots, shock tube pressure histories, 45 Hot-wire anemometry, 158 Humid gas cleanup, 23–24 Hybrid combustion-gasification system, chemical looping, 209 Hybrid fuel cells, 213–214 Hydrocarbon emissions hydrogen-assisted SI methane combustion, 309–310, 311 VOC formation and control, 182–185 Hydrocarbon fuels conventional; See Conventional fuels SOFC charge transfer pathways, 342–343 Hydrocarbon reforming; See Reforming; Steam reforming Hydrogen carbon monoxide emissions, 182 and carbon monoxide production, 181–182 chemical looping systems, 208–209, 210 combustion characteristics studies, 100–101 diesel duel-fuel engines experimental configuration, 304, 305 results and discussion, 314–317 extinction strain rates, 94–95 feedstock properties, 5 and flame extinction, 71–72 and flame propagation speed, 75 and flame structure, 75, 76 flame temperature comparisons, 196 gasification technology, 3 internal combustion engines; See Internal combustion engines, hydrogenassisted combustion physical and chemical properties, 294 pollutant production and control, natural gas and syngas properties, 188 reburning with, 178 SOFCs, 330, 342 turbulent combustion, 154–157 variability of syngas, 194 waste gas composition, 205 Hydrogen, pure adiabatic flame temperature, 72, 73 carbon monoxide production, 181 diluent effects versus, 89 flame speed and adiabatic temperature profiles, 106 flame structure, 77

387 Hydrogen/air mixtures catalytic combustion, 231–232, 239–245 gas-phase chemistry impact, 239–242 light-off temperatures, 244–245 pressure and, 242–244 Hydrogen atoms laminar flame properties diluent effects, 89, 90 nonpremixed flames, 95 premixed flames, 74, 75, 76–77 pressure-dependence of flame propagation, 84–85, 86 reaction mechanisms and kinetics R1: H + O2 = O + OH, 38, 39, 57, 58, 60, 61, 354 R2: H + O2(+M) = HO2(+M), 33, 35, 36, 38, 39, 40, 57, 58, 61, 354 R6: HO2 + H = H2 +O2, 33, 35, 43, 57, 58, 59, 60, 61, 343, 354 R7: HO2 + H = OH + OH, 33, 35, 42–43, 57, 58, 59, 60, 61 R9: NO2 + H = NO + OH, 36 R17: H + OH + M = H2O + M, 57, 59, 61 spatial variations of major and minor species for different equivalence ratios, 105, 106 Hydrogen/carbon monoxide autoignition, 281–282, 283 catalytic combustion, 247–256 hetero-/homogeneous chemistry coupling, 247–250 light-off temperatures, 252–256 reaction schemes, 237 surface temperatures, 250–252 combustion characteristics counterflow diffusion flames, 117 ignition delay in PSR and HR, 102 ignition delay times, 283 laminar flame properties, 81, 82; See also Laminar flame properties reaction mechanisms and kinetics characteristic reaction times, 37 explosion characteristics of H2-containing systems, 36 kinetic model, recent and proposed updates, 37–43 premixed flame propagation in highpressure media, 53 shock tube ignition studies, 62 variability of syngas, 194 Hydrogen/carbon monoxide/carbon dioxide, 56 Hydrogen/carbon monoxide/oxygen, 40 Hydrogen/carbon monoxide/oxygen/helium, 53 Hydrogen/carbon monoxide/oxygen/nitrogen/ argon, 37, 40, 41

388 Hydrogen chloride cold gas cleanup, 20, 21 SOFC effects, 346–347 syngas cleanup requirements, 19 Hydrogen cyanide cold gas cleanup, 20, 21 fuel cleanup, 197 staged fuel injection, 178 Hydrogen-fueled engines; See Internal combustion engines, hydrogenassisted combustion Hydrogen fumigation, 320 Hydrogen/methane flames, turbulent flame speed, 154–155 Hydrogen/oxygen, homogeneous chemical reaction mechanism for H2/CO, 38 Hydrogen/oxygen/carbon monoxide/carbon dioxide diluent, burning rates, 52 Hydrogen/oxygen diluent, premixed flame propagation in high-pressure media, 56 Hydrogen peroxide catalytic combustion of H2/CO, 238 explosion characteristics of H2-containing systems, 35 homogeneous reaction mechanism for H2/ CO, 38 ignition studies, 45 R4: H2O2 (+M) = OH + OH (+M), 33, 35, 43, 100 spatial variations of major and minor species for different equivalence ratios, 105, 106 temperature/pressure dependencies, 279–280 Hydrogen production, 215–216 Hydrogen radicals; See also Hydrogen atoms NNH formation, 172 premixed flame propagation in high-pressure media, 60, 61 Hydrogen sulfide, 179–180 cold gas cleanup, 21 purification of syngas, 18 SOFCs, 347, 349 Hydrogenation, cascading, 183 Hydroperoxyl (HO2) radical, 279–280 catalytic combustion of H2/CO, 238 explosion characteristics of H2-containing systems, 35 laminar flame properties flame chemical structure, 75 flame structure, 76 preheat temperature effects, 87, 89 NO2-to-NO reaction, 317 reaction mechanisms and kinetics, 37 explosion characteristics of H2-containing systems, 32, 33–34 ignition studies, 45

Index premixed flame propagation in highpressure media, 59, 61 R3: HO2 + HO2 = H2O2 + O2, 33, 100, 354 R5: HO2 + H2 = H2O2 + H, 33, 35, 343 R6: HO2 + H = H2 +O2, 33, 35, 43, 57, 58, 59, 60, 61, 343, 354 R7: HO2 + H = OH + OH, 33, 35, 42–43, 57, 58, 59, 60, 61 R8: NO + HO2 = NO2 + OH, 36, 354 R13: CO + HO2 = CO2 + OH, 40, 41, 43, 58, 62 R14: HO2 + OH = H2O + O2, 41, 42, 43, 57, 58, 59, 60, 61, 63 rate constants, 42 spatial variations of major and minor species for different equivalence ratios, 105, 106 Hydroxyl (OH) radical map, 240 Hydroxyl (OH) radicals catalytic combustion, 244 explosion characteristics, 35 flame structure, 75, 76–77 internal combustion engines, 324 nonpremixed flame properties, 95 reaction mechanisms and kinetics R10: CO + OH = CO2 + H, 40, 42 R14: HO2 + OH = H2O + O2, 41, 42, 43, 57, 58, 59, 60, 61, 62 R15: H2 + OH = H2O + H, 57, 58 R17: H + OH + M = H2O + M, 57, 59, 61 shock tube modeling, 51 spatial variations of major and minor species for different equivalence ratios, 105, 106

I Idealized one-dimensional double infinite planar turbulent flame, 143–145 Ideal voltage, fuel cells, 211 IGCC; See Integrated gasification combinedcycle systems Ignition carbon monoxide production, 180, 181 catalytic combustion carbon monoxide/air mixtures, 246, 247, 252–255 hydrogen/air mixtures, 244–245 in CHRs and PSRs, premixed, 101–104 combustion characteristics studies, 104 high-pressure/low-temperature systems, kinetic implications, 43–52 modeling approaches, 49–51 other systems, 51–52 shock tube pressure histories, 46–49 high-pressure rapid compression machine, 40, 41

Index internal combustion engines; See Internal combustion engines, hydrogenassisted combustion shock tube pressure histories, 45 Ignition-assisted, high-compression diesel engines, 205 Ignition delay autoignition, 278, 279–284 combustion characteristics studies, 125 diesel duel-fuel engines, 298–300 dual-fuel compression ignition (CI) combustion, 317–318 experimental studies, 43, 44 induction chemistry and, 62 internal combustion engines, 320 in perfectly stirred reactors, 101–102 perturbations and, 51 prediction of, 38–39 premixed syngas combustion, 202 sensitivity to experimental perturbations, 62 shock tube, 45, 47, 50 Ignition timing; See Timing, ignition Impurities and catalyst performance, 235 chemical looping systems, 210 fuel cell syngas cleanup, 212 gasification technology, 3 induction chemistry sensitivity to, 62 and reactivity, 6 removal of; See Cleanup/purification SOFCs, 342, 346–347 In-furnace control, NOx, 175–176 Incomplete combustion carbon monoxide emissions, 180, 181–182 emissions controls, 200 extinction, 93 stretch effects, 90 Induction chemistry ignition studies, 45, 62 sensitivity to experimental perturbations, 51–52, 62 Industrial burners, turbulent combustion, 142, 154 Inhomogeneity, shock tube, 45 Inlet conditions catalytic combustion, 225, 232, 233 combustion in counterflow diffusion flames, 120, 123 combustion in premixed systems, 113, 114, 115–116 diffusion flame combustors, 200 emissions controls, 200 hydrogen/air combustion, 279–280 SOFCs, modeling, 365 strained counterflow premixed flames, 124 Instabilities

389 Lewis number; See Lewis number operability issues, 273–278 premixed syngas combustion, 202–203 pressurized laminar flames, 158, 159 Intake air fumigation, diesel engine, 320, 324 Intake-induced swirl, 311–312, 313, 314 Integral length scale, turbulent combustion, 133 Integral time scale, turbulent combustion, 133 Integrated gasification combined-cycle (IGCC) systems, 18 carbon monoxide production, 180 circulating fluid bed systems, 13 combustor pressure drop, 198 comparison of gasifier types and approaches, 14 emissions controls, 199 entrained flow gasifiers, 13 iron deposits, 54–55 pollutant formation and control, 169, 170, 188 mercury, 185–186 particulates, 186–187 sulfur pollutants, 180 thermal management, 18 warm gas cleanup, 24 Internal combustion engines exhaust gas recirculation in, 35 pollutants, 180 utilization of syngas, 204–206 Internal combustion engines, hydrogen-assisted combustion, 290–325 compression ignition (CI) fuels, hydrogenassisted and dual-fuel combustion, 296–302 ignition delay, 298–300 performance and emissions, 300–302 dual-fuel combustion, diesel with hydrogen experimental configuration, 304, 305 results and discussion, 314–317 dual-fuel combustion, diesel with syngas experimental configuration, 304–306 results and discussion, 317–324 methane, spark timing study experimental configuration, 302, 303 results and discussion, 306–311 natural gas (HCNG) experimental configuration, 303–304 results and discussion, 311–314 spark ignition (SI) fuels, 290–296 early use in engines, 290–291 hydrogen impacts on methane and gasoline combustion, 291, 294–295 Iron carbonyls and laminar burning velocity, 62–63 premixed flame propagation in high-pressure media, 53, 54 syngas cleanup requirements, 19 Iron catalysts, Fischer-Tropsch process, 215

390 Iron-chromium alloy, catalyst substrates, 235 Iron oxide, 210 Isotropy, turbulence, 131–132

J Jet penetration, 197

K Karlovitz number, 137 Kernels, turbulent combustion, 148, 149–150, 156, 157 Kinetic energy, turbulent combustion, 132–133 Kinetics blowout modeling, 262–268 chemical reaction mechanisms; See Reaction mechanisms and kinetics combustion characteristics studies, 100 flashback, vortex breakdown-driven flame propagation in swirling flow core, 273 hydrogen-assisted internal combustion, 291 laminar flame properties complex mixtures, 74 diluent effects, 88, 89–90 flame thickness, 77 pressure and, 81, 84–85 oxy-combustion, 206–207 Knockout, 20 Knudsen diffusion coeffcients, 345 Kolmogorov scale, 133–134, 137 Koppers-Totzek medium-BTU gas, 179 Koppers-Totzek oxygen process, 174–175

L Laboratory burners, turbulent combustion studies; See Turbulent combustion properties, premixed syngas Laminar boundary layers, flashback in, 270–271 Laminar burning rates; See Burning rates/speed/ velocity Laminar counterflow premixed flame simulation, 124 Laminar flame properties, 71–96; See also Combustion characteristics combustion in premixed systems, 72–92, 126, 202 adiabatic flame temperature, 72–74 diluent effects, 107 flame extinction, 92–95 flame structure and flame thickness, 74–80 flashback, 269 gasoline, hydrogen, and methane, 294 hydrogen-assisted internal combustion, 291

Index nonpremixed systems, 95–96 operability issues; See specific issues (blowout, flashback, autoignition) premixed flame propagation, 80–92 diluents, 88–90 flame stretch, 90–92 H2/CO ratio, 81, 82 preheat temperature, 86–88 pressure, 81–86 turbulent combustion; See also Turbulent combustion properties, premixed syngas high-pressure and temperature conditions, 158 wrinkled laminar flame model, 142 Laminar flames, VOC emissions, 182–183 Laminar flame speed; See also Flame speeds and flame propagation, 80 flashback in boundary layer, 270–271 premixed combustion conditions, 104, 105 turbulent combustion, 140, 142, 157 Laminar mass burning rates, 54 Lanthanum strontium-doped lanthanum manganate (LSM), 333, 348, 350, 355, 356 strontium-doped lanthanum vanadate (LSV), 347 Lanthanum strontium cobalt ferrite (LSCF), 348 Laser-induced fluorescence (LIF), 173–174, 175 Laser sheet tomography, 135–136 Lead, 185 Lean direct injection (LDI), diffusion flame combustors, 201 Least squares analysis, explosion characteristics of H2-containing systems, 42 Length scales, turbulent combustion, 131–134, 136 Lewis number, 155, 163 catalytic combustion, 226, 227, 231 and flame instability, 139, 140, 142, 156 flame wrinkles, 141, 161 in high-pressure media, 52 and flame temperature, 72, 91–92 and NOx emissions, 201 premixed flame propagation in high-pressure media, 52 strain rate and, 92 thermal management, 230, 231, 233 turbulent combustion, 154 Lift-off, flame, 160 Light-off temperatures catalytic combustion hydrogen/air mixtures, 244–245 hydrogen/carbon monoxide mixtures, 252–256 palladium catalysts and, 234

Index Lignite, 5 Limit regimes, explosion characteristics, 31, 32, 35, 36–37, 38, 39 Liquid fuel production circulating fluid bed systems, 13 moving bed gasifiers, 10 Lister-Petter engine, 302, 303 Livengood and Wu integral, 298–299 Local consumption speed, 147, 153 Local displacement speed, 147, 150, 151–152, 155 Local flame curvature, 141, 144, 162 Local rate minimum, 59–60 Lockhopper feed system, 8 Low-pressure systems, explosion characteristics, 31 Low-swirl burner, 146, 148, 150–151, 153, 154 Low-swirl injector, 155–156, 159, 161 Low-temperature systems; See also Highpressure/low-temperature ignition fluid bed gasifiers, 6 ignition studies, 48–49 Lower flammability limit, carbon monoxide production, 181 Lower heating value (LHV) chemical looping systems, 209 fuel composition, 194–195 gasoline, hydrogen, and methane, 294 variability of syngas, 194 Lowest achievable emissions rate (LAER), 223 LSCF (lanthanum strontium cobalt ferrite), 348 LSM (strontium-doped lanthanum manganate), 333, 348, 350, 355, 356 LSM/YSZ cathode, 350 LSV (strontium-doped lanthanum vanadate), 347 Lubricating oil decomposition carbon monoxide production, 180 VOC emissions, 185 Lurgi low-BTU gas, 179 Lurgi oxygen process, 174–175

M Mach Hebra nozzle burners, 80 Markstein length, 91, 92, 139, 157 Markstein numbers, 144 Mass balance, flame speed calculations, 152 Mass burning rate flame propagation in high-pressure media, 56, 58 pressure and, 82–83, 86 preheat temperature effects, 88 Mass flow diffusion flame combustors, 201 turbulent combustion, 152 Mass fractions catalytic combustion, hydrogen/air mixtures, 242

391 combustion in counterflow diffusion flames, 117 combustion in premixed systems, diluent effects, 108, 109, 110, 111, 112 Maximum brake torque (MBT) spark timing, 311, 312 Maximum flame speeds, 105, 126 Maximum flame temperature counterflow diffusion flames, diluent effects, 119, 120 factors affecting, 126 premixed systems diluent effects, 107 H2/CO equivalence ratios and, 106 strained counterflow premixed flames, 125 stretch effects, 90, 91 MCFC (molten carbonate fuel cells), 19, 210 Membrane-electrode assemblies (MEA) modeling, 355, 357, 358–359 SOFCs, 332–337 anode, 335–336 cathode, 333–334 electrolyte membrane, 334–335 planar stacks, 350 Membranes warm gas cleanup, 24 YSZ, 348 Mercury cold gas cleanup, 20, 21, 22 pollutant production and control, 185–186 Metal carbonyls and ignition process, 284 and laminar burning velocity, 62–63 premixed flame propagation in high-pressure media, 53, 54 syngas cleanup requirements, 19 Metal oxides catalytic combustion carbon monoxide/air mixtures, 247 catalysts, 234 chemical looping systems, 208 production of, 186 Metallic reactors, catalytic combustion, 223–224 Metals catalytic combustion catalysts, 234 cold gas cleanup, 20, 22 fuel cleanup, 197 and ignition process, 284 premixed flame propagation in high-pressure media, 54 SOFC effects, 331, 346, 347 Methane biogas composition, 205 combustion characteristics, 99; See also Combustion characteristics flame temperature comparisons, 196

392 laminar flame speed, 154 natural gas composition, 195 pollutant production and control, 188 carbon monoxide emissions, 182 carbon monoxide production, 181 VOC emissions, 183–184 pure, adiabatic flame temperature, 72, 73 SOFCs, 343 charge transfer pathways, 342 modeling, 356, 358–361, 362 spark ignition (SI) engines carbon monoxide production, 181 and gasoline combustion, 291, 294–295 spark timing study, 302, 303, 306–311 variability of syngas, 194 Methane diluents, 107, 109, 111 Methane/hydrogen/carbon monoxide, autoignition, 281 Methanol, 23, 215 Methanol fuel cell, 182 Methyl radical, 184 MIEC (mixed ionic and electric conductivity), 343, 348 Mixed coal/biomass gasification, feedstock preparation, 8 Mixed ionic and electric conductivity (MIEC), 343, 348 Mixing carbon monoxide production, 181, 182 combustion characteristics studies, 104, 125–126 emissions controls, 200 gasification technology, 2 ignition delay in perfectly stirred reactors, 101 induction chemistry sensitivity to, 62 VOC destruction, 185 Mixture-specific heat, diluent effects, 88 Moisture and ash content (maf) coal, 4 Moisture content feedstock properties, 4, 5 and process efficiency, 4 Moisture free (mf) coal, 4 Molar heating value, 73 Molecular collisions; See Collision parameters/ phenomena Molecular transport, 154, 157 Molten carbonate fuel cells (MCFCs), 19, 210 Moving bed gasifiers, 9–10, 15, 16 Mueller mechanism, 282, 283

N Natural frequency of combustor, 275 Natural gas gasification technology, 3 hydrogen-assisted internal combustion

Index combustion studies, 303–304, 311–314 lean limits, 291 reaction mechanisms, 294–295 physical and chemical properties, 294 variability of composition, 195 Natural gas power cycle applications, oxycombustion, 206 Navier-Stokes equations, 131 Nernst equation, 211 Nickel arsenide, 346 Nickel-based anodes, SOFCs charge transfer pathways, 342, 343 syngas cogeneration, 354 syngas impurity effects, 346 Nickel carbonyls acceleration of oxidation in shock tubes, 54 syngas cleanup requirements, 19 Nickel catalyst, SOFCs, 335, 336 charge transfer pathways, 342 impurity effects, 347 Nickel-free anodes, SOFCs, 343 Nickel oxide formation, SOFCs, 349 Nickel/YSZ anodes, 336, 340, 349, 357, 369 impurity effects, 346–347 modeling, 354, 355, 356 planar stacks, 350 surface mechanisms, 343–344 Nitrogen (N2) feedstock properties, 5 natural gas composition, 195 oxy-combustion, 208 pollutant production and control, 188 variability of syngas, 194 Nitrogen, molecular, 197 Nitrogen compounds cold gas cleanup, 21 fuel cleanup, 197 gasification technology, 3 Nitrogen diluents, 203, 204 diffusion flame combustors, 201 flame propagation, 88 premixed system combustion characteristics, 107–112, 113, 114 Nitrogen dioxide, 171 Nitrogen gas, NOx species conversion to, 179 Nitrogen injection, emissions controls, 199 Nitrogen oxide(s) (NOx) combustor design and development, 194 and explosion characteristics, 35–37, 51 and ignition process, 284 reaction mechanisms and kinetics R8: NO + HO2 = NO2 + OH, 36, 354 R9: NO2 + H = NO + OH, 36 Nitrogen oxide (NOx) emissions/products adiabatic flame temperature and, 72 catalytic combustion, 223, 226, 229

Index diffusion flame combustors, 202 exhaust aftertreatment, 204 fuel composition and, 196 hydrogen effects on, 160 internal combustion engines dual-fuel compression ignition (CI) combustion, 301, 320, 323 emissions controls, 199 hydrogen-assisted SI methane combustion, 307–308, 309, 311 oxy-combustion, 208 pollutant formation and control, 171–179 combustion conditions and, 173–175 control technologies, 175–179 formation mechanisms, 171–172 premixed flame propagation in high-pressure media, combustion dilution and, 56 premixed syngas combustion, 202–203 purification of syngas, 18 Nitrogen system, explosion characteristics, 39–40 Nitrous oxide, 171 NNH radical, 172 Noble metal catalysts, 234, 235 Non-Arrhenius behavior, explosion characteristics of, 41 Nonlinearity hydrogen/air mixtures, 280 NOx effects on explosion characteristics, 36 premixed flame propagation in high-pressure media, 52 Nonuniformity, shock tube, 45, 47–48 Nozzle flashback, 270 iron deposits, 55 Nozzle burners, 80, 90

O Ohmic overpotential, 338–339 Oil decomposition products, 180, 181, 182 One-dimensional double infinite planar turbulent flame, 143–145 Operability issues, 261–284 autoignition, 278–284 blowout, 262–268 combustion characteristics; See Combustion characteristics combustion instability, 273–278 flashback, 268–273 in boundary layer, 270–271 combustion pulsation-induced flashback, 270 turbulent flame propagation in core flow, 268–270

393 vortex breakdown-driven flame propagation in swirling flow core, 271–273 Operating conditions, 104 catalytic combustion, minimum pressure drop, 225–226 catalytic partial oxidation (CPO), 228–229 Fischer-Tropsch process, 215 fuel cells, 210 oxy-combustion, 206 and prediction departure from experimental results, 125–126 Operational performance; See Performance OPPDIF, 90, 95 Optical measurements, blowout, 265 Oscillations and combustion instability, 203, 273–278 kinetically driven, in catalytic combustion, 255 Outflow, turbulent combustion, 152 Outlet temperature, comparison of gasifier types and approaches, 16 Overpotentials, SOFC electrochemistry, 337–342 Oxidation catalytic partial oxidation (CPO), 227–230, 232–233 flame structure, 75 gasification technology, 2, 4, 7 NOx effects on explosion characteristics, 36 SOFCs, 343; See also Solid oxide fuel cells Oxidizer and adiabatic flame temperature, 72, 73 catalytic combustion, 224 Oxy-combustion efficiency comparisons, 214 utilization of syngas, 206–208 Oxygen chemical looping systems, 208 comparison of gasifier types and approaches, 15–16 diluent options, 203 explosion characteristics, 41 feedstock analysis, 4 feedstock properties, 5 fractional conversion, 233 gasification technology, 2, 3 minimizing in exhaust stream, 206 premixed flame propagation in high-pressure media, 57, 59 pure, and adiabatic flame temperature, 72 SOFC anode protection, 349 SOFC electrochemistry, 330, 332, 333, 334–337 Oxygen-blown gasifiers, diffusion flame combustors, 201

394 Oxygen radicals nonpremixed flame properties, 95 premixed flame propagation in high-pressure media, 57, 59 R16: O + H2 = H + OH, 57 spatial variations of major and minor species for different equivalence ratios, 105, 106

P PAFC (phosphoric acid fuel cell), 210 Palladium catalysts, 234, 235 Paraffins, 182–183, 215 Parameterized reaction, explosion characteristics, 41–42 Partial gas quench, thermal management, 17 Particle filters, 20, 21, 185, 186 Particle size, feedstock feedstock preparation, 7, 8 moving bed gasification, 10 Particulates, 182 cold gas cleanup, 20, 21 dual-fuel compression ignition (CI) combustion, 301, 302 and ignition, 284 internal combustion engines, 320 pollutant formation and control, 186–187 syngas cleanup requirements, 19 warm gas cleanup, 24 Pattern factors, turbine inlet temperature profiles, 200 Peak flame temperature combustor design and development, 194 fuel composition and, 196, 197 nonpremixed flame properties, 96 Peak fluxes, premixed flame propagation in highpressure media, 61 PEM (polymer electrolyte membrane fuel cell), 210 Perfectly stirred reactors (PSR) blowout modeling, 264–265 ignition and extinction in, 101–104 Perfect mixing, ignition delay, 102 Performance diesel duel-fuel engines, 300–302 feedstock description, 4 NOx in exhaust gas recirculation stream and, 36 premixed syngas combustion, 203 reciprocating internal combustion engines, 205 SOFCs, modeling, 356, 358–367 spark ignition (SI) engines, 290–291 Perovskites, 234, 347, 348 Petcoke, 5, 6, 7

Index Petroleum, 3 Phase relationships, combustion instability, 273, 274, 278 Phase spaces, turbulent combustion, 136–139 Phosphoric acid fuel cell (PAFC), 210 Phosphorus, 346 Physical solvents, acid gas scrubbing, 21–22 Physicochemical processes, catalytic combustion, 223–226 Pipelines, 206 Piston engines, carbon monoxide production, 180 Planar laminar flame, interaction with vortex, 134, 135 Planar laser-induced fluorescence, 139, 140, 159, 161, 162, 163 Planar SOFC systems, 350–351, 352–353 Platinum catalysts, catalytic combustion, 234, 235 carbon monoxide/air mixtures, 247 hydrogen/air mixtures, 241, 244 syngas, 237 PLIF; See Planar laser-induced fluorescence Plug-flow reactor, reburning effects, 178 Pollutant formation and control, 169–188 carbon dioxide, 187–188 carbon monoxide, 180–182 combustor design and development, 194 emissions controls chemical looping system and, 210 diffusion flame combustors, 201, 202 oxy-combustion, 208 premixed syngas combustion, 202–203 gas turbines emissions requirements, 199–200 exhaust aftertreatment, 204 nitrogen oxides (NOx), 171–179 combustion conditions and, 173–175 control technologies, 175–179 formation mechanisms, 171–172 particulates, 186–187 sulfur species, 179–180 trace elements, 185–186 volatile organic compounds (VOCs), 182–185 Polyaromatic hydrocarbons (PAHs) pollutant production and control, 182, 186 SOFCs, 343, 344 Polymer electrolyte membrane (PEM) fuel cell, 210 Porous media transport, SOFCs, 345 Postcombustion control NOx, 175–176 sulfur pollutants, 180 Potassium, 19 Potential energy surface, R13 kinetic models, 40 Power generation

Index catalytic combustion, 226–229 IGCC; See Integrated gasification combinedcycle systems oxy-combustion, 206 pollutant production and control, 186 reciprocating internal combustion engines, 204–205 SOFCs, 330 water/steam dilution issues, 204 Prakash model, ignition delay, 299–300 Preburners, catalytic combustion, 234 Preheat temperature and adiabatic flame temperature, 73, 74 diluent effects, 89 extinction strain rates, 93 flame propagation, 80 and flame thickness, 78–80 and laminar flame properties, 86–88 and premixed flame properties, 71 stretch effects, 91 Preheat zone thickness and flame thickness, 77 hydrogen levels and, 76 turbulent combustion, 138–139 Preheated pressurized turbulent flow, 158–159 Preheating and ignition delay, pressure effects, 280 SOFCs, with recovered exhaust, 352–353 Preignition, 37 ignition delay times of syngas–air combustion, 44 modeling, 49, 50 shock tube pressure histories accuracy of measurements, 51 heat release, 45 pressure-time history record, 47, 48 PREMIX, 74, 104 Premixed combustors, 198 blowout modeling, 265, 266–267 flashback, 268–269 operability issues autoignition, 262 flashback, 262 Premixed swirling combustor, 265, 266–267 Premixed systems catalytic combustion, emissions, 230 combustion characteristics, 101–114 CO/H2 mixtures, 104–106, 107 diluents and, 107–112, 113, 114 ignition and extinction in CHRs and PSRs, 101–104 pressure and temperature effects, 112–114, 115–116 strained counterflow diffusion flames, 124, 125 turbulence; See Turbulent combustion properties, premixed syngas

395 combustion instability, 273–278 flame propagation in high-pressure media, 52–61 gas turbines, 194, 202–203 internal combustion engines, 319, 321 laminar flame properties, 72–92 adiabatic flame temperature, 72–74 combustion characteristics studies, 100 flame propagation, 80–92 flame structure, 74–77 flame thickness, 77–80 pollutant production; See Pollutant formation and control Premixer autoignition, 278 flashback and flame anchoring in, 198 and ignition delay times, 283 Pressure and adiabatic flame temperature, 73–74 catalytic combustion hydrogen/air mixtures, 242–244 minimum pressure drop, 225–226 NOx formation, 230 chemical looping systems, 210 combustion characteristics counterflow diffusion flames, 120–121, 123, 124 early experiments, 99–100 ignition delay in PSR and HR, 104 premixed systems, 112–114, 115–116 explosion characteristics of H2-containing systems, 31, 35 extinction strain rates, 94 feedstock transport systems, 8 flame propagation, 80 and flame thickness, 77–78, 79 gasifier types; See specific gasifiers gas turbines combustor pressure drop, 194, 198 fuel pressure drop and mixing, 194, 197–198 and ignition delay, 104, 283 and laminar flame properties, 81–86, 87 methanol production, 215 and operability issues autoignition, 279–281, 282 combustion instability, 262, 273–278 palladium oxide decomposition, 234 shock tube modeling, 49 shock tube pressure histories, 46–49 stretch effects, 91 turbulent combustion properties, premixed syngas, 157–161 Pressure-driven convective fluid flow, 345 Pressure-heat release phase, combustion instability, 278

396 Pressure-temperature conditions; See also Highpressure/low-temperature ignition Pressure variations ignition studies, 45 and premixed flame properties; See Laminar flame properties premixed syngas combustion, 203 shock tube modeling, 49–51 shock tube pressure histories, 46–49 Pressure waves, 203 Pressurized gasifiers, 13, 15 Probability density functions, turbulent combustion, 131, 141–142 Process efficiency/energy inputs adiabatic flame temperature, 73 comparison of gasifier types and approaches, 14 cooling and, 19 feedstock properties and, 4 steam generation, 204 syngas conversion to other fuels, 216 water addition, coal slurry creation, 7 Process heat, circulating fluid bed systems, 12–13 Producer gas, 2 Production technology; See Gasification technology Prompt NO, 172, 229 Propane, 188, 195 PSR; See Perfectly stirred reactors Pulsation, combustion pulsation-induced flashback, 270 Pulverized coal (PC)-fired boilers, 169, 170, 180 Purge gas, 8 Purification of syngas; See Cleanup/purification Purity of fuel combustor design and development, 194 gas turbines, 197

Q Quartz diffuser cone, 152–153 Quench cold gas cleanup, 20, 21 combustion in counterflow diffusion flames, 117 flashback in boundary layer, 270–271 hydrogen-assisted internal combustion, 294 NO2-to-NO reaction, 316, 317 strained counterflow premixed flames, 124 thermal management, 16, 17, 18 turbulent combustion, 138

R Radial outflow, turbulent combustion, 152 Radiant syngas cooler, 17, 18 Radiative heat transfer

Index diluent effects, 88, 90 oxy-combustion, 206 Radical coupling, catalytic combustion of syngas, 236 Radical recombination catalytic combustion, 225 premixed flame propagation in high-pressure media, 61 Radicals branching; See Branching reactions combustion characteristics studies, 104 explosion characteristics of H2-containing systems, 31–32 nonpremixed flame properties, 95, 96 Radical termination reaction flame propagation, pressure and, 85 heat release mechanisms, 75 Rankine cycle, 213 Rapid compression machine ignition, 40, 41, 43 ignition delay in PSR and HR, 102 modeling approaches, 50 theory-experiment disparities, 62 Rate parameters combustion characteristics studies, 100 ignition studies, 284 premixed flame propagation in high-pressure media, 58 Rayleigh scattering, 138 Rayleigh scattering images, 138–139, 163 Reaction mechanisms and kinetics, 29–63, 279 combustion in premixed systems, 100–101 diluent effects, 107 pressure and temperature effects, 113–114 DMCF code, 114 explosion characteristics, 30–37 NOx impurities and, 35–37 flame structure, 74–75 flame thickness, 77 H2/CO kinetic model, recent and proposed updates, 37–43 high-pressure/low-temperature syngas ignition and kinetic implications, 43–52 modeling approaches, 49–51 other systems, 5–52 shock tube pressure histories, 46–49 laminar flame propagation, 74–76 predictive ability of models, real system perturbation sources, 62–63 premixed flame propagation in high-pressure media, 52–61 premixed syngas combustion, 202 reburning effects, 178 SOFCs, thermal and heterogeneous catalytic chemistry, 343–344 Reaction order, flame propagation, 84, 85 Reaction rates

397

Index diluent effects, 107 explosion characteristics, 32 pressure and temperature effects combustion in premixed systems, 113–114 preheat temperature and, 79–80, 86 pressure and flame propagation, 84–85 pressure and flame speed, 81 Reaction temperature, gasification technology, 2 Reaction zone region flame propagation in high-pressure media, 61 pressure effect modeling, 83–84 strain and, 92 turbulent combustion, 136, 137–138, 157 Reactor geometry, catalytic combustion, 241, 242 Reburning (staged fuel injection), 177–179 Reciprocating engines carbon monoxide production, 180 exhaust gas recirculation in, 35 utilization of syngas, 204–206 Recirculation; See also Heat recovery exhaust gas (EGR) combustion in reciprocating engines, 35 dual-fuel compression ignition (CI) combustion, 320 flue gas (FGR) NO control, 176, 177 waste heat recovery, 157 SOFC anode exhaust, 349 steam reforming; See Steam reforming turbulent combustion, 154 Recirculation zone as ignition source, 104 vortex breakdown-driven flame propagation in swirling flow core, 272 Recombination of radicals, catalytic combustion, 225 Recuperative gas turbine cycle, 213, 214 Recuperators preheating reactants with, 157 SOFCs, 352 Reducing conditions ash temperatures, 7 feedstock properties, 5 Reflected shock wave, 45, 47, 50 Reforming catalytic, 232 chemical looping systems, 208–210 fuel cells, 212, 214 SOFCs, 330, 331, 342 electrochemistry, 336 modeling, 354, 355, 356 on-anode, 349 performance with syngas from, 361–365 upstream, 343 SOFC syngas cogeneration, 354

Regime diagrams, turbulent combustion, 136–139 Regulatory standards carbon monoxide emissions, 181 cleanup requirements, 19 Residence time blowout modeling, 265–266 comparison of gasifier types and approaches, 16 diffusion flame combustors, 202 and extinction, 104 ignition delay in perfectly stirred reactors, 101 nonpremixed flame properties, 96 oxy-combustion, 207 strain rate and, 91–92 Resistive loss, fuel cells, 212 Reynolds-averaged Navier-Stokes equations (RANS), 131 Reynolds number, 246 blowout, 266 catalytic combustion, 244, 246 swirl flow dynamics, 284 turbulent, 137 Reynolds stresses, 131, 132 Rhodium catalysts, 234 Ricardo Hydra single-cylinder engine, 303 Rich-quick-lean (RQL) combustion, 229 Root mean square (rms) pressure fluctuation, 276 Root mean square (rms) velocity component fluctuations, 131 Rotations, R13 kinetic models, 40

S Samaria-doped ceria, 348 Sawdust, 8 Scalar fields, turbulent combustion, 153 Scaling laws, turbulent combustion, 132, 143–144 Scatter, turbulent combustion, 145 SCR; See Selective catalytic reduction Scrubbing acid gas, 22–23 cold gas cleanup, 20 particulate removal, 186 purification of syngas, 18 tars, 186 Second explosion limit explosion characteristics of, 35, 38 ignition delay, 283 Second limit, extended; See Extended second limit Selective catalytic reduction (SCR), 176, 204 emissions controls, 199 exhaust aftertreatment, 204 fuel cleanup, 197 purification of syngas, 18 Selective noncatalytic reduction (SNCR), 176

398 Selenium, 185 SENKIN (CHEMKIN II), 49, 50 Separation autoignition, 278 flashback, 270 Shadowgraph, ignition studies, 46 Shear layers blowout modeling, 262, 263 turbulent combustion, 161 Shear stresses blowout, 262 turbulent combustion, 131 Shock tubes combustion characteristics, 99 explosion characteristics of H2-containing systems, 35, 41 ignition delay, 45 ignition delay studies, 102 ignition studies, 43, 44, 45 metal carbonyl acceleration of reactions in, 54 modeling approaches, 49–51 pressure histories, 46–49 theory–experiment disparities, 62 S-H sigma bond, VOC emissions, 183–184 SI engines; See Spark ignition engines Single-cylinder engine research, hydrogenassisted combustion, 291 Slag/ash, 3 cold gas cleanup, 20, 21 feedstock properties, 4, 5 gasification technology, 3 gasifier types and; See specific gasifiers moving bed gasifiers, 10 particulates, 186 thermal management, 18 Slagging gasifiers ash properties, 6–7 entrained flow, 13–14, 16 Smoke, 301 Sodium, 19 SOFC; See Solid oxide fuel cells Solid fuel feedstocks, 1, 2, 3–4 Solid oxide fuel cells (SOFCs), 210–212, 214, 216–217, 330–369 cleanup requirements, 19 electrochemistry, 337–343 activation overpotentials, 339–341 cell potential and overpotentials, 337–338 charge transfer pathways, 342–343 concentration overpotentials, 341–342 ohmic overpotential, 338–339 materials, 347–349 anode, 349 cathode, 348–349 electrolyte, 347–348 membrane-electrode assemblies, 332–337 anode, 335–336

Index cathode, 333–334 electrolyte membrane, 334–335 modeling, 354–367, 368 approaches, 355–356, 357 performance of syngas from CPOx of hydrocarbons, 356, 358–361 performance of syngas from steam reforming of hydrocarbons, 361–365 performance with syngas from coal and biomass gasification, 365–367, 368 porous media transport, 345 stacks and systems, 350–354 cogeneration of syngas, 353–354 integration of systems, 352–353 planar, 350–351 tubular, 351–352 syngas cleanup requirements, 19 syngas impurity effects, 346–347 thermal and catalytic chemistry, 343–345 Solid-state reactions, SOFCs, 342, 348 Solvents acid gas scrubbing, 21–22 warm gas cleanup, 24 Soot, 324 Soot temperature, 302 Sorbents, mercury, 186 Sour shift, 20 Spark energy, and flame stability, 150 Spark ignition (SI) engines carbon monoxide production, 181, 182 VOC emissions, 185 Spark ignition (SI) fuels, 290–296 early use in engines, 290–291 hydrogen impacts on methane and gasoline combustion, 291, 294–295 methane, spark timing study experimental configuration, 303 results and discussion, 306–311 natural gas (HCNG) experimental configuration, 303–304 results and discussion, 311–314 Spatial distributions/variations of major and minor species combustion in counterflow diffusion flames, 121, 122 for different equivalence ratios, 105, 106 Species profiles, pressure and temperature effects, 115–116 Specific heat and adiabatic flame temperature, 73 fuel dilution and, 88, 90, 107, 204 hydrogen effects on gasoline and methane combustion, 291 modeling ignition, 50 oxy-combustion, 206 Speed, flame; See Flame speeds Spherical bomb, 99–100

Index Spherically expanding flames, 80 Spinning tubes, flame acceleration in, 272 Stability, flame catalytic partial oxidation (CPO), 228 combustion characteristics studies, 100 Lewis number; See Lewis number premixed syngas combustion, 202 turbulent combustion, stabilizing mechanisms, 140 Stability maps, 275, 276 Stack emissions, purification of syngas, 18, 19 Stacks and systems, SOFCs, 350–354 cogeneration of syngas, 353–354 integration of systems, 352–353 planar, 350–351 tubular, 351–352 Staged fuel injection (reburning), 177–179 Stagnation flame extinction, 92 flow strain and, 90 nonpremixed flame properties, 95 Stagnation and low swirl barrier studies, 150–151, 152 Stagnation burner, 146 Stagnation flow burner, 141, 148, 150–151, 154 Standard potential, fuel cells, 211 Statistical measurements, turbulent combustion, 153 Steam chemical looping systems, three-reactor, 210 combustion characteristics; See Combustion characteristics combustor design and development, 194 diffusion flame combustors, 201 fuel cell systems, 212 gasification technology, 2 incoming, SOFCs, 331 oxy-combustion, 206 syngas cooling and, 19 syngas reheating after purification, 195 Steam cycles, hybrid fuel cells, 213 Steam dilution, oxy-fuel, 206, 207 Steam injection emissions controls, 199 moving bed gasifiers, 10 Steam reforming catalytic, 232 SOFCs, 330, 331, 343 electrochemistry, 336 modeling, 354, 356 performance with syngas from, 361–365 SOFC syngas cogeneration, 354 Steam requirements feedstock analysis, 4 gasifier types and; See specific gasifiers Stirred reactors, ignition and extinction in PSRs, 101–104

399 Strained counterflow premixed flames, 124, 125 Strain resistant fuels, CIVB-induced flashback Strain/strain rates; See also Stretch, flame and blowout, 262, 267 combustion in counterflow diffusion flames, 117, 118 diluent effects, 117, 119–120, 121, 122 pressure effects, 120–121, 123, 124 combustion instability, 277 extinction, 92–95 and flame propagation, 71 flame stretch, 90–92 nonpremixed flame properties, 95–96 strained counterflow premixed flames, 124, 125 turbulent combustion, 139, 141–142 Stretch, flame; See also Strain/strain rates laminar flame properties, 80, 90–92, 142 turbulent combustion, 139, 154 reaction to curvature and strain, 139–142 stretch factor measurement, flame surface density and, 144–145 Strontium barium strontium cobalt iron oxide (BSCF), 348–349 lanthanum strontium cobalt ferrite (LSCF), 348 Strontium-doped ceria, 342 Strontium-doped lanthanum manganate (LSM), 333, 348, 350, 355, 356 Strontium-doped lanthanum vanadate (LSV), 347 Strontium titanate, yttrium-doped, 347 Strouhal numbers, 274 Structure; See Flame structure Sub-bituminous coal, 5 Substitute natural gas (SNG) production, 215, 216 Substrate, catalytic combustion catalysts, 235 Sulfur impurities and catalyst performance, 235 feedstock properties, 5 fuel cleanup, 197 pollutant formation and control, 179–180 and pollutant production, 175 purification of syngas, 18 SOFCs, 331, 347 syngas cleanup requirements, 19 Sulfur products exhaust aftertreatment, 204 gasification technology, 3 Superadiabaticity, 162, 230–232 Surface area, flame flame speed calculations, 152 stretch definition, 90 Surface area, flame front, 142, 143 Surface chemistry catalytic combustion, 231 catalytic combustion of syngas, 236, 237 and ignition, 284 Surface density, flame, 144, 145, 147

400 Surge, compressor, 270 Sweet shift, 20 Swelling coals, moving bed gasifiers, 10 Swirl autoignition, 278 intake-induced, 311–312, 313, 314 operability issues, 261 combustion instability, 273, 277–278 flashback, 262, 270 vortex breakdown-driven flame propagation in, 271–273 Swirl burner, stability map, 275 Swirl-stabilized flames, premixed syngas combustion, 202 Swirled vanes, flashback, 270 Swirling combustor blowout modeling, 265, 266–267 combustion instability, 275, 277 flashback, 268–269 Synergy, coal-biomass, 8

T Tail-gas combustor, SOFCs, 352 Tangential strain, 90 Tars/tar formation, 182, 186 cold gas cleanup, 23 comparison of gasifier types and approaches, 15 entrained flow gasifiers, 14 fuel cells, 212 transport reactors, 13 Taylor’s hypothesis, 134 Temperature; See also High-pressure/lowtemperature ignition; Thermal management and autoignition, 279–280 and catalyst stability, 234 combustion in counterflow diffusion flames, 117, 118 diluent effects, 119 pressure effects, 123 combustion in premixed systems, 112–114, 115–116 diluent effects, 107 H2/CO composition and, 105–106, 107 ignition delay in PSR and HR, 102, 103 spatial variations of major and minor species for different equivalence ratios, 105, 106 strained counterflow premixed flames, 124, 125 diffusion flame combustors, 200, 201 energy inputs; See Process efficiency/energy inputs explosion characteristics of H2-containing systems, 31, 35

Index Fischer-Tropsch process, 215 fuel cell operating conditions, 210 gasification technology, 2 ash composition, 6–7 ash fusion, 4 ash properties, 7 cold gas cleanup, 21 comparison of gasifier types and approaches, 15, 16 process efficiency, 4 ignition delay times, 283 laminar flame properties, 72–74 diluent effects, 88 extinction strain rates, 94–95 flame extinction, 92 flame propagation, pressure and, 81–82 flame structure, 74, 75, 76 premixed flame propagation in highpressure media, 61 strain and, 92 stretch effects, 90, 91 palladium oxide decomposition, 234 pollutant production and control NO production, 176 trace element fates, 186 VOC destruction, 185 reaction mechanisms shock tube, pressure gradients behind reflected shock and, 47 transition into explosion regime, 38 SOFCs, endothermic reforming and, 336 thermal/diffusive effect; See Lewis number turbulent combustion properties, premixed syngas, 157–161 Temperature profiles, conventional hydrocarbon fuels, 106 Temperature window sensitivities, premixed flame propagation in high-pressure media, 56, 59, 60, 61 Termination catalytic combustion of syngas, 236 diluent effects, 90 explosion characteristics, 31–32, 33–34 flame structure, 76 pressure and, 86 three-body radical termination reaction, 75, 85 Ternary collisions, explosion characteristics of H2-containing systems, 32 Thermal ballast, transport reactors, 13 Thermal barriers, water/steam dilution issues, 204 Thermal chain explosive regime, 51 Thermal chemistry, SOFCs, 343–345 Thermal diffusion explosion characteristics of H2-containing systems, 33

Index flame propagation, pressure and, 81 and flame thickness, 77 turbulent combustion, 154 Thermal/diffusive effect; See Lewis number Thermal dilution effects, NOx production, 173 Thermal efficiency, spark ignition (SI) engines, 290–291 Thermal energy transfer diluent effects, 90 flame propagation, pressure and, 81 Thermal management catalytic combustion, 230–233 catalytically stabilized thermal (CST) combustion, 227 hydrogen/air mixtures, 241–242 cleanup methods and, 19 fuel cells, 212–213, 214 gasification technology, 16–18 Thermal NO, 171–172, 175, 176, 177, 230 Thermal stability, catalysts, 234 Thermal swing regeneration, 22 Thermoacoustic stability, premixed flame propa­ gation in high-pressure media, 56 Thermodynamic state, shock tube experiments, 45, 47 Thermogravimetric analysis, 6 Thickened flamelet properties, blowout modeling, 262 Third-body concentration, explosion characteristics of H2-containing systems, 38 Three-body radical termination reaction, 75, 85 Three-phase boundary (TPB), 333, 340–341, 342, 348 Three-reactor chemical looping system, 210 Time delay, combustion instability, 273, 277 Time relationships blowout modeling, 263, 264, 265–266 explosion characteristics, 31, 32, 33, 35 shock tube modeling, 49 shock tube pressure-time histories, 47 turbulent combustion, 131–134 Time scales autoignition, 278 blowout modeling, 268 explosion characteristics, 31 turbulent combustion, 133, 137 Timing, ignition, 320 NOx in exhaust gas recirculation stream and, 36 spark ignition (SI) engines, 302, 303, 304, 306–311 Tomographic images, turbulent combustion, 135–136 Torrefaction, 8 Town gas, 2 Trace impurities

401 chlorine removal, 186 pollutant formation and control, 185–186 SOFCs, 331, 346 syngas cleanup requirements, 19 Transfer functions, flame response, 274 Transport, feedstock, 8 Transport properties complex mixtures, 74 DMCF code, 114 turbulent combustion, 154 Tubular SOFC systems integration of systems, 352–353 modeling, 355 structure, 351–352 Turbine cycle, fuel cells, 213 Turbines carbon monoxide emissions, 182 cleanup requirements, 19 combined cycle, 209 diffusion flame combustors, 200 emissions controls, 199 gas, fuel cells, 213, 214 premixed syngas combustion, 202–203 purification of syngas, 19 sulfur pollutants, 180 water/steam dilution issues, 204 Turbulence isotropy, 131–132 Turbulent combustion flashback, 268–271 laminar flame properties; See Laminar flame properties Turbulent combustion properties, premixed syngas, 129–165 classification of turbulent flames and turbulent flame interactions, 134–142 curvature and strain, stretched flame model, 139, 141–142 Lewis numbers, 139–140, 142 reaction zone thickness, 136 wrinkled flame regime, 134–139, 141 flashback, 262 general description of turbulence length and time scales, 131–134 heat release rate and flame speed, 142–154 diffuser burner, 152–153 experimental configurations of laboratory burners, 145, 146, 147, 148 flame kernel, 149–150 flame surface area ratio, 142, 144 flame surface density and stretch factor measurement, 144–145 global displacement speed, 149 idealized one-dimensional double infinite planar turbulent flame, 143–145 scaling laws, 143–144 stagnation and low swirl barrier studies, 150–151, 152

402 high temperature and high pressure conditions, 157–161 modeling considerations for syngas and hydrogen flames, 161–163 turbulent flame speed heat release rate and, 142–154 syngas and hydrogen effects on, 154–157 Turbulent flame speed, heat release rate and, 142–154 Turbulent flame speed ratio, 157 Turbulent flow fields, sulfur pollutants, 180 Two-body reaction, flame propagation, 84, 85 Two-stage combustion, NO control, 176 Two-zone structure, flame propagation, 83

U Ultimate analysis, feedstock properties, 4, 5 Updraft (countercurrent) moving bed gasifiers, 9, 10 Upflow, entrained flow gasifiers, 14 Urea, 176 Utah surrogate mechanisms, VOC emissions, 183 Utilization of syngas, 193–217 chemical looping systems, 208–210 design and development considerations, 194 fuel and chemical production from syngas, 214–216 fuel cells and, 210–214 gas turbines, 193–204 combustor configuration, 200–203 combustor pressure drop, 198 composition of fuel, 194–197 diluent options, 203–204 emissions requirements, 199–200 exhaust aftertreatment, 204 fuel pressure drop and mixing, 197–198 purity of fuel, 197 oxy-combustion, 206–208 reciprocating engines, 204–206

V Vanadium, trace elements, 185 Vanadium catalysts, 6 Vapor, particulate classification, 186 Vaporization, syngas cooling and, 19 Velocity gradient, critical, 270–271 Velocity vector, flame surface density measurement, 145, 147 v-flame, 145, 146, 147–148, 151, 152 Vinyl radical formation pathway, 183 Viscosity ash properties, 6–7 turbulence, 131, 133, 136 VM Motori/DDC Turbodiesel engine, 304

Index Volatile organic compounds (VOCs), pollutant formation and control, 182–185 Volatiles, feedstock properties, 5 Volatility, trace elements, 185 Voltage, fuel cells, 211 Volume, shock tube modeling, 49 Volume flow, diffusion flame combustors, 200, 201–202 Vortex breakdown bubble, 277 Vortex breakdown-driven flame propagation in swirling flow core, 271–273 Vortex shedding, and combustion instability, 273 Vortices/vorticity and combustion instability, 273–274 combustion pulsation-induced flashback, 270 turbulent combustion, 134 VTIM, 49, 50

W Wakes, flashback, 270 Wall deposits, premixed flame propagation in high-pressure media, 54 Wall effect, explosion characteristics of H2-containing systems, 31 Wall gradient, critical, 271 Warm gas cleanup, 23–24 Waste gas, reciprocating internal combustion engines, 205 Waste heat; See Heat recovery Water chemical reactions and kinetics and carbon monoxide oxidation, 31 explosion characteristics of H2-containing systems, 39–40 cold gas cleanup, 20, 21 combustion characteristics; See also Combustion characteristics diluent effects, 107–112, 113, 114 mass fraction in burned gases, 107, 108 combustor design and development, 194 diluent options, 204 flame structure, 76 gasification technology, 2, 3 feedstock preparation, 7 and process efficiency, 4 thermal management, 16, 17, 18 laminar flame properties flame propagation, 88 nonpremixed flame properties, 95 NO control, 177 separation from postcombustion products, 206 SOFCs, 211, 330, 331, 342 variability of syngas, 194 Water gas, 2

403

Index Water–gas shift diluent options, 203 gasification technology, 2, 3 cold gas cleanup, 20, 21 thermal management, 16, 17, 18 SOFCs, 343, 368 charge transfer pathways, 342 electrochemistry, 336 modeling, 355, 364, 366 syngas cogeneration, 354 Water knockouts, 21 Water quench cold gas cleanup, 21 purification of syngas, 20 thermal management, 17 Wave numbers, turbulent combustion, 133 Well-stirred reactor (WSR) conditions, 38 blowout modeling, 262, 264 NOx formation, 172 Wellman-Galusha air process, 175 Wet scrubbing, 20, 21, 186 Wiebe’s functions, 319 Winkler process, 174–175

Wobbe index, 197–198 Woody biomass, 5 Wrinkle structure, flame, 134–141, 156, 159

Y Yttrium-doped strontium titanate, 347 Yttrium-doped zirconia (YSZ), 334–335 anode material; See also Nickel/YSZ anodes catalytic and thermal chemistry, 343–344 impurity effects, 346–347 cathode materials, LSM/YSZ composites, 348 electrolytes, 334–335, 347–348 modeling, 355, 356 planar stacks, 350

Z Zeldovich mechanism, NOx formation, 72, 171, 230 Zeldovich number, 61, 139 Zinc oxide/chromium oxide catalysts, 215 Zirconia; See Nickel/YSZ anodes; Yttrium-doped zirconia