No title

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

Quality of ladies’ Relationship between the knitted fabrics warm/cool feeling of fabric and the subjective evaluation of the quality 7 of ladies’ knitted fabrics Takako Inoue and Akira Nakayama School of Life Studies, Sugiyama Jogakuen University, Nagoya, Japan, and

Received 14 November 2008 Accepted 12 May 2009

Masako Niwa Nara Women’s University, Nara, Japan Abstract Purpose – The purpose of this paper is to analyze the relationship between the warm/cool feeling of the heat properties of fabrics and the subjective evaluation of the quality of ladies’ garment fabrics. Design/methodology/approach – Regression analysis is conducted using stepwise block regression applied to the expert judges’ judgment value total hand value, using six blocks of the mechanical properties and one block of the initial maximum values qmax of the heat flux of the heat properties of spring and summer tailored-type jacket fabrics, as the seven blocks of fabric properties, including the secondary term of each property. Findings – The results of the regression analysis show that the qmax values do not affect the subjective evaluation of the quality of spring and summer tailored-type jacket fabrics. The results of the regression analysis of ladies’ knitted fabric properties applied to the subjective evaluation value have confirmed that the qmax values affect the subjective evaluation of the quality of ladies’ knitted fabrics. Originality/value – This paper usefully describes the relationship between the warm/cool feeling of fabric and the subjective evaluation of the quality of ladies’ knitted fabrics. Keywords Mechanical properties of materials, Clothing, Fabric testing, Heat measurement Paper type Research paper

Introduction For the objective evaluation of fabrics that are used in diverse ways in ladies’ garments, we have been conducting analyses using basic mechanical properties, and heat properties such as heat transport properties and air permeability. The subjective evaluation of the quality of ladies’ garment fabrics involves not only basic mechanical properties, but also heat properties. In this study, we analyze the relationship between the warm/cool feeling of the heat properties of fabric and the subjective evaluation of the quality of ladies’ garment fabrics. We also clarify the effects of the warm/cool touch feeling on the subjective evaluation of the quality of ladies’ woven fabrics and knitted fabrics. Experimental Test sample collection The test samples were 134 types of woven fabrics for spring and summer tailored-type jackets which, we have used in previous research (Inoue and Niwa, 2009), and ladies’

International Journal of Clothing Science and Technology Vol. 22 No. 1, 2010 pp. 7-15 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011008767

IJCST 22,1

8

knitted fabrics newly collected from around the world (70 types for spring and summer and 65 types for autumn and winter). The samples of ladies’ knitted fabrics were collected broadly in diverse ways covering various uses. The mechanical properties of woven fabrics for spring and summer tailored-type jackets were measured in a previous research (Inoue and Niwa, 2009) (see the Appendix). Sensory tests of fabric warm/cool touch and subjective hand-evaluation of the quality of fabric Sensory tests of the fabric warm/cool touch were conducted to investigate the relationship between the measured qmax value and the fabric warm/cool touch. The tests were conducted once in the summer and once in the winter (for a total of two times). The room temperature at the time of the tests was 258C in the summer, and 208C in the autumn. Sensory tests of the subjective hand-evaluation of the quality of the fabrics were also conducted two times. The subjects representative of the consuming public were 20 females, and five males were included as reference subjects. Sensory tests of the fabric warm/cool touch were made using ranking numbers 5 (feels cool), 4 (feels slightly cool), 3 (no distinction between warm/cool touch), 2 (feels slightly warm), and 1 (feels warm). The total hand value (THV) evaluation was standardized following the standardization of the total hand, using ranking numbers 5 (excellent), 4 (good), 3 (average), 2 (below average), and 1 (poor). The initial maximum values qmax of heat flux and air resistance The initial maximum values qmax of the heat flux of the fabric heat properties were measured using a KES-F7 thermo labo, and these values were used as the values related to fabric warm/cool touch (Imai et al., 1987). The air resistance AR was measured using a KES-F8 air permeability tester. Results and discussion The test samples are 134 types of woven fabrics for spring and summer tailored-type jackets which have been used in the previous research (Inoue and Niwa, 2009) of the objective evaluation of the quality of ladies’ garment fabrics to compare ladies’ knitted fabrics. For the spring and summer ladies’ knitted fabrics, cotton and blended fibers each make up more than 30 percent of the group, while other fibers used in this group include linen, silk, and polyester. The blended fibers consist of cotton or rayon blended with polyurethane or nylon. More than 50 percent of the autumn and winter ladies’ knitted fabrics are blended fibers; other fibers used in this group include cotton and wool. Many of the blended fibers consist of wool or acrylic blended with polyurethane or nylon. Weft-knit fabrics comprised 94 percent of the spring and summer fabrics and 96 percent of the autumn and winter fabrics. Figure 1 is a distribution map of the qmax value and fabric weight. The weight per unit area of autumn and winter ladies’ knitted fabrics is higher than that of spring and summer ladies’ knitted fabrics. The distribution is wide, from heavy to light. The qmax values of spring and summer ladies’ knitted fabrics are higher than those for autumn and winter ladies’ knitted fabrics, and the distribution is slightly wider than that of autumn and winter ladies’ knitted fabrics.

Quality of ladies’ knitted fabrics

qmax (kW/m2)

2

1

9 6 5 4 3 9 10

2

3

Weight (mg/cm2) Notes: : Spring and summer knitted fabrics (n = 70); : autumn and winter knitted fabrics (n = 65)

The qmax values when a knitted fabric is arranged in two layers were measured using the same procedure. A difference in the qmax values between two- and one-layers knitted fabrics was not recognized for spring and summer ladies’ knitted fabrics. However, the distribution was narrow when the knitted fabric was in two layers. For autumn and winter ladies’ knitted fabrics, the qmax values when the knitted fabric was in two layers show a tendency to be lower than the qmax values of one-sheet fabric. From the results of the correlation of the sensory test of the warm/cool feeling in summer and autumn, the reproducibility of the warm/cool feeling of ladies’ knitted fabric is recognized collectively. The relationship between the mean value of the sensory tests with the 25 “consuming public” judges and qmax values is shown in Figure 2. A correlation with the qmax values was recognized collectively in the summer test and the autumn test. However, the correlation coefficient in the summer test is slightly lower, and the same results are seen in research by Imai et al. (1987). The correlation coefficient between the mean value of the female sensory test and the qmax values is higher than those between the mean value of the male sensory test and the qmax values, and for both the summer test and the autumn test, the correlation coefficients with the qmax values are high collectively. The relationship between the mean value of the sensory test and the qmax values was plotted by dividing the test samples into spring and summer ladies’ knitted fabrics and autumn and winter ladies’ knitted fabrics. The qmax values of the spring and summer ladies’ knitted fabrics were distributed more widely than the qmax values of the autumn and winter ladies’ knitted fabrics, and a correlation with the value of the sensory test is recognized. The correlation coefficient between the qmax values of autumn and winter ladies’ knitted fabrics and the mean value of the sensory test is low. The reason for this is thought to be the fact that the distribution of qmax values is narrow. The relationship between the thickness of the knitted fabrics and the standard deviation of the sensory test was plotted, but the effect of fabric thickness is not seen. The initial maximum values qmax are shown in Figure 3. The initial maximum values of heat flux ranked highest with the spring and summer ladies’ knitted fabrics,

Figure 1. Distribution of qmax and weight

IJCST 22,1

6.0

10

Mean values of sensory tests

5.0

4.0

3.0

2.0

1.0

Figure 2. Correlation between the value of warm/cool evaluation and qmax

0

0

1.0

2.0

3.0

qmax (kW/m2) Notes: : First sensory test; : second sensory test; deviation

: ±standard

qmax (kW/m2)

2

1 9 8 7 6 14

16

18

20

22

24

26

28

Weight (g/cm2)

Figure 3. Relation between qmax and fabric weight

Notes: The values of the mean and standard deviation of each group of fabrics are plotted in this figure; : tailored-type spring and summer fabrics (n = 134); : spring and summer knitted fabrics (n = 70); : autumn and winter knitted fabrics (n = 65); : ±standard deviation

followed by the spring and summer tailored-type jacket fabrics, and then the autumn and winter ladies’ knitted fabrics. It was clear that spring and summer ladies’ knitted fabrics provide the coolest feeling of the fabrics tested. The relationship between the weight per unit of fabric and air resistance AR is shown in Figure 4. The air resistance of the spring and summer ladies’ knitted fabrics

Quality of ladies’ knitted fabrics AR (kPa·s/m)

0.6

0.4

11

0.2

0 9 10

2

3

Figure 4. Air resistance of ladies’ knitted fabrics

Weight (g/cm2) Notes: : Spring and summer knitted fabrics (n = 70); : autumn and winter knitted fabrics (n = 65)

is low and the weight per unit of fabric is low. The air resistance and the weight per unit of fabric of autumn and winter ladies’ knitted fabrics are distributed widely, and the standard deviation is high. It is conceivable that this is related to the fact that there are many types of fabric designs, such as tailored jackets, over-alls, trousers, skirts, etc. The accuracy of the equations for the objective evaluation of the quality of ladies’ tailored-type jacket fabrics derived from six blocks of the mechanical properties was lower for spring and summer ladies’ tailored-type jacket fabrics than for autumn and winter ladies’ tailored-type jacket fabrics (Inoue and Niwa, 2009). Therefore, we used a thermo labo to measure the initial maximum values qmax of heat properties as heat transport properties. To analyze the subjective evaluation of spring and summer tailored-type jacket fabrics, a regression analysis was conducted using stepwise block regression applied to the expert judges’ judgment value THV (Inoue and Niwa, 2009), using six blocks (tensile, bending, shearing, surface, compression, and construction) of the 19 mechanical properties and one block of the initial maximum values qmax of heat properties of the fabrics, as the seven blocks of 20 fabric properties, including the secondary term of each property. For this analysis, the effect of the initial maximum values qmax was added to the six blocks of mechanical properties for spring and summer tailored-type jacket fabrics for a total of seven blocks. The regression formula is as follows: THV ¼ C 0 þ

20 X i¼1

X i 2 M i1 X 2 2 M i2 C i1 þ C i2 i si1 si2

where: C0, Ci1, Ci2 ¼ constant coefficients of the ith variable terms. Xi

¼ mechanical property of the ith variable term.

Mi1, si1

¼ the population mean and standard deviation.

Mi2, si2

¼ the square mean and standard deviation.

!

IJCST 22,1

12

The results are shown in Table I. The effects of qmax were lowest with the seventh block, and the accuracy of the regression was at the same level as the accuracy of the regression using the six blocks of the mechanical properties. The correlation between the initial maximum value qmax of heat flux and the THV was also low, at 0.017, with multiple regression analysis using the seven blocks of fabric properties with THV. We found no effects of the initial maximum values qmax of heat flux on the subjective evaluation of the quality of spring and summer tailored-type jacket fabrics. To analyze the subjective evaluation of 70 types for spring and summer ladies’ knitted fabrics and 65 types for autumn and winter ladies’ knitted fabrics, a regression analysis was conducted for the knitted fabrics of each season, using stepwise block regression applied to the subjective evaluation values of the 25 “consuming public” judges’ judgment values, and using three blocks (the initial maximum value qmax of heat flux, air resistance, and weight per unit of fabric). The results are shown in Table II. Based on the accuracy of the regression of knitted fabrics, an effect of the initial maximum values qmax of heat flux on the subjective evaluation values is recognized. Importance order Xi 1

2

3

4

5

6

Tensile LT logEM1 logEM2 RT Compression LC logWC RC Shear logG log2HG log2HG5 Surface MIU logMMD logSMD Bending logB1 LogB2 log2HB1 log2HB2 Construction logT logW

R

RMS

Importance order Xi 1

0.499 0.930

2 0.618 0.846 3 0.666 0.805 4 0.705 0.765 5 0.731 0.738

6 0.741 0.725 7

Table I. Accuracy of the regression to THV

Tensile LT logEM1 logEM2 RT Compression LC logWC RC Shear logG Log2HG Log2HG5 Surface MIU logMMD logSMD Bending logB1 logB2 log2HB1 log2HB2 Construction logT logW Heat absorption logqmax

R

RMS

0.499 0.930

0.618 0.846

0.666 0.805

0.705 0.765

0.731 0.738

0.741 0.725 0.741 0.724

Notes: Suffix 1, wrap direction; suffix 2, weft direction; Sn, number of subjects; R, accuracy of the regression; RMS, root mean square of regression error; regression based on significant judgments (Sn ¼ 6) (Inoue and Niwa, 2009) for tailored-type fabrics for spring and summer

The accuracy of the regression of the initial maximum value qmax of heat flux, air resistance and weight per unit of fabric for spring and summer ladies’ knitted fabrics was 0.655 and these three properties related more closely to the subjective evaluation values of the consumers than did the same values for autumn and winter ladies’ knitted fabrics. This is thought to be due to the fact that knitted fabrics are worn in more direct contact with the skin than spring and summer tailored-type jacket fabrics. It can be clarified that the initial maximum values qmax of heat flux affected the subjective evaluation of the quality of ladies’ knitted fabrics. The contributions of these three properties to THV were investigated for each spring and summer fabric and each autumn and winter knitted fabric The contributions of each property to THV are shown in Figure 5. For spring and summer knitted fabrics, when the initial maximum value qmax of heat flux is high, THV is high, and when the weight per unit of fabric is low, THV is recognized as being high. For autumn and winter knitted fabrics, the contribution of air resistance was recognized. The tendency of THV to be low when air resistance is high is recognized. However, the whole accuracy is based on three variables, a small number, and for the subjective evaluation of the quality of knitted fabrics, relationships with factors other than the initial maximum value qmax of heat flux and air resistance are suggested. The judges of the quality of knitted fabrics were consumers in this research, but it is also necessary to further examine the judgments of textile experts.

Quality of ladies’ knitted fabrics

13

Conclusions . The initial maximum values qmax of heat flux are recognized as having a tendency to decrease, especially when autumn and winter ladies’ knitted fabrics form two layers rather than when there is only a single layer fabric. . The reproducibility of the warm/cool feeling of ladies’ knitted fabric is recognized in the summer and autumn evaluations of the sensory test. . The spring and summer ladies’ knitted fabrics provide the coolest feeling of the fabrics tested from the initial maximum values of heat flux of the spring and summer ladies’ knitted fabrics, the spring, and summer tailored-type jacket fabrics, and the autumn and winter ladies’ knitted fabrics. . The regression analysis was conducted using stepwise block regression applied to the expert judges’ judgment value THV, using six blocks of the mechanical properties and one block of the initial maximum values qmax of the heat flux of heat properties of the fabrics, as the seven blocks of fabric properties, including S/S knitted fabrics (n ¼ 70) Importance order Xi R

RMS

1 2 3

0.489 0.449 0.437

logqmax logW logAR

0.529 0.629 0.655

A/W knitted fabrics (n ¼ 65) Importance order Xi R 1 2 3

logAR logqmax logW

0.307 0.363 0.374

RMS 0.544 0.533 0.530

Notes: Sn, Number of subjects; N, number of samples; S/S, spring and summer; A/W, autumn and winter; W, weight; AR, air resistance; R, accuracy of the regression; RMS, root mean square of regression error; regression based on general consumer judgments (Sn ¼ 25) for knitted fabrics

Table II. Accuracy of regression to THV

IJCST 22,1

Spring and summer knitted fabrics

Contribution to THV

4

14

3 2

logqmax logW

1 0 logAR

–1 –2 –4

–2

2

0

4

Xi – Mi1 s i1 Autumn and winter knitted fabrics

Contribution to THV

4 3 2 logqmax 1 0 –1 –2 –4

Figure 5. The contribution of each property to the THV of knitted fabrics

logAR

logW –2

0

2

4

Xi – Mi1 s i1 Notes: Number of samples = 75 (spring and summer), 60 (autumn and winter); number of subjects = 25

.

.

the secondary term of each property. The results of the regression analysis show that the initial maximum values qmax of the heat flux of the heat properties of the fabrics do not affect the subjective evaluation of the quality of spring and summer tailored-type jacket fabrics. The results of the regression analysis of ladies’ knitted fabric properties for 70 types of spring and summer fabrics and 65 types of autumn and winter fabrics applied to the subjective evaluation value have confirmed that the initial maximum values qmax of heat flux affect the subjective evaluation of the quality of ladies’ knitted fabrics. For the subjective evaluation of the quality of knitted fabrics, relationships with factors other than the initial maximum value qmax of heat flux and air resistance are suggested.

References Imai, J., Yoneda, M. and Niwa, M. (1987), “Sensory tests for objective evaluation of fabric warm/cool touch”, Jpn. Res. Assn Text. End-Uses, Vol. 28, pp. 414-22.

Inoue, T. and Niwa, M. (2009), “Objective evaluation of the quality of fabrics for ladies’ tailored-type jackets for spring and summer”, J. Text. Eng., Vol. 55, pp. 1-11. Matsudaira, M., Kawabata, S. and Niwa, M. (1984), “Measurements of mechanical properties of thin dress fabrics for hand evaluation”, J. Text. Mach. Soc. Japan (predecessor Journal of J. Text. Eng.), Vol. 37, pp. T49-T57.

15

Appendix

Mechanical properties

Symbols Characteristic value

Unit

Measuring conditions High sensitivity (Matsudaira et al., 1984)

Tensile

EM

%

Strip biaxial deformation

LT

Bending Surface

WT RT B 2HB MIU MMD SMD

Shearing

G 2HG

Compression

Quality of ladies’ knitted fabrics

2HG5 LC WC

RC Thickness and T weight W

Tensile strain at maximum load Linearity Tensile energy Resilience Bending rigidity Hysteresis

– 2

Upper limit tensile force (maximum load): 50 gf/cm

gf cm/cm % gf cm2/cm gf cm/cm

Pure bending Maximum curvature, K ¼ ^2.5/cm Coefficient of friction – Contactor for friction measurement: ten parallel steel piano wires with 0.5 mm dia. and 5 mm length Mean deviation of – Simulating finger skin geometry. MIU Contact force: 50 gf Geometrical mm Contactor for geometrical roughness roughness: a steel piano wire, with 0.5 mm dia. and 5 mm length. Contact force: 10 gf Shear stiffness gf/cm degree Shear deformation under constant tension of 10 gf/cm Hysteresis at gf/cm Maximum shear angle, f ¼ ^88 f ¼ 0.58 Hysteresis at f ¼ 58 gf/cm Linearity – Upper limit pressure: 10 gf/cm2 2 Compressional gf cm/cm energy Resilience % Thickness at mm Thickness at 0.5 gf/cm2 pressure 0.5 gf/cm2 Weight of specimen per unit area Weight per unit area mg/cm2

Corresponding author Takako Inoue can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

Table AI. Characteristic values of basic mechanical properties and measuring conditions for KESF measurements

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

IJCST 22,1

Computerized pattern making focus on fitting to 3D human body shapes

16

Young Sook Cho Faculty of Home Economics, Tokyo Kasei University, Tokyo, Japan

Received 6 November 2008 Accepted 9 June 2009

Keiichi Tsuchiya Graduate School of Shinshu University, Nagano-ken, Japan

Masayuki Takatera and Shigeru Inui Faculty of Textile Science and Technology, Shinshu University, Nagano-ken, Japan

Hyejun Park Department of Clothing and Textiles, College of Human Ecology, Chungham National University, Daejon, South Korea, and

Yoshio Shimizu Faculty of Textile Science and Technology, Shinshu University, Nagano-ken, Japan Abstract Purpose – This paper aims to describe the development of a method of constructing three-dimensional (3D) human body shapes that include a degree of ease for purpose of computerized pattern making. Design/methodology/approach – The body shape could be made with ease allowance to an individual’s unique body shape using sweep method and a convex method. And then generates tight skirt patterns for the reconstructed virtual body shape using a computerized pattern making system. Findings – This paper obtains individual patterns using individually reconstructed 3D body shapes by computerized pattern development. In these patterns, complex curved lines such as waist lines and dart lines are created automatically using the developed method. The method is successfully used to make variations of a tight skirt to fit different size women. The author also used the method to make other skirts of various designs. Originality/value – The method described in this paper is useful for making patterns and then garments, without the need for the garments to be later adjusted for the subject. Keywords Modelling, Computer applications, Textile technology, Human anatomy Paper type Research paper International Journal of Clothing Science and Technology Vol. 22 No. 1, 2010 pp. 16-24 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011008776

This research was supported by the Ministry of Education, Science, Sports and Culture, Grant-in-Aid for twenty-first century COE Program and Scientific Research (C), 16500124, 2005.

Introduction Traditional pattern making systems use two-dimensional perimeter element information, which is not enough to correspond to individual variations in the shape of the human body. By using three-dimensional (3D) data measured by 3D scanning devices, it is possible to express individual unique characteristic of body shape for pattern making. There are number of current research efforts in the field of 3D body modeling and computerized pattern making systems using 3D data (Cho et al., 2005; Bigliani et al., 2000; Watanabe, 1999). However, almost all of them use scanned 3D dummy shapes, not 3D human body shapes (Heisey et al., 1990a, b; Kim and Park, 2007; Miyoshi and Hirokawa, 2001). When making patterns using traditional techniques, pattern makers used dummies (Miyoshi, 2002). These dummies are shaped to include a degree of ease. Ease allowance (looseness in certain areas) in pattern making allows for body movement and unrestricted fit. Too much ease can be sloppy and unattractive. When making individually personalized clothing patterns, it is difficult to map the pattern to actual 3D representations of the body, because ease allowance must be given. Therefore, pattern makers (dummy used) using traditional techniques need not allow any additional ease. However, it is necessary to confirm that they actually fit well on real people body shape. Adjustments to garments are often required. Since we are using scanned 3D real human body data as opposed to pattern makers’ dummies, our system must generate patterns that include a degree of ease. Our method first constructs, a virtual body shape, which includes ease, and then generates tight skirt patterns for the constructed virtual body shape using a computerized pattern making system. By using 3D real human body shape directly, it is not necessary to confirm actually fit well on real people body shape. This system has the potential to be much more efficient than traditional pattern making techniques. It also generates patterns which are much more suitable for each unique human body. Methods The lower body, waist, stomach, and hip shape express unique characteristics of an individuals’ shape. Thus, these factors need to be considered when reconstructing 3D body shapes with ease allowance during the pattern making process. In our research, we develop a method of reconstructing 3D individual body shapes which retain individuals’ unique characteristics (Figure 1). Method of reconstruction of real human body shape In the reconstruction process, it is necessary to divide the body in three parts, from waist to stomach, from stomach to hip, and from hip to hemline. The following steps describe our method: (1) First, we construct a line model using 3D scanned body data. (2) We extract waist line (WL) having maximum Z-value of back shape line in the side view of 3D body model. Stomach line (SL) is extracted by line having maximum Z-value in front shape line under waist. Hip line (HL) is extracted by line having minimum Z-value in back shape line under stomach. Figure 2 shows WL, SL, and HL on lower body so that it is possible to divide into three parts, from WL to SL (I), from SL to HL (II) and from HL to hemline (III).

3D human body shapes

17

IJCST 22,1

18 Figure 1. Individual shape of stomach and hips

Front shape line

Back shape line

Y WL I SL

II

HL

Figure 2. 3D body shape and line model extracted WL, SL, and HL

III Z

(3) Extracted SL is arranged and copied to HL at regular intervals using sweep method. As a result, there are existing two lines on the same Y coordinates in the II area (Figure 3). These two lines are connected using convex hull method for gaining the besieging lines at each position in II area. The convex hull of a set of points is the smallest convex set that includes the points. For a two-dimensional finite set the convex hull is a convex polygon. When creating the besieging lines, it is possible to give ease allowance retaining an individuals’ shape in areas such as stomach and HLs shapes (Figure 4). (4) Using sweep method, the besieging line is copied from HL to hem line as III area (Figure 5). (5) Uneven lines from waist to stomach back shape line on I area are smoothed using a convex method. As shown in Figure 6, we can make the body shape with ease allowance for making patterns unique to an individual’s unique body shape.

3D human body shapes SL

19

II HL

Figure 3. The result of II area using sweep method

HL

HL

SL

SL

Figure 4. The convex method for creating the besieging lines of SL and HL (a) SL and HL

(b) Connected SL and HL

II

(c) Besieging line

HL HL

III

Figure 5. The result of sweeped HL on III area

Experiments 3D measurement of human body shape We used scanned ten subjects’ body data to examine the effectiveness of our reconstruction method. Tables I and II show size information of the ten subjects based on JIS size indication. For example, number three subject had difficulty choosing skirts because an M size, which is suitable for her waist is too loose around her hips. Six other subjects in the Tables I and II had similar problems with size choice. We try to reconstruct their body shape for individual pattern making and make unique skirts for each subject. We then examine how well they fit.

IJCST 22,1

I

I

20 Figure 6. The results of smoothed waist and stomach back lines on I area

Table I. JIS size indication (cm)

Waist size Hip size

Subject number

Table II. JIS size information of subjects (cm)

1 2 3 4 5 6 7 8 9 10

S

M

L

LL

58-64 82-90

64-70 87-95

69-77 92-100

77-85 97-105

Waist size

Hip size

M L M LL M L LL L L L

S L S L S L L L L M

Individual pattern development process Our development method is presented simply here (Cho et al., 2006). First, we make the 3D reconstructed body surface of the clothes using triangular patches and sets grainlines for weft and warp on 3D body surface. We arrange 12 cross-sectional grainlines at 158 intervals to make 14 sections. After we set grainlines accurately for weft and warp, we fit the fabric lattice to the contour surface. We form a fabric lattice with a mesh structure in weft and warp direction. We cut 3D surfaces using plane. We create patterns by making angle of fabric lattices at right angles from three dimensionally contoured panels into two dimensions. Panels can then be described using curved lines in this process, 3D cutting line is flattened on the two-dimensional pattern. Finally, we can achieve the pattern (Figure 7).

WL

WL

HL

HL

Sewing line

3D human body shapes

21

Figure 7. Individual pattern development process

Results Results of reconstructed 3D human body shapes Figure 8 show the results of reconstructions for subjects 5 and 8. As shown, our reconstructed body shapes include ease allowance. These reconstructed body shapes, which include ease allowance are used as body shapes for pattern development. Results of individual pattern development We obtained individual patterns using individually reconstructed 3D body shapes by computerized pattern development. Figure 9 shows the completed pattern of four subjects with four dart lines. In individual patterns, complex curved lines such as waistlines and dart lines are created automatically using our developed method. Even though subjects 3 and 5 have same size as shown Table II, their shape, amounts and length of waist and dart lines are different depending on individual body shapes. These lines are one of individual characters in pattern making. Result of making skirts using personalized patterns We made tight skirts using patterns created using our method and then examined fitness on each individual subject’s body. Figure 10 shows tight skirts for four subjects made using our pattern making system. Subjects indicated that the fit of our skirt. It shows how our method works for making patterns for different body sizes and shapes.

Figure 8. The reconstructions for Subjects 5 and 8

IJCST 22,1

22

Figure 9. The results of individual pattern

Figure 10. The result of making skirts using unique created patterns

Notes: Subjects of 3, 5, 7, 8

Notes: Subjects of 3, 5, 7, 8

Application for various designs There are various designs based on a basic skirt pattern. In this research, we tried to apply our method to make various design skirts with patterns unique to a given subject. Two subjects used our method to create various design. Figure 11 shows the results for the two subjects.

3D human body shapes

23 Front view

Front view

Side view

Side view

Notes: Subjects of 9,10

Conclusions We developed a method of reconstructing 3D individual body shapes, which retain an individual’s unique characteristics. It is part of a computerized pattern making system, which we developed. We could make the body shape with ease allowance for making patterns unique to an individual’s unique body shape using sweep method and a convex method. We obtained individual patterns using individually reconstructed 3D body shapes by computerized pattern development. In our patterns, complex curved lines such as WL and dart lines are created automatically using the developed method. We successfully used our method to make variations of a tight skirt to fit different size women. We also used our method to make other skirts of various designs. Our method is useful for making patterns and then garments, without the need for the garments to be later adjusted for the subject. References Bigliani, R., Eischen, J.W., House, D.H. and Breen, D.E. (2000), “Collision detection in cloth modeling”, Cloth Modeling and Animation, A.K. Peters, Natick, MA, pp. 199-205. Cho, Y.S., Komatsu, T., Park, H.J., Inui, S., Takatera, M. and Shimizu, Y. (2006), “Individual pattern making using computerized draping method for clothing”, Textile Research Journal, Vol. 76 No. 8, pp. 646-54. Cho, Y.S., Okada, N., Park, H.J., Takatera, M., Inui, S. and Shimizu, Y. (2005), “An interactive body model for individual pattern making”, International Journal of Clothing Science & Technology, Vol. 17 No. 2, pp. 91-9. Heisey, F., Brown, P. and Johnson, R.F. (1990a), “Three-dimensional pattern drafting – part 1: projection”, Textile Research Journal, Vol. 60 No. 11, pp. 690-6. Heisey, F., Brown, P. and Johnson, R.F. (1990b), “Three-dimensional pattern drafting – part 2: garment modeling”, Textile Research Journal, Vol. 60 No. 12, pp. 731-7.

Figure 11. The results of application for various designs

IJCST 22,1

24

Kim, S. and Park, C.K. (2007), “Basic garment pattern generation using geometric modeling method”, International Journal of Clothing Science & Technology, Vol. 19 No. 1, pp. 7-17. Miyoshi, M. (2002), The Dress Making, Bunka Women’s University, Tokyo. Miyoshi, M. and Hirokawa, T. (2001), “Study on the method of measuring a vacant space distance in a worn jacket for clothing pattern design”, Journal of the Japan Research Association for Textile End-Uses, Vol. 42 No. 4, pp. 233-42. Watanabe, Y. (1999), “Ordering your cloth to fit yourself”, The Journal of the Institute of Electronics, Information and Communication Engineers, Vol. 82 No. 4, pp. 404-11. Corresponding author Masayuki Takatera can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

Lipase treatment to improve hydrophilicity of polyester fabrics Hye Rim Kim and Wha Soon Song Department of Clothing and Textiles, Sookmyung Women’s University, Seoul, South Korea

Hydrophilicity of polyester fabrics 25 Received 17 February 2009 Accepted 12 June 2009

Abstract Purpose – The purpose of this paper is to investigate the conditions of the treatment using commercial lipase to improve the hydrophilicity of the polyethylene terephthalate (PET) fabrics. Design/methodology/approach – The lipase treatment conditions, such as the pH, temperature, treatment time, and concentration, are controlled by measuring the hydrolytic activity, moisture regain, and wettability of the treated fabrics. The effects of calcium ions on the moisture regain and wettability of the treated fabrics are also evaluated. Findings – The lipase treatment conditions for PET fabrics are controlled at a pH of 7.5, a temperature of 308C, a treatment time of 60 min, and a lipase concentration of 50 percent (owf). The moisture regain of the PET fabrics that are treated with lipase improved 3.3 times that of the untreated PET fabric. Calcium chloride did not affect the moisture regain of the treated fabrics but affected their wettability. The surface of the PET fabrics that are treated under optimum conditions and in the presence of calcium chloride showed many cracks and voids, unlike the surface of the untreated PET fabrics. Research limitations/implications – The lipase treatment did not affect the handle of the PET fabrics in the present paper because the weight loss is very small. Originality/value – In this paper, the control conditions for the improvement of the hydrophilicity of PET fabrics using the low-cost commercial lipase are determined. The results of the study could further the environment-friendly finishing of PET fabrics. Keywords Fabric production processes, Moisture measurement, Textile testing Paper type Research paper

1. Introduction Concerns regarding health, energy, and the environment drive the improvement of enzyme technology in the textile industry (Bielen and Li, 2002). Enzymatic processing has been developed for natural fibers in wide-ranging operations, from cleaning preparations to finishing (Cavaco-Paulo and Gu¨bitz, 2003; Kirk et al., 2002). In addition to natural fibers, enzymatic hydrolysis on synthetic fibers has been explored to enhance their hydrophilicity (Cavaco-Paulo and Gu¨bitz, 2003; Guebitz and Cavaco-Paulo, 2007), using lipases, polyesterases, and cutinases (Alisch-Mark et al., 2006; Chaya and Kitano, 1999; Chaudhary et al., 1998; Fischer-Colbrie et al., 2004; Guebitz and Hsieh and Cram, 1998; Kim and Song, 2006, 2008b; Vertommen et al., 2005; Walter et al., 1995; Yoon et al., 2002). Among these enzymes, lipases have been reported as hydrolyzing ester linkages in polyethylene terephthalate (PET), thus producing polar hydroxyl and carboxylic groups (Guebitz and Cavaco-Paulo, 2007; Hsieh and Cram, 1998; Kim and Song, 2006, 2008a; Yoon et al., 2002;). Lipases have been obtained from bacterial, fungal, and animal pancreases and are used as crude mixtures with

International Journal of Clothing Science and Technology Vol. 22 No. 1, 2010 pp. 25-34 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011008785

IJCST 22,1

26

other hydrolases or in purified form (Hassan et al., 2006). The industrial applications of lipases in synthetic fiber processing have been limited, however, due to their high-production cost and the limited number of lipases available in industrial amounts (Houde et al., 2004). The enzyme hydrolysis of PET has used novel lipases made for laboratory use (Alisch et al., 2004; Chaya and Kitano, 1999; Yoon et al., 2002) or high-cost commercial lipases (Kim and Song, 2006). Owing to these problems, PET hydrolysis using lipase has been limited to broaden its industrial application. The use of low-cost commercial lipases for PET hydrolysis is a possible way to further environmentally friendly finishing of PET fabrics. The majority of this research is directed towards improving the hydrophilicity of PET fabrics using the commercial lipase, Lipolase 100L from Novozymes, and relatively few studies have examined the use of commercial lipases to hydrolyze PET fabrics. Commercial lipases have low-hydrolytic activity on PET. This low activity can be improved, however, by adding activators during the processing. Lipases from bacteria and fungi are known to demonstrate improved activity in the presence of inorganic salts such as calcium ions (Jung, 2003). Several studies on the effects of calcium ions on lipase activity have been limited to substrates such as olive oil or trybutyrin (Decker, 1997; Sharma et al., 2001). Few studies have investigated the effect of calcium ions on PET fabrics during enzymatic processing (Kim and Song, 2006, 2008a). This study investigates the conditions of the treatment of PET fabrics with commercial lipase to improve the hydrophilicity of the PET fabrics. The hydrolytic activity of lipase is evaluated via the number of carboxylic groups, using the titration method. Each treatment condition, such as the pH, temperature, treatment time, and concentration, is controlled by measuring the hydrolytic activity, moisture regain, and wettability of the treated fabrics. The effects of calcium ions on the moisture regain and wettability of the treated fabrics are evaluated. 2. Experimental 2.1 Materials For the experiment, 100 percent PET fabric (test fabric from KS K 0905) was used (Table I). The PET fabric consisted of filament fibers and had plain weave structures. Aspergillus oryzae lipase abbreviated as AOL (EC 3.1.1.3, Lipolase 100L, Novozymes) was used without further purification. The activity of lipase is 100KLU/g at pH 7.0, 308C using tribytyrin as a substrate. Tris (hydroxymethyl) amino methane abbreviated as Tris (pKa 8.3 at 208C, Sigma Chemical Co.) was used as a buffer. Tris buffer solution was used as the basis for all applications. The pH of the buffer solution was adjusted with 1 M HCl (Duksan Pure Chemicals, Korea) and 0.1 M NaOH (Junsei Chemicals, Japan). Thymophthalein (TPH) from Aldrich Chemical Co. was used as an indicator. A total of 95 percent ethanol (Duksan Pure Chemicals, Korea) was used to inactivate the enzymes in the test solution during the titration. All the chemicals were used without further purification.

Fiber Table I. Characteristics of fabric

100 percent polyester

Yarn count (denier)

Fabric count (yarns/in.)

Fabric weight (g/m2)

Thickness (mm)

70

113 £ 95

70 ^ 5

0.094 ^ 0.02

2.2 Hydrolytic activity on PET fabric The hydrolytic activity of lipase was measured by the titration method (Kim and Song, 2008b; Sigmaaldrich.com, 2006; Japanese Standard JIS K 0601, 1995; Walter et al., 1995). Each fabric sample was cut into 4 £ 4 cm pieces that weighed approximately 0.1 g. Each sample was placed into a 20 ml vial bottle that contained 8 ml of the 50 mM Tris buffer. The PET fabrics were then treated with different temperature, pH, treatment time, and concentration. All treatments were performed at 150 rpm using a shaking water bath. After the treatments, the test solution was transferred to a 50 ml Erlenmeyer flask, and 20 ml ethanol was added to the test solution to inactivate its enzymes. Four drops of 0.9 percent TPH indicator was added to the test solution. The test solution was titrated with 0.1 M NaOH into a light blue color. Then, the volume of the 0.1 M NaOH used in the sample test was recorded. Each test was carried out ten times. The blank test performed in the same condition with the sample test, but enzymes were not included. Each test was carried out five times. The hydrolytic activity of lipase was calculated using the following equation (Kim and Song, 2008b; Sigmaaldrich.com, 2006; Japanese Standard JIS K 0601, 1995; Walter et al., 1995): Hydrolytic activity ðml=gÞ ¼

Vs 2 Vb PET

where Vs: volume of 0.1 M NaOH for the sample test, Vb: volume of 0.1 M NaOH for the blank test, and PET: weight of the PET sample (g). 2.3 Enzymatic treatment Each fabric sample was cut into specific dimensions and weighed approximately 1 g. Depending on pH, temperature, concentration, and treatment time, the PET fabrics were treated with lipase in Tris buffer solution, using a liquor ratio 80:1. All the lipase treatments were performed at 150 rpm using a shaking water bath (BS-21, Jeio Tech., Korea). The enzyme inactivation was performed at 808C for 10 min. The treated fabrics were thoroughly washed with water and dried at room temperature. Then, weight loss, moisture regain, wettability, and scanning electron microscope (SEM) micrographs of treated fabrics were measured. The weight loss was evaluated by the dry weight of the fabrics before and after the treatment. The moisture regain was evaluated according to ASTM D629-99. It was measured via the dry weight and the moisture conditioning weight of the treated PET fabrics. The samples were dried at 1058C for 1 h using a drying oven. After cooled, they were weighed in a desiccator for 30 min. The moisture conditioning was carried out at 208C with a 65 percent relative humidity for 24 h. The moisture regain of the PET fabrics was calculated using the following equation: Moisture regain ð%Þ ¼

Wm 2 Wd £ 100 Wd

where Wm: weight of the fabrics in a moisture equilibrium at 208C at 65 percent relative humidity and Wd: weight of the fabrics dried at 1058C for 1 h. The wettability of the treated fabrics was evaluated via water absorbency and water contact angle (WCA). The water absorbency of the PET fabrics was evaluated

Hydrophilicity of polyester fabrics 27

IJCST 22,1

according to the AATCC test method 79-1992. The WCA of the PET fabrics was measured using the contact angle measurement system (KRUSS DSA100, KRU¨SS, Inc., Germany). Each test was carried out ten times. Changes in the surface of the treated fabrics were analyzed with a SEM (JSM-5410, Japan) after the samples were plated with gold.

28

3. Results and discussion 3.1 Effects of the pH and temperature Figure 1 shows the hydrolytic activity of lipase on PET fabrics. PET fabrics were treated with 50 percent (owf) lipase for 60 min, using different pH levels from 7 to 8.5 and different temperatures from 25 to 608C. The highest value was achieved at a pH of 7.5 and a temperature of 308C. The hydrolytic activity of the treatment at 308C showed approximately a twofold increase over that of the treatment at 258C, when the pH was controlled from 7.0 to 7.5. When the pH was controlled from 8.0 to 8.5, the hydrolytic activity of lipase remained low because the lipase was inhibited and denaturalized above the critical pH. The hydrolytic activity showed the highest values at a temperature of 308C and a pH of 7.5. Since the lipase actively hydrolyzed the ester linkages in the PET fabrics in this critical condition, the number of carboxylic groups increased. Also, the decrease in the hydrolytic activity at temperatures above 308C and a pH of 7.5 was related to the sensitive nature of the biocatalyst to the temperature and the pH. Enzymatic treatment above the critical temperature and pH would denaturalize the enzymes (Cavaco-Paulo and Gu¨bitz, 2003; Chaudhary et al., 1998; Kim and Song, 2008b). Figure 2 shows the effects of temperature on the moisture regain and wettability of PET fabrics. PET fabrics were treated with 50 percent (owf) lipase for 60 min at different temperatures, from 25 to 508C. The pH level was adjusted to 7.5. The moisture regain showed the highest value (1.579 percent ^ 0.06) at 308C, and improved 3.3 times

Hydrolytic activity (ml/g)

2.0

7.0 (pH) 7.5 8.0 8.5

1.5

1.0

0.5

Figure 1. Effects of the pH and temperature on the hydrolytic activity of the lipase-treated PET fabrics

0.0

25

30

35

40 45 50 55 60 Temperature (°C) Notes: Treatment conditions: 50 percent (owf) lipase; treatment time, 60 min

that of the untreated PET fabric. The WCA and the water absorption time of the lipase-treated PET fabrics decreased when the temperature was increased to up to 308C. The WCA and the water absorption time of the treated fabrics at 308C decreased 1.42 and 2.03 times those of the untreated PET fabrics, respectively. The highest degree of wettability of the PET fabrics was thus achieved at a temperature of 308C. The lipase hydrolytic activity affected the improvement of the moisture regain and wettability of the treated PET fabrics. The number of carboxylic groups increased because of the lipase-hydrolyzed ester linkages in the PET fabrics under the controlled treatment conditions, which improved the fabrics’ moisture regain and wettability. The highest levels of moisture regain and wettability of the PET fabrics were achieved at 308C and a pH of 7.5.

Hydrophilicity of polyester fabrics 29

25

30 35 40 45 Temperature (°C)

105 100 95 90 85 80 75 70 65 0.0

50

110 Water absorbency (sec.)

1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

WCA (degree)

Moisture regain (%)

3.2 Effects of the treatment time Figure 3 shows the effects of the treatment time on the hydrolytic activity. Over different time periods from 10 to 360 min, the fabrics were treated with 50 percent (owf)

25

30 35 40 45 Temperature (°C)

50

100 90 80 70 60 50 40 0.0

25

30 35 40 45 Temperature (°C)

Notes: Treatment conditions: 50 percent (owf) lipase; pH, 7.5; treatment time; 60 min; , lipase-treated PET

50

,untreated PET;

Figure 2. Effects of temperature on the moisture regain and wettability of the lipase-treated PET fabrics at varying temperatures

Hydrolytic activity (ml/g)

1.8 1.6 1.4 1.2 1.0 0.8

0.00

0

50

100 150 200 250 300 350 400 Treatment time (minutes)

Notes: Treatment conditions: 50 percent (owf) lipase; pH, 7.5; temperature, 30°C

Figure 3. Effects of the treatment time on the hydrolytic activity of the lipase-treated PET fabrics

IJCST 22,1

30

lipase at a pH of 7.5 and a temperature of 308C. After the 60 min treatment, the hydrolytic activity improved 1.5 times that after the 10 min treatment. Above 60 min, the hydrolytic activity gradually decreased because a longer treatment time could cause denaturalization of enzymes or aggregation between enzyme molecules (Cavaco-Paulo and Gu¨bitz, 2003). Figure 4 shows the effects of the treatment time on the moisture regain of the PET fabrics. The fabrics’ moisture regain and wettability improved when the time was increased to up to 60 min, and showed the highest values at 60 min. The moisture regains of the PET fabrics that were treated for 60 min improved 3.3 times those of the untreated PET fabrics. The WCA and the water absorption time decreased rapidly as the treatment time was increased to 60 min, at which time the lowest value emerged. From the measurement of the hydrolytic activity and the wettability, the treatment time was controlled at 60 min.

Figure 4. The moisture regain and wettability at various treatment times

0

30 60 90 120 150 180 Time (minutes)

110

105 100 95 90 85 80 75 70 65 0.0

Water absorbency (sec.)

1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

WCA (degree)

Moisture regain (%)

3.3 Effects of the enzyme concentration Figure 5 shows the effects of the enzyme concentration on the hydrolytic activity. The PET fabrics were treated with lipase at different concentration levels that ranged from 10 to 100 percent (owf). At the 50 percent concentration, the hydrolytic activity improved almost 1.7 times that at the 30 percent concentration (owf). At concentrations above 50 percent (owf), however, the value of the hydrolytic activity stayed within the acceptable error range. Since the cleavage site of the PET fabrics was limited, the excess enzyme supplies could not help improve the hydrolytic activity of lipase (Cavaco-Paulo and Gu¨bitz, 2003; Kim and Song, 2008b). In addition, the excess enzyme supplies could have caused aggregation between the enzyme molecules. The enzymes would have been partly inactivated depending on the type of aggregation formed (Cavaco-Paulo and Gu¨bitz, 2003; Kim and Song, 2008b). Figure 6 shows the effects of the enzyme concentration on the moisture regain and wettability of the PET fabrics. Enzymatic treatment was carried out at a temperature of 308C and a pH of 7.5 for 60 min. The enzyme concentration was controlled from 10 to 100 percent (owf). The moisture regain of the treated fabrics improved when the lipase concentration was increased from 10 to 70 percent; and at lipase concentrations above 70 percent, the moisture regain decreased slightly. The WCA and the water absorption time decreased rapidly as the enzyme concentration increased to 50 percent (owf). The lowest values of the WCA and the

0

30 60 90 120 150 180 Treatment time (minutes)

100 90 80 70 60 50 40 0.0

0

30 60 90 120 150 180 Treatment time (minutes)

Notes: Treatment conditions: 50 percent (owf) lipase; pH, 7.5; temperature, 30°C; , lipase-treated PET

,untreated PET;

Hydrophilicity of polyester fabrics

2.5

Hydrolytic activity (ml/g)

2.0

31 1.5

1.0

0.00

Figure 5. Effects of the enzyme concentration on the hydrolytic activity of the lipase-treated PET fabrics

0

10 20 30 40 50 60 70 80 90 100 Concentration (%, owf) Notes: Treatment conditions: pH, 7.5; temperature, 30°C; treatment time, 60 min

2.0 110

1.8

WCA (degree)

Moisture regain (%)

1.4 1.2 1.0 0.8 0.6

Water absorbency (SEC.)

100

1.6

90 80 70

0.4

100 90 80 70 60 50 40

0.2 0.0

0.0 0 10 20 30 40 50 60 70 80 90 100

0 10 20 30 40 50 60 70 80 90 100

Concentration (%, owf)

Concentration (%, owf)

Notes: Treatment conditions: pH, 7.5; temperature, 30°C; treatment time, 60 min;

0.0

0 10 20 30 40 50 60 70 80 90 100 Concetration (%, owf)

, untreated PET;

, lipase-treated PET

water absorption time were obtained at a 50 percent lipase concentration. At above 50 percent lipase concentrations, the WCA and the water absorption time increased. The lipase concentration was controlled at 50 percent (owf) via the measurement of the hydrolytic activity, moisture regain, and wettability of the PET fabrics. The lipase treatment condition was controlled at a pH of 7.5, a temperature of 308C, a treatment time of 60 min, and a lipase concentration of 50 percent (owf). 3.4 Effects of calcium chloride Figure 7 shows the effects of the calcium chloride on the moisture regain and wettability of the PET fabrics. The PET fabrics were treated in the presence of different amounts of calcium chloride, from 1 to 50 mM. The moisture regain did not significantly differ when there was calcium chloride and when there was none. The WCA and the water absorption time, however, decreased considerably where there was calcium chloride. The WCA and the water absorption time

Figure 6. The moisture regain and wettability at various enzyme concentrations

IJCST 22,1

32

of the PET fabrics that were treated in the presence of 3 mM of calcium chloride decreased almost 1.2 and 1.35 times their levels, respectively, when there was no calcium chloride. The improved wettability of the PET fabrics was probably due to the increase in the number of voids and cracks on the surface of the PET fabrics in the presence of calcium chloride. The increase in the number of voids and cracks was confirmed as shown in Figure 8. Also, these surface changes caused the weight loss of the treated fabrics. The weight losses of the treated fabrics in the presence of calcium chloride and in the absence of it were 0.317 percent ^ 0.075 and 0.152 percent ^ 0.067, respectively. In the presence of calcium chloride, the weight loss of the treated fabrics improved because the lipase retained its activity in the presence of calcium ions (Sharma et al., 2001). In the authors’ previous study on porcine pancreas lipase (Kim and Song, 2008b), the wettability of the treated fabrics improved in the presence of 30 mM calcium chloride. Unlike the previous study, the small number of calcium ions in this study helped improve the hydrolytic activity of AOL because the calcium ions helped activate more lipases from bacteria and fungi than from animal pancreases (Jung, 2003). The weight loss of the treated fabrics with calcium ions was too small, however, to affect the handle of the fabrics. 3.5 SEM micrographs Figure 8 shows the SEM micrographs of the PET fabrics during their lipase treatment. The surface of the PET fabrics that were treated under optimum conditions and in the presence of calcium chloride had many cracks and voids, unlike the surface of the untreated PET fabrics. Even though the voids and cracks were largely responsible for the improved wettability of the treated fabrics, they did not affect the handle of the fabrics because the weight loss was too small (Kim and Song, 2006; Kim and Song, 2008b). 2.0

90

1.8

80

80

1.6 1.4

Water absorbency (sec.)

WCA (degree)

Figure 7. The moisture regain and wettability of the lipase-treated PET fabrics in the presence of calcium chloride

Moisture regain(%)

70

70

60

1.2 50

60 50 40 30 20 10

1.0 0.0

0.0

0

0 5 10 15 20 25 30 35 40 45 50

0 5 10 15 20 25 30 35 40 45 50

0 5 10 15 20 25 30 35 40 45 50

CaCl2 (mM)

CaCl2 (mM)

CaCl2 (mM)

Notes: Treatment conditions: 50 percent (owf) lipase; pH, 7.5; temperature, 30°C; treatment time, 60 min;

Figure 8. SEM micrographs of the lipase-treated PET fabrics

Untreated PET

a Lipase-treated PET (pH 7.5, 30, 50% (owf), 60min.)

b Lipase-treated PET in the presence of 3mM CaCl2

4. Conclusion This study investigated the conditions of commercial lipase treatment of PET fabrics to improve the fabrics’ hydrophilicity. Each treatment condition, such as the pH, temperature, treatment time, and concentration, was controlled by measuring the hydrolytic activity, moisture regain, and wettability of the treated fabrics. The effects of calcium ions on the moisture regain and wettability of the treated fabrics were evaluated. The lipase treatment conditions were controlled at a pH of 7.5, a temperature of 308C, a treatment time of 60 min, and a lipase concentration of 50 percent (owf). The moisture regain of the PET fabrics that were treated with lipase improved 3.3 times that of the untreated PET fabric. Calcium chloride did not affect the moisture regain but affected the wettability of the treated fabrics. The surface of the PET fabrics that were treated under optimum conditions and in the presence of calcium chloride showed many cracks and voids, unlike the surface of the untreated PET fabrics. Even though the voids and cracks were largely responsible for the improved wettability of the treated fabrics, they did not affect the handle because the weight loss was very small. This study determined the control conditions for the improvement of the hydrophilicity of PET fabrics using commercial lipase. Using a large amount of lipase could limit lipase application in PET processing, though. This problem can probably be solved if enhanced lipases for PET processing are developed or if activators and auxiliaries for the improvement of the hydrolytic activity of lipases are studied.

References Alisch, M., Feuerhack, A., Mu¨ller, H., Mensak, B., Andreaus, J. and Wimmermann, W. (2004), “Biocatalytic modification of polyethylene terephthalate fibers by esterases from actionomycete isolates”, Biocatalysis and Biotrasformation, Vol. 22 Nos 5/6, pp. 347-51. Alisch-Mark, M., Herrmann, A. and Zimmermann, W. (2006), “Increase of the hydrophilicity of polyethylene terephthalate fibers by hydrolases from Thermomonospora fusca and Fusarium solani f. sp. pisi”, Biotechnol. Lett, Vol. 28, pp. 681-5. Beilen, J.B.V. and Li, Z. (2002), “Enzyme technology: an overview”, Current Opinion in Biotechnology, Vol. 13 No. 4, pp. 338-44. Cavaco-Paulo, A. and Gu¨bitz, G.M. (2003), Textile Processing with Enzymes, The Textile Institute, New York, NY, pp. 96-191. Chaudhary, A.K., Beckman, E.J. and Russell, A.J. (1998), “Enzymes for polyester synthesis”, ACS Sym. Ser., Vol. 684, American Chemical Society, Washington, DC, pp. 18-57. Chaya, E. and Kitano, M. (1999), “Possibility of modifying polyester fibers using lipases”, Sen-I Gakkaishi, Vol. 55 No. 5, pp. 150-4. Decker, L.A. (1997), Worthington Enzyme Manual: Enzymes, Worthington Biochemical, Lakewood, NJ, pp. 112-29. Fischer-Colbrie, G., Heumann, S., Liebminger, S., Almansa, E., Cavaco-Paulo, A. and Guebitz, G.M. (2004), “New enzymes with potential for PET surface modicition”, Biocatal. Biotransfor., Vol. 22 Nos 5/6, pp. 341-6. Guebitz, G.M. and Cavaco-Paulo, A. (2007), “Enymes go big: surface hydrolysis and functionalisation of synthetic polymers”, Trends Biotechnol., Vol. 26 No. 1, pp. 32-8. Hasan, F., Shah, A.A. and Hameed, A. (2006), “Industrial applications of microbial lipases”, Enzyme Microb. Tech., Vol. 39 No. 2, pp. 235-51.

Hydrophilicity of polyester fabrics 33

IJCST 22,1

34

Houde, A., Kademi, A. and Leblanc, D. (2004), “Lipases and their industrial applications”, Appl. Biochem. Biotech., Vol. 118, pp. 155-70. Hsieh, Y.L. and Cram, L.A. (1998), “Enzymatic hydrolysis to improve wetting and absorbency of polyester fabrics”, Text. Res. J., Vol. 68 No. 5, pp. 311-19. Japanese Standard JIS K 0601 (1995), Determination of lipolytic activity of lipase for industrial use. Jung, D.H. (2003), Enzymology, Dae Kwang Seo Rim, Seoul, pp. 177-86. Kim, H.R. and Song, W.S. (2006), “Lipase treatment of polyester fabrics”, Fiber. Polym., Vol. 7 No. 4, pp. 339-43. Kim, H.R. and Song, W.S. (2008a), “Effects of Triton X-100 and calcium chloride on the porcine pancreas lipase treatment of PET fabrics”, J. Korean Society of Clothing and Textiles, Vol. 32 No. 6, pp. 911-17. Kim, H.R. and Song, W.S. (2008b), “Optimization of enzymatic treatment of polyester fabrics by lipase from porcine pancreas”, Fiber. Polym., Vol. 9 No. 4, pp. 423-30. Kirk, O., Borchert, T.V. and Fuglsang, C.C. (2002), “Industrial enzymes applications”, Curr. Opin. Biotech., Vol. 13 No. 4, pp. 345-51. Sharma, R., Chisti, Y. and Banerjee, U.C. (2001), “Production, purification, characterization, and applications of lipases”, Biotechnology Advances, Vol. 19, pp. 627-62. Sigmaaldrich.com (2006), “Sigma quality control test procedure”, available at: www.sigmaaldrich. com/sigma/enzyme%20assay/l3126enz.pdf (accessed 11 November 2006). Vertommen, M.A.M.E., Nierstrasz, V.A., van der Veer, M. and Warmoeskerken, M.M.C.G.J (2005), “Enzymatic surface modification of poly(ethylene terephthalate)”, J. Biotechnol., Vol. 120, pp. 376-86. Walter, T., Augusta, J., Muller, R.J., Widdecke, H. and Klein, J. (1995), “Enzymatic degradation of a model polyester by lipase from Rhizodus delemar”, Enzyme Microb. Tech., Vol. 17, pp. 218-24. Yoon, M.Y., Kellis, J. and Poulose, A.J. (2002), “Enzymatic modification of polyester”, AATCC Review, Vol. 2, pp. 33-6. Corresponding author Wha Soon Song can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

Geometric disposition of threads in single-layer woven structures

Geometric disposition of threads

Elena Chepelyuk Department of Design, Kherson National Technical University, Kherson, Ukraine

Valeriy Choogin Department of Mechanical Technology of Fibre Materials, Kherson National Technical University, Kherson, Ukraine, and

35 Received 20 March 2009 Revised 1 June 2009 Accepted 1 June 2009

Jenny Cousens and Michael Hann The School of Design, University of Leeds, Leeds, UK Abstract Purpose – The purpose of this paper is to analyse the advantages of a new interpretation of the geometric disposition of threads within woven fabric structures, and to develop a method of determining the parameters of threads, with reference to each order of their disposition. Design/methodology/approach – Based on the analysis of the geometrical models proposed by Barker and Midgely, by Pierce and by Novikov, the substantiation of the advantages of a stricter model, offered by the authors, for determining the geometric disposition of threads within single layer woven fabric structures with the help of the tangent function is given. This model allows the substantial expansion of the actual bounds of the interval of the order of the geometric disposition of threads in woven fabric structures to 0.2-9.8. Findings – The tangent function can approximate the crimp height ratio of the warp threads within the woven fabric structure with accuracy within the limits of geometric disposition angle change from 18 to 898. Research limitations/implications – The work has applications in the industrial production of woven fabrics. Practical implications – This research will allow the design of a woven fabric with practically any ratio of crimp height for the warp and weft threads to effectively achieve the required performance characteristics of the cloth. Originality/value – This paper extends the knowledge of the geometrical characteristics of woven fabric structure, and proposes intelligent methods of determining the parameters of thread cross-sections in accordance with the orders of the geometric disposition of threads in woven fabric structure. Keywords Fabric testing, Thread, Modelling, Geometry Paper type Research paper

1. General introduction This paper presents theoretical developments of the geometrical models of woven fabric structure proposed previously by Barker and Midgley (1914), Peirce (1937) and Novikov (1946a, b, c). These established geometrical models conform, to varying degrees to the actual arrangement of threads in a woven fabric, on which the quality and speed of the development of new woven fabric structures to a large extent depends. 1.1 Literature considerations The study of woven fabric geometry was pioneered by Barker and Midgley (1914) and Peirce (1937), and many of the later investigations have been based on their models.

International Journal of Clothing Science and Technology Vol. 22 No. 1, 2010 pp. 35-48 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011008794

IJCST 22,1

36

Figure 1 shows the two geometrical models of ordinary woven fabric with a plain weave interlacing proposed by Barker and Midgley. Figure 1(a) shows a section of the woven fabric with the warp threads at an extreme position in relation to the central line of the cloth (A0 ), the weft thread bending without straight sections, equal diameters of warp and weft threads, and with identical crimp height hwp ¼ hwft ¼ a. In Figure 1(b), the warp threads are twice the diameter of the weft threads. The model is again of a plain weave, but woven as a weft rib. There is greater space between the warp ends and they do not deviate from the centre line (hwp ¼ 0). The thinner weft threads are bent significantly and have the maximum crimp height – hwft ¼ Max. These two extreme variants of thread crimp in a woven fabric are the basis for subsequent research by various authors. A simpler geometrical model, proposed by Peirce is shown in Figure 2. This geometrical model assumes that the threads have circular cross-sections and are Y

Z

Weft

Warp

B

B hwp/2

a

c

A′

A hwft/2 C

C′

b Y′

Z′ (a) Repeat Weft

Warp A

A

Figure 1. Models of Barker and Midgley

(b)

Weft

Warp D C θ1

d2 θ1

Figure 2. Geometrical model of Peirce

d1

A

θ1

h1/2

O h2/2

B P2

incompressible. The geometrical parameters are determined by the following conventional signs: †

ABCD

– the length of the warp thread in these circumstances consists of circular arcs (AB and CD) and straight lines (BC).

†

u1

– the inclination angle of warp in relation to the plane of the woven fabric for the two warp and weft systems.

†

d1 , d 2

– diameters of the warp and weft threads.

†

h1 , h 2

– maximum displacements of the warp and weft threads axes, normal to the plane of the woven fabric.

†

p2

– distance between the weft threads.

This model identifies a property of significant interest for this research. For a single-layer fabric, Peirce showed that the sum of the diameters of the warp and weft threads is equal to the sum of the displacement heights of the crimp in the structure of a single-layer woven fabric: d 1 þ d 2 ¼ h 1 þ h2

ð1Þ

The Peirce model is of limited practical use unless supported by computer calculation. Further research shows an updating of the geometrical model of the unit cell of plain weave by Peirce, in particular, the replacement of the circular cross-section of the warp and weft with an elliptical one (Peirce, 1937). In an attempt to reflect the actual behaviour of threads within woven fabrics many authors have offered other geometrical models which take into account the deformation of the threads within the structure (Zvorikina, 1946; Surnina, 1973). Based on the full range of thread arrangements in woven fabric, from the equal crimping of warp and weft threads to the absence of crimp in warp or weft, identified by Barker and Midgley (1914), and the development of the theory by Peirce (1937) and Novikov (1946a, b, c) developed a theory of woven fabric structure phases, which allows the specification of the range of probable variants of geometrical thread disposition within a woven fabric structure. The term “phase” is indicative of a particular form of thread disposition, between the two extremes of equal crimping and the absence of crimping in either warp or weft. The object of this paper is the substantiation of the advantages of the author’s interpretation of these phases of geometrical thread disposition within single-layered woven fabrics, based on the fundamental work by Novikov (1946a, b, c). The substantive postulates of Novikov’s theory can be considered as the initial position for this research. 2. Geometrical analysis of the disposition of threads within single-layered woven structures by Novikov According to the theory of geometric thread disposition by Novikov (1946a, b, c), the arrangement of threads, with a circular cross-section, in a single-layer fabric of plain weave can be identified by one of nine structure “phases” (Figure 3). The first phase relates to the extreme case where the warp remains straight within the woven fabric, and only the weft bends. At the ninth phase, the contrary is true: the weft remains straight, and the warp bends to the maximum degree. Between the first

Geometric disposition of threads 37

IJCST 22,1

dwp = dwft Weft

I

Warp hwp = 0

hwft = 4r

Warp

Weft

38

II hwft = 3.5

hwp = 0.5r

III hwp = 1r

hwft = 3r

hwp = 1.5r

hwft = 2.5

hwp = 2r

hwft = 2r

IV

V

VI hwp = 2.5r

hwft = 1.5

VII hwp = 3r

hwft = 1r

hwp = 3.5r

hwft = 0.5

VIII

Figure 3. Phases of geometrical disposition of threads in single-layer woven structures by Novikov

Warp hwp = 4r

Weft

IX

Weft

hwft = 0 Warp

and ninth phases the warp and weft both bend to varying degrees. The fifth phase corresponds to the model of Barker and Midgely, shown in Figure 1(a), where there is equal crimp height of warp hwp and weft hwft (designations are ours hereinafter). Novikov suggested determining the difference between each phase by a rate of half of the thread radius (0.5 r). An important assumption is made. In transition to the following phase the decrease in the crimp height of the warp thread hwp is equal to increase in the crimp height of the

weft thread hwft. For example, with equal diameters dwp ¼ dwft of warp and weft threads, the crimp height of the warp thread hwp increases by 0.5 r, and the crimp height of the weft thread hwft decreases by the same value. Thus, it is observed that dwp þ dwft ¼ hwp þ hwft ¼ 4r. As a basic characteristic of woven fabric structure Novikov suggested using the crimp height ratio of warp hwp and weft hwft. Surnina (1973) suggested that the crimp height ratio of the warp and weft threads be expressed by the coefficient K h ¼ hwp =hwft (designated from this point as K Nh ). Table I presents the value of the coefficient K Nh ¼ hwp =hwft for all nine phases identified by Novikov. It is necessary to note that for the first phase K Nh ¼ 0=8 ¼ 0; and for the ninth, K Nh ¼ 8=0 ¼ 1. At the first phase, the warp threads can theoretically occupy a position one above the other and similarly at the ninth with the weft threads. It is noted, however, that this is theoretical conjecture, and that it is practically impossible to obtain these positions. Another essential disadvantage is the integer designation of the phases. In practice, we are compelled to use imprecise expressions (designation), for example: “the given woven fabric has a structure between the third and fourth phase, closer to the third”. In his work Novikov also considered a variation of his phase model for woven structures with unequal thread diameters, using the example dwp ¼ 2dwft. Considering the phenomenon whereby threads deform or flatten, Novikov provided an opportunity to interpret the cross-section of threads as elliptical, as shown by the work of Zvorikina (1946). In order to eliminate the disadvantages, while preserving the valuable essence of Novikov’s theory, a regularity of change in the value of the crimp height ratios of threads in a woven fabric, in the process of transition from a considered phase to the subsequent, was established:

Geometric disposition of threads 39

K Nh ¼ 0:143ðN F 2 1Þ þ 0:0235ðN F 2 1ÞðN F 2 2Þ þ 0:005ðN F 2 1ÞðN F 2 2ÞðN F 2 3Þ þ 0:00108ðN F 2 1ÞðN F 2 2ÞðN F 2 3ÞðN F 2 4Þ þ 0:000425ðN F 2 1ÞðN F 2 2ÞðN F 2 3ÞðN F 2 4ÞðN F 2 5Þ þ 0:0001958ðN F 2 1ÞðN F 2 2ÞðN F 2 3ÞðN F 2 4ÞðN F 2 5ÞðN F 2 6Þ þ 0:0001972ðN F 2 1ÞðN F 2 2ÞðN F 2 3ÞðN F 2 4ÞðN F 2 5ÞðN F 2 6ÞðN F 2 7Þ

Thread disposition phase I II III IV V VI VII VIII IX

Crimp height of threads Warp hwp Weft hwft 0 0.5 r 1.0 r 1.5 r 2.0 r 2.5 r 3.0 r 3.5 r 4.0 r

4.0 r 3.5 r 3.0 r 2.5 r 2.0 r 1.5 r 1.0 r 0.5 r 0

ð2Þ

Crimp height ratio of threads K Nh ¼ hwp =hwft 0 0.143 0.333 0.6 1.0 1.666 3.0 7.0 1

Table I. Value of coefficient K Nh for the crimp height ratio of the threads in a woven fabric proposed by Novikov

IJCST 22,1

40

Analysis of the cumbersome formula (2) shows an essential limitation of its use. Geometrical approximation of function (2) is shown on Figure 4 in the form of dotted curve K Nh which, as an example, allows the search for a simpler law Kh ¼ f(NF). 3. Tangential law of distribution of the phases of geometric disposition of threads in single-layer woven structures In Figure 4 point M(Kh ¼ 1, NF ¼ 5), corresponding to the average (fifth) phase of geometric thread disposition in woven fabric structures on curve K Nh ¼ f ðN F Þ also coincides with curve K Nh on the chart of the tangent function. It is equal to the tangent 458, corresponding to the middle of the range of change of argument NF. On Figure 4, the continuous curve K tg h ¼ f ðN F Þ shows the law of change Kh according to function tg aF in the interval of actual values of thread disposition angle aF ¼ 18-898. This interval was not casually chosen. An attempt to keep the existing representation about phases of mutual bend of threads in the woven fabric to a maximum led to the following initial position of searches of the range of argument variation. It is known, from Novikov, that at the fifth phase of geometrical thread disposition, the warp and weft threads have identical crimp height values: hwp ¼ hwft, corresponding to coefficient Kh ¼ 1.0. The tangent of angle aF ¼ 458 is also equal to 1.0. Therefore, on Figure 4 point M with coordinates (5; 1) and (458, 1) is common for curves K Nh and K tg h . With the exception of the values unsuitable for practical use tg aF ¼ 0 at aF ¼ 08 and tg aF ¼ 1 at aF ¼ 908, order NF(1) of the first phase of thread disposition (corresponding to the first phase by Novikov) is approximated by thread 11

Ratio between values of warp and weft crimp Kh = hwp/hwft

10 9

Novikov KhN

8 7 6

The tangent function Khtg

5

∆ϕM

4 3 ϕN (M) ϕtg (M)

2 M

1

Figure 4. Comparison of the phases of thread disposition by Novikov and the tangent function

0 0.2

1

2

1°

5°

15° 25° 35° 45° 55° 65° 75° 85° 89°

3

4

5

6

7

8

9

9.8

Phase number NF Phase angle

αF

disposition angle aF1 ¼ 58, and the order of the ninth phase – aF9 ¼ 858. The interval of change between each subsequent order NF(iþ 1) from previous NF(i ) corresponds to a change of thread disposition angle aFi of 108. Thus, the range of possible variations in the crimp height ratios of threads in woven structures, under the tangent law K tg h ¼ hwp =hwft in limits NF ¼ 1-9, changes for the tangent function from 0.087 (by Novikov Kh(1) ¼ 0.0) to 11.430 (by Novikov Kh(9) ¼ 1). Of practical interest is the increased possibility of using the tangent law for a wider range of variation of argument – thread disposition angle aFi: from 18 to 898 corresponding to final numerical values of coefficient of orders K tg h ¼ tg aF ¼ 0:0175 and 57:29. Thus, using the tangent function essentially enables us to increase the quantity of thread disposition phases (from NF ¼ 0.2 up to NF ¼ 9.8 inclusive). It allows the calculation of practically any set arrangement of threads in a single-layer fabric for implementation on a weaving machine. Here, it is appropriate to pay attention to the necessity of coordinating the thread disposition order with the physical parameters of threads and structures of the woven fabric, in order to exclude the possibility of designing unworkable structures or an unreasonable overestimation of woven fabric density. Table II shows numerical values for the coefficient of thread disposition order K tg h under the tangent law in the specified limits of variation aF, and also K Nh by Novikov, for convenience of comparison (Chepelyuk and Choogin, 2007). With the purpose of essentially increasing the information value, the following universal designation of the coefficient of the crimp ratio Kh of the given thread disposition order NF of the structure of single-layer fabrics is offered for practical use: 3;2 3;2 K NhaFF . For example, the notation K h27 ¼ tg a27 should be read as “Kh is calculated for a tangent of thread disposition angle aF ¼ 278 and corresponds to the thread disposition order NF ¼ 3.2”. The fractional part 0.2 is determined in view of the interval between the integer values of the orders of 108 from the ratio: (278 2 258)/108. Designations for the thread disposition orders for the above-mentioned values aF ¼ 18 and 898 will 0:2 9:8 accordingly have the appearance: K h1 and K h89 . Here, fractional parts of the orders are determined as follows:

Thread disposition orders of woven fabric structure, NF 0 0.2 1 2 3 4 5 6 7 8 9 9.8 10

Geometric disposition of threads 41

Coefficients of crimp height ratios of threads in a woven fabric Thread disposition Under N.G. Novikov Using the tangent N angle, grad aF K h ¼ hwp =hwft function K tg h ¼ tg aF 0 1 5 15 25 35 45 55 65 75 85 89 90

– – 0:8 ¼ 0.000 1:7 ¼ 0.143 1:3 ¼ 0.333 3:5 ¼ 0.600 1:1 ¼ 1.0 5:3 ¼ 1.666 3:1 ¼ 3.00 7:1 ¼ 7.000 8:0 ¼ 1 – –

0 0.0175 0.087489 0.267949 0.466308 0.700207 1.0 1.428148 2.144507 3.732051 11.430052 57,29 1

Table II. Conformity of coefficients of the crimp height ratio of threads in a woven fabric to the thread disposition orders under Novikov’s theory and with use of the tangent function

IJCST 22,1

42

½ð18 2 08Þ ¼ 18
¼ 0:2; 58

½ð898 2 858Þ ¼ 48
¼ 0:8 58

4. Thread parameters in single-layer woven structures The theory offered for the geometric disposition of threads in two-dimensional woven fabric structures, based upon the tangential law of thread disposition order variation, gives the designer the possibility of predicting more precisely the properties of an elaborate woven fabric, ensuring fitness for purpose. 4.1 Parameters of the crimped thread During weaving, threads are in a stressed state and undergo deformations of stretching, bending and compression. It results in a condensing of the elementary fibres, and a reduction in the primary area of thread cross-section. At the fell of the cloth, on the weaving machine, the effects of condensing and crimping are caused by the physical properties of weft and warp threads to resist the bending and stretching, and also by the action of the driven elements of the weaving machine. The cross-section of the threads, due to mutual pressure, cannot remain in their initial form of a circle (a special case of the ellipse). The cross-section of threads is naturally transformed into the ellipse. The major (awft for weft and awp for warp) and the minor (bwft for weft and bwp for a warp) semi-axis of ellipses have to be determined with consideration for coefficients of crimp hawft , hbwft for weft and hawp , hbwp for warp:

hawp hbwp hawft hb ; awp ¼ d wp · ; bwft ¼ dwft · wft ; bwp ¼ dwp · ; ð3Þ 2 2 2 2 where dwft and dwp – initial conditional diameters of weft and the warp threads before their crimp, in mm. awft ¼ dwft ·

4.2 Transformation of thread parameters when changing the thread disposition order of single-layer woven fabric structures With the purpose of minimising the tensely-deformed conditions of threads by producing a defined thread disposition order of woven fabric structure during weaving, and the essential simplification of process of forming a woven fabric structure upon the release of external loading, an obvious model of weft arrangements, with changeable parameters in a unit cell of the woven fabric, with a limiting density at all 11 phases: from 0.2 up to 9.8 (Figure 5) is offered below. Using only the physical properties of the warp and weft threads, without special measures, it is possible to obtain any desired geometric thread disposition on a weaving machine. For this purpose, it is necessary to identify certain ratios of conditional diameters of weft and warp, with reference to the given thread disposition of the structure. For example, if the average fifth thread disposition order is obtained simply enough with an equality of conditional diameters of weft and warp threads, then for achievement of the extreme thread disposition orders 0.2 and 9.8 it is necessary to precisely understand the following two key rules: (1) For reducing the thread disposition order from the fifth down to 0.2 it is necessary to considerably reduce the cross-section of the weft threads

Warp (0.2)

hwft (5)

Weft (1)

Weft (5)

Lwft (0.2)

Area of crumple

hwft (0.2)

Weft (9)

Ο1

Ο2 (0.2)

βwft (1)

Lwft (5)

Ο2 (1)

Transformation of warp

Weft (0.2)

Weft (9.8)

Ο2 (2)

Ο2 (3)

βwft (5) Ο2 (4)

βwft (9)

Ο2 (5)

Ο2(6)

Weft (5)

Ο2 (7)

Ο2 (8)

Ο2(9)

Ο2 (9.8)

Transformation of weft

hwft (9.8)

Weft (9.8)

Warp (9.8)

Geometric disposition of threads 43

Figure 5. The transformation of threads under alteration of the thread disposition order number

IJCST 22,1

44

whilst simultaneously increasing the corresponding cross-section of the warp threads. (2) For increasing the thread disposition order from the fifth up to 9.8 it is necessary to considerably increase the cross-section of weft threads whilst simultaneously reducing the cross-section of the warp threads. As a basis for the search for and analysis and choice of the most effective woven fabric structure it is recommended that the designer carry out calculations of the limiting density of thread arrangement at the first cycle of the development of the woven fabric structure. For further calculations we shall make use of the property of the single-layer fabric, proposed by Peirce (1937) that the sum of crimp heights of the warp and the weft is equal to the sum of their diameters ðhwp þ hwft Þ ¼ ðd wft þ dwp Þ. Accepting the replacement of round thread cross-section with the elliptical, we can write the equality for the fifth thread disposition order as: ðhwp þ hwft Þ ¼ ð2bwft þ 2bwp Þ ¼ T wf ;

ð4Þ

followed by the important property: to keep the set thickness of the woven fabric Twf it is necessary to reduce the thickness of the warp threads 2bwp accordingly when increasing the thickness of the weft threads 2bwft (and vice versa). Further, it is necessary to pay attention to a significant feature: the value of Khi thread disposition order is expressed in relative units. It allows the acceptance of the crimp height of the weft for one, i.e. hwft ¼ 1, whilst the crimp height of the warp is considered as a part of hwft: hwp ¼ hwft £ K h

or

hwp ¼ T wf 2 hwft

ð5Þ

Then taking into account of crimp of threads it is possible to write: hwft þ hwp ¼ 2bwft þ 2bwp ¼ ð1 þ K h Þhwft

and

hwft ¼

2bwft þ 2bwp T wf ¼ ð6Þ 1 þ Kh 1 þ Kh

Then you should serially take the values of the thread disposition order coefficient under the tangent law K tg h , further – Kh(i ), and carry out calculations of the weft crimp height in the woven fabric hwp(i ) under the formula (6) and of the warp crimp height hwp(i ) under the formula (5) for every (i ) order. For further calculations, it is necessary to know the values of the semi-axes of the elliptical cross-sections of the weft and warp threads awft(i ) and bwft(i ) for each thread disposition order of the woven fabric structure. It is known that, by researching the 1-4th thread disposition orders, the thickness of the woven fabric is determined by means of the equation: T h ¼ hwft þ 2bwft ;

ð7Þ

and by researching the 6-9th thread disposition orders the thickness of the woven fabric is created by the sum: T h ¼ hwp þ 2bwp

ð8Þ

Upon the calculation of the value of the weft crimp height for each order hwft(i ) with the equation (6) and of the warp crimp heights hwp(i ) with equation (5) it is possible to

proceed to the calculation of the thickness of the threads in the direction of small semi-axis of the ellipse: For orders 0:2 2 4:0 : 2bwftði Þ ¼ T h 2 hwftði Þ and

Geometric disposition of threads

2bwpði Þ ¼ ð1 þ K hði Þ Þhwft 2 2bwftði Þ ð9Þ For orders 5:0 2 9:8 : 2bwpði Þ ¼ T h 2 hwpði Þ and 2bwftði Þ ¼ ð1 þ K hði Þ Þhwpði Þ 2 2bwpði Þ Further, it is necessary to determine the size of threads along the major semi-axis of the ellipse for each thread disposition order. It is known, that: awft hawft ¼ bwft hbwft

and

awp hawp ¼ ; bwp hbwp

ð10Þ

whence: awftði Þ ¼

bwftði Þ £ hawftði Þ

hbwftði Þ

and

awpði Þ ¼

bwpði Þ £ hawpði Þ

hbwpði Þ

ð11Þ

Table III demonstrates the effect a change in the geometrical parameters of warp and weft has on the various elements of geometrical thread disposition within in the woven fabric structure: the thread disposition order, the angle of thread disposition and the coefficient K tg h using the tangent law. The examples assume the use of warp and weft threads of equal linear density and the same fibre composition and construction for the middle equation (5) order of thread disposition. A designer can clearly and quickly identify the parameters of warp and weft threads, and therefore the ideal order of geometrical thread disposition in the woven fabric structure to achieve the required aesthetic properties. Here, it is necessary to emphasize a very important circumstance: any single variant of geometrical parameter combinations of warp and weft threads guarantee the minimum of process intensity in forming the fabric on the weaving machine. 4.3 Practical experimental research of two-dimensional woven fabric structures Figure 6 shows examples of variations in crimp of the warp thread. The warp was R29.2/2 tex yarn in a single-layer woven structure of plain weave interlacing with different weft yarn thicknesses: . R7.5/2 tex; . R29.2/2 tex; and . R50/2 tex. The warp yarn is coloured blue, with the weft threads woven in white or cream. To photograph the cross-section, the fabric is cut along the length of a warp. Figure 6(a) shows the finest thickness of weft (R7.5/2 tex). The warp threads show minimal bending. The fabric exhibits the thread disposition order < 1. Figure 6(b) shows equal

45

18 28080 58 158 258 358 458 558 658 758 858 878540 898

0.2 0.4266 1 2 3 4 5 6 7 8 9 9.58 9.8

0.017455 0.037245 0.087489 0.267949 0.466308 0.700207 1.0 1.428148 2.144507 3.732051 11.430052 27.271486 57.289962

K tg h

Crimp height ratio of warp and weft threads

Notes: Centi-milli-metre (cmm) ¼ 0.01 mm ¼ 102 5 m

aGDT grade

NGDT

Table III. Effect of warp and weft parameters on the variations in geometrical disposition of threads in woven fabric structures

Thread disposition angle

16.2 18.46 24.59 47.52 71.71 97.15 124.51 155.26 191.79 237.88 301.29 324.79 334.6

Weft Twft tex

Parameters of elliptical cross-section

334.6 324.79 301.29 237.88 191.79 155.26 124.51 97.15 71.71 47.52 24.59 18.46 16.2

70.76 69.42 66.2 56.78 49.1 42.35 36.0 29.65 22.9 15.22 5.79 2.55 1.24

1.24 2.58 5.79 15.22 22.9 29.65 36.0 42.35 49.1 56.78 66.2 69.45 70.76

6.49 6.93 8.0 11.12 13.66 15.9 18.0 20.1 22.34 24.88 28.0 29.07 29.51

29.51 29.07 28.0 24.88 22.34 20.1 18.0 15.9 13.66 11.12 8.0 6.93 6.49

9.74 10.4 12.0 16.68 20.49 23.85 27.0 30.15 33.51 37.32 42.0 43.61 44.26

44.26 43.6 42.0 37.32 33.51 30.15 27.0 23.85 20.49 16.68 12.0 10.39 9.74

Warp Weft Warp Weft Warp Weft Warp Twp hwft hwp bwft bwp awft awp tex (0.01 mm) (0.01 mm) (0.01 mm) (0.01 mm) (0.01 mm) (0.01 mm)

Crimp height

46

Threads disposition order

Number of filling threads per metre

IJCST 22,1

Geometric disposition of threads 47 Weft

Warp

(a)

Warp

Weft

(b)

Warp

Weft

(c)

bending of warp and weft threads, with the use of yarn of equal thickness (R29.2/2 tex in warp and weft). The fabric exhibits the thread disposition order < 5. Figure 6(c) shows the greatest thickness of weft (R50/2 tex), therefore the warp thread shows significant bending. The fabric exhibits order < 9.

Figure 6. Variation in the disposition of warp threads (R29.2/2 tex) in a plain weave woven structure interlacing with weft of (a) R7.5/2 tex; (b) R29.2/2 tex; (c) R50/2 tex

IJCST 22,1

48

The designer has individual control of the thread disposition within the woven fabric structure using this technology of experimental fabric specimen research. 5. Conclusion The approximation of the possible variations in the ratios of crimp height of warp and weft in woven fabric structure hwp/hwft with the precise tangent law essentially expands the limits of the actual use of the crimp height ratio coefficient of threads in the woven fabric structure K tg hi ¼ tg aF from 0.2 up to 9.8 (instead of 2-8 by Novikov (1946a, b, c)), and raises the efficiency and speed of calculations. Using the basic property of a single-layer fabric proposed by Peirce (1937), about the equality of the sum of diameters of warp and weft threads to the sum of their crimp heights, it is proposed that it is necessary to take into account an important property of the woven fabric: to keep the set thickness of the woven fabric it is necessary to reduce the thickness of the warp threads accordingly when increasing the thickness of the weft threads (and vice versa). For the practical calculation of the parameters of the elliptical section of threads in the warp and the weft, the use of the method of ratio of thickness of weft and warp threads with reference to each order of geometrical disposition of threads in the woven fabric structure is proposed. It will allow, without applying special measures and without excessive stress (overstrain), using only the physical properties of the warp and weft threads, the minimisation of the tensely deformed thread conditions and the production of practically any desired woven fabric structure (or phase) on the weaving machine. References Barker, A.F. and Midgley, E. (1914), Analysis of Woven Fabric, Scott, Greenwood & Son, London, pp. 72-9. Chepelyuk, E.V. and Choogin, V.V. (2007), “Tangent law of the distribution of phases order of one-layer woven fabric structures”, The Bulletin of Kiev National Univ. of Technol. and Design, No. 6, pp. 111-7 (in Russian). Novikov, N.G. (1946a), “On the woven fabric structure and its designing applying the geometrical method”, J. Technol. of Text. Ind., No. 2, pp. 9-17 (in Russian). Novikov, N.G. (1946b), “On the woven fabric structure and its designing applying the geometrical method”, J. Technol. of Text. Ind., No. 6, pp. 24-8 (in Russian). Novikov, N.G. (1946c), “On the woven fabric structure and its designing applying the geometrical method”, J. Technol. of Text. Ind., Nos 11/12, pp. 17-25 (in Russian). Peirce, F.T. (1937), “Geometry of cloth structure”, J. Text. Inst., Vol. 28, pp. T45-T96. Surnina, H.F. (1973), The Design of Woven Fabric on Set Parameters, Light Industry, Moscow, pp. 27-48 (in Russian). Zvorikina, E.K. (1946), “Investigation of the phenomenon of weft contraction on weaving”, dissertation, Textile Industry, Moscow (in Russian).

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

A new methodology for the development of sizing systems for the mass customization of garments Maria L. Mpampa, Philip N. Azariadis and Nickolas S. Sapidis Department of Product and Systems Design Engineering, University of the Aegean, Ermoupolis, Greece

Mass customization of garments 49 Received 27 June 2008 Revised 12 March 2009 Accepted 12 March 2009

Abstract Purpose – The purpose of this paper is to derive a new method for developing sizing systems for the mass customization of garments. Design/methodology/approach – A range of recently published works has been studied. A new method is derived by following a basic statistical analysis on anthropometric data which are supported by an iterative mass customization model and introduced “satisfaction performance” indices. The derived method is applied successfully to an anthropometric data consisting of 12,810 Greek men. Findings – With the proposed method, it is possible to control the degree of mass customization and the corresponding number of garment sizes. Under this way, a balance between the number of sizes (in other words: production cost) and the percentage satisfaction of consumers can be achieved. The proposed method consists of six subsequent tasks which are applied to the target population data for the development of mass customization models for male shirts, coats and trousers. Research limitations/implications – Future work could be focused on the development of methods for the automatic garments grading with respect to the proposed mass customization models and practise. Originality/value – The methodology presented in this paper can be applied to the development of mass customization models for other categories of garments and target population. Keywords Garment industry, Human anatomy, Mass customization Paper type Research paper

1. Introduction Garments are manufactured massively using predefined size charts which allow for the reduction of production cost. It is, therefore, practically impossible to obtain a perfect fit between a piece of cloth and an individual buyer (Kotha, 1995; Pine, 1993). Owing to their low cost, preˆt-a-porter garments are dominating the modern markets, while partial individualization is achieved using sizing systems with normalized dimensions. Under this way, absolute individualization is sacrificed to the benefit of production economy (Fralix, 2000; Walter, 2006). The concept of mass customization is devised to serve the individualized needs of consumers and increase their satisfaction percentage (Anderson et al., 1999). This term implies a strategy for producing customized garments with maximum differentiation through a low-cost production process (Davis, 1987). Nowadays, this manufacturing model is enabled by modern information technologies, computer-aided design and manufacture systems, three-dimensional (3D) body scanners, interactive web-based

International Journal of Clothing Science and Technology Vol. 22 No. 1, 2010 pp. 49-68 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011008802

IJCST 22,1

50

applications, etc. (Ashdown and Delong, 1995; Salusso-Deonier, 1989; Gazzuolo et al., 1992). From the manufacturers’ point of view, there is an apparent trend on producing individualized garments with the greatest possible differentiation without affecting the production cost (DeLong et al., 1993). However, the relation between the size charts and body dimensions is not constant because of the changes that occur in the human population. Recent body surveys in Germany proved that a garment sizing system for a certain body type does not cover more than the 25 per cent of the population in which it is addressed (Lanenegger and van Osch, 2002; Walter, 2002). Hence, body measurements should be updated regularly in order to provide current information on essential sizes and their geographical distribution in the population (Istook et al., 2003). Consequently, for a successful garment mass customization model, the development and maintenance of up-to-date anthropometric databases of the target market population is essential (Salusso-Deonier, 1982; LaBat and Delong, 1990; Workman, 1991; Goldsberry et al., 1996; Ashdown, 1998). Such an approach necessitates the existence of a proper methodology for producing sizing systems (which are not proportionally graded) with respect to a target population and the corresponding garments type(s). These sizing systems should satisfy the majority of the target population and at the same time should imply a cost-effective and affordable production process by the garment manufacturer (Fralix, 2000; Walter, 2006). In the present work, a new methodology for the development of effective sizing systems is proposed, aiming at the introduction of mass customization in the manufacturing process of a garment company. With the proposed methodology one is able to derive a sizing system appropriate for a particular group of people (i.e. children, elderly, solders, etc.). The effectiveness of the proposed methodology is illustrated by applying it to the development of size charts for a target population consisting of 12,180 Greek men between 20 and 30 years old. With the obtained charts, we were able to produce garments that satisfy up to 92.4 per cent of the population. In the next section, we introduce the main methodology for producing sizing systems with respect to a specific target population and garment type, and with a variable degree of mass customization. Sections 3 and 4 present an application of the new methodology for producing certain garments for a Greek population consisting of 12,810 men. Finally, Section 5 concludes the paper giving also some remarks for future work. 2. The proposed methodology for the development of sizing systems 2.1 Review of existing works and main definitions One early empirical sizing system, the CS 215-58 Standard, was developed in the USA in 1958 based on the manufacturers’ experience. In 1970, another empirical sizing system, the PS 42-70 Standard of the USA, was developed using military anthropometrical data and a “trial-and-error” approach. Both sizing systems were based on out-of-date measurements taken at 1941 (Ashdown, 1998). Later, the development of new anthropometric databases demanded the classification of human bodies under various body types. This classification was achieved by Salusso-Deonier (1982) through a method called as “Principal component sizing system (PCSS)”. Soon after, Tryfos (1986) proposed another method based on the “integer programming” approach in order to minimize the number of the different sizes in a size chart. This method attempts to optimise an objective function (or indicator) named as the

“aggregate loss of fit” which measures the difference between real body dimensions and the produced size charts. In 1994, the American Society for Testing and Materials (ASTM) developed the ASTM D5585-94 Standard of the USA utilizing the experience of garment designers and market information. This sizing system was validated using US army and navy anthropometrical data. Ashdown (1998) and McCulloch et al. (1998) focused on the development of sizing systems more satisfactory than the ASTM D5585-94 using the “aggregate loss of fit” indicator in combination with the PCSS method. Gupta and Gangadhar (2004) proposed a “statistical” method for the minimisation of the “aggregate loss of fit” indicator with respect to the population’s body-type distribution. These models, however, were not developed with respect to the mass-customisation concept but to support the mass production of garments. Loker et al. (2005) describe a variety of size-specific statistical and visual analysis methods than can be applied to market body scan data to improve the apparel fit of a sizing system. They apply their method to modify an existing sizing system of a garments company. Finally, Hsu and Wang (2005) use a decision tree-based data-mining approach to establish a sizing system for the manufacture of garments, which allows for a wider coverage of body shapes with a fewer number of sizes and generates regular sizing patterns and rules. Their method is specialised for men’s pants. In this paper, we adopt two definitions originated in Lanenegger and van Osch (2002) for the characterization of garment dimensions. These definitions will be referred later by using the terms constraints A and B: . Definition 1 (constraint A). A dimension is called as primary when it plays an essential role in assessing when a garment is wearable by a consumer or not. In other worlds, a primary dimension tells when a piece of garment is able to cover the entire body of a consumer or not. . Definition 2 (constraint B). A dimension is called as secondary when it plays an essential role in assessing a garment’s fitting. In other words, a secondary dimension measures how a piece of garment (that fulfils constraint A) fits in a consumer’s body. Both constraints A and B are mentioned in the present paper as manufacturing constraints and vary with respect to the garment’s type (i.e. upper/lower body clothing, etc.). The purpose of the introduced method is to provide a feasible mass customization model for a particular target group and a garment type, which will be also called as related garment. The chest girth, waist girth and height are commonly used for the manufacture of the upper body garments. The neck girth and sleeve/arm length are commonly used for the manufacture of shirts. The waist girth and inside leg length are commonly used for the manufacture of lower body garments. In the methodology proposed in this paper, a set of body measurements (linear dimensions of height, inside leg length, sleeve/arm length, and girth dimensions of chest, neck and waist) of a target population is analyzed producing primary and secondary sizes that correspond to the primary and secondary dimensions of the related garment(s), respectively. The input body measurements should comply with the manufacturing constraints of the related garment(s). Then a sizing system is derived with the minimum number of different

Mass customization of garments 51

IJCST 22,1

52

garment sizes according to a chosen mass customization model. Consumers’ satisfaction is assessed with a new tool that is proposed in this paper. 2.2 The proposed mass customization methodology The overall methodology consists of six subsequent stages, which are shown in Figure 1 and are detailed in the following paragraphs. Input: body measurements

Stage 1: Statistical analysis

Stage 2: Linear regression analysis

Stage 3: Classification of the target population in body types

Stage 4: Determination of primary sizes

Stage 5: Determination of secondary sizes for every primary size

Stage 6: Output mass customization model

Figure 1. The flow chart of the proposed mass customization methodology

Mass customization model 1

Mass customization model 2

Mass customization model 3

Case A

Mass customization model 4

Case B

Stage 1: statistical analysis of body data. A statistical analysis of body measurements takes place in order to determine the range of the various body dimensions. This analysis includes the calculation of the minimum (Min), maximum (Max) and mean value (Mean) as well as the standard deviation (SD) for every distinct set of body measurements. Stage 2: linear regression analysis. Using the least squares linear regression analysis technique, the correlation between different body dimensions is determined. In this way, one is able to distinguish the primary from the secondary dimensions of the sizing system; essential information for the manufacture of garment patterns (Robertson and Minter, 1996; Jarosz, 1999; Barroso et al., 2005). Generally, in this method, the selection of the primary and secondary dimensions is performed with respect to Constraints A and B in order to ensure that the produced sizing system will comply with the manufacturing constraints of the related garment and with the specific body measurements of the target population. This selection is facilitated using the correlation coefficient R. According to the BS 7231 Standard (BS 7231, 1990) two dimensions are related according to the following rules: if the correlation coefficient R , 0.5, then no relationship exists; if 0.5 , R , 0.75, then there is a mild relationship; if R . 0.76, then a strong relationship exists. Using R, it is possible to reduce the number of independent measurements by removing those which exhibit a strong correlation along the primary and secondary dimensions. Stage 3: classification of the target population in body types. The purpose of this stage is to classify the target population in body shapes according to the body measurements of each subject. The classification is performed utilizing: . the height dimension; and . the “drop value” DV ¼ [(Chest girth) 2 (Waist girth)] (Cooklin, 1999; Gupta and Gangadhar, 2004).

Mass customization of garments 53

The “drop value” parameter is used to classify the different body shapes of the target population by determining distinct relationships between key dimensions (Cooklin, 1999; prEN 13402-3, 2004). Based on drop values, the population is classified into categories which correspond to generally perceived body shapes. In this stage, we separate the target population in six body types according to their height (Table I) and in seven body types with respect to their drop value (Table II). This partition will allow of a sizing system covering a large range of different body shapes. Stage 4: determination of primary sizes. The determination of the primary dimension depends on the garment type and Constraint A. Therefore, using the results Body type Very short , Mean 2 2SD Mean 2 2SD , Short , Mean 2 SD Mean 2 SD , Normal , Mean Mean , Tall , Mean þ SD Mean þ SD , Very tall , Mean þ 2SD Too much tall . Mean þ 2SD

Height (cm) , 164 164-171 171-178 178-185 185-192 192 .

Table I. Classification of the population vs height

IJCST 22,1

54

of stage 2 analysis, we are able to associate a body dimension with the primary dimension that is used for the manufacture of the related garment. In this way, the target population is classified in primary sizes derived along the selected primary dimension. In Gupta and Gangadhar (2004), the primary size ranges between [(Mean 2 SD), (Mean þ SD)]. In this paper, we chose a larger range between [(Mean 2 2SD), (Mean þ 2SD)], in order to take into account a greater part of the population (. 95 per cent). This choice, however, does not necessarily affect the final number of the produced garment sizes, since in the proposed method one is able to select the degree of mass customization. A threshold d is determined afterwards in order to define the difference between two successive primary sizes xk, xkþ 1. In this way, the obtained sizing system satisfies Constraint A as long as the corresponding part of the population lies within the interval [Mean 2 2SD, Mean þ 2SD]. The resulted number of primary sizes is: n¼

ðMean þ 2SDÞ 2 ðMean 2 2SDÞ 4SD ¼ d d

ð1Þ

Although the development of a sizing system can be based either on a constant or a variable d (McCulloch et al., 1998), in the proposed method, a constant d is selected for compatibility with the prEN 13402-3 (2004) standard. The effectiveness of this particular selection is verified by the application of the proposed method in recent anthropometric data. Stage 5: determination of secondary sizes for every primary size. For every primary size, a set of secondary sizes is developed according to the type of garment and Constraint B. Similarly to the primary dimension, the secondary dimension lies also within [Mean 2 2SD, Mean þ 2SD] and the number of obtained secondary sizes n is calculated through equation (1). All secondary sizes are grouped together into two-dimensional tables with respect to the selected primary dimension and are utilized for the development of the required sizing system. Stage 6: selection of a mass customization model. Each mass customization model defines a sizing system that starts with the “medium body shape” (according to Stage 3) and proceeds to covering “neighboring body shapes” according to this model. Each model depends on the desired mass customization degree which is iteratively improved in sequential steps. For each obtained mass customization model, a sizing system is developed according to the following scheme: (1) Step 0: mass production. Development of medium sizes (primary sizes) only, which are traditionally used for the mass production of preˆt-a-porter garments. Body type

Table II. Classification of population according to the “drop value”

Very small Small Medium Full Large Extra large Very extra large

Difference of chest – waist girth (cm) .18 14-18 10-14 6-10 2-6 2 2 to 2 , 22

(2) Step 1: mass customization model 1. Development of two secondary sizes for every primary size according to the first secondary dimension. (3) Step 2: mass customization model 2. Development of eight secondary sizes for each primary size according to the first and second secondary dimensions. This model is valid for garments with two secondary dimensions. (4) Step 3: mass customization model 3: . Case A (only one secondary dimension exists). Development of four secondary sizes for every primary size according to the secondary dimension. . Case B (at least two secondary dimensions exist). Development of up to 25 sizes along the two secondary dimensions for every primary size. These sizes are produced for the two body categories which are “before” and “after” the medium body shape. (5) Step 4: mass customization model 4. Development of all possible sizes for every primary and secondary dimension. This model corresponds to a sizing system with the maximum population satisfaction and the larger number of different garment sizes. This is the best possible degree of mass customization for a particular target population. Our experiments show that the resulted garments fit to the 99.9 per cent of the target population. The selection of a mass customization model is facilitated using a new assessment tool called as “Total satisfaction percentage.” The new index is based on the value of “satisfaction percentage” according to the following definitions. Definition 3 (“satisfaction percentage”). For each size k of the sizing system of a given mass customization model j, and for each garment type i, the satisfaction percentage aijk of the target population is equal to the percentage of the target population that its three characteristic (one primary and two secondary) body dimensions (xi, yi, zi) fall within the corresponding garment sizes (xik, yik, zik) up to a maximum tolerance value (Dxi,max, Dyi,max, Dzi,max), i.e. jxi 2 xik j # Dxi;max , jyi 2 yik j # Dyi;max , jzi 2 zik j # Dzi;max . Here: 1 ðDxi;max ; Dyi;max ; Dzi;max Þ ¼ ðdp ; ds1 ; ds2 Þ; 2 where dp, ds1, ds2 correspond to the constant difference between two successive values of the primary dimension, and the first and second secondary dimensions, respectively. Definition 4 (“total satisfaction percentage”). The “Total satisfaction percentage” aij of a target population with respect to a given garment type i and a mass customization model j is calculated by: n X aijk ð2Þ a
ij ¼ k¼1

Contrarily to existing approaches (see, e.g. the “aggregate loss” of fit (Tryfos, 1986; McCulloch et al., 1998; Gupta and Gangadhar, 2004)) the proposed indicator is able to measure the degree (or percentage) to which a certain sizing system “satisfies” the target population. On the other hand, the “aggregate loss” index simply expresses the average Euclidean distance between the dimensions of individuals and their allocated garment size. Therefore, the proposed index is able to assist a garment manufacturer to

Mass customization of garments 55

IJCST 22,1

56

select the mass customization model that is most appropriate for its production process and market pursues. 3. Application of the proposed methodology for the development of mass customization models for a Greek-men population The proposed methodology is applied to anthropometric data that correspond to 12,180 Greek men from the entire country aging between 20 and 30 years old. Six basic body dimensions were measured the time period between 2003 and 2004 by the traditional technique. These include the linear dimensions height, inside leg length, sleeve/arm length and the girth dimensions of chest, neck and waist. The purpose of this application is to derive mass customization models for the production of men: . shirts; . garments of the upper body (excluding shirts and underwear); and . garments for the lower body. In order to avoid redundant analysis, this section presents the first five stages of the application of the proposed method, while Section 4 summarizes the different mass customization models according to stage 6. 3.1 Stage 1: statistical analysis of body data The results of the statistical analysis of the six body dimensions of the 12,180 individuals are listed in Table III. The mean values (Mean) express the body dimensions of the “average” 20-30 years old Greek man. 3.2 Stage 2: linear regression analysis For each pair of body dimensions, the linear regression correlation coefficient R is calculated in order to determine the body dimensions that will be utilized for satisfying constraints A and B with respect to the related garment type. Table IV summarizes the results of this analysis. According to Table IV, there is a strong relationship (R ¼ 0.9948) between the chest and waist girth as it is also shown in Figure 2. For clarity reasons, the diagram points (rhombs) depict the mean value of the waist girths that correspond to a certain chest-girth value. Similarly, the chest girth vs the neck girth has very strong relationship, while there is a mild relationship between the chest girth and the height. There is no strong relationship between the waist girth and linear dimensions. The height has strong relationship with the linear dimensions sleeve and inside leg length. Dimension

Table III. The results of the statistical analysis of the six body dimensions of the target population

Mean

SD

Min Max Mean 2 SD Mean þ SD Mean 2 2SD Mean þ 2SD

Height 178 7 154 Chest girth 96.7 8.7 82 Waist girth 88.3 10.8 70 Neck girth 38.8 2.5 34 Inside leg length 80 5 65 Sleeve length 61.3 3 54

208 137 135 51 117 71

171 88 77 36 75 58

185 106 99 41 85 64

164 80 67 34 69 56

192 114 110 44 90 67

Summarizing, there is a strong linear correlation within length and girth dimensions, but there is no linear correlation between length and girth dimensions. This is an important observation since most empirical size diagrams are based on a linear increment of the sizing systems. In other words, it is mistakenly supposed that when the girth body size is increased, the height size is increased, respectively. Because of this issue, existing sizing systems result to clothes which are suitable to a limited percentage of the target population (Ashdown, 1998; Lanenegger and van Osch, 2002). Based on the above, the chest girth is selected as the primary dimension for upper body garments, while the neck girth is chosen as the primary dimension for the case of shirts. Also, the waist girth is considered as the primary dimension for the development of lower body garments. The rest of the dimensions like the height, sleeve or leg length will be selected as secondary dimensions with respect to one of the related garment types.

Mass customization of garments 57

3.3 Stage 3: classification of the population in body types 3.3.1 Population vs height. The resulted classification of the target population is depicted in Table V. The population is divided into six categories with respect to the height dimension.

Dimension

Height

Chest girth

Waist girth

Neck girth

Height Chest girth Waist girth Neck girth Inside leg length Sleeve length

1.0000 0.6354 0.6740 0.3886 0.9802 0.9927

0.7875 1.0000 0.9948 0.9866 0.0387 0.8531

0.7024 0.9928 1.0000 0.9712 0.2131 0.7664

0.7543 0.9634 0.9700 1.0000 0.4704 0.7671

Table IV. Correlation coefficient R of the body dimensions

140 130

Mean waist girth

120 110 100 90 y = 1.0445x – 12.121 R = 0.9948

80 70 60 80

90

100

110 Chest girth

120

130

140

Figure 2. Correlation of chest girth vs mean waist girth

IJCST 22,1

3.3.2 Population vs “drop value”. The resulted classification of the target population according to the drop value is listed in Table VI. Based on this analysis, it results that a large part of Greek male population has greater waist girth compared to the “medium” body type and it is categorized under the “full” body type. Thus, the difference between chest and waist girth for the major percentage of the population equals to 8 cm.

58

3.4 Stage 4: determination of primary sizes In this stage, the target population is classified in sizes according to the chosen primary dimension. More specifically: . the chest girth is selected for the manufacture of the upper body garments; . the neck girth is selected for the manufacture of shirts; and . the waist girth is selected for the manufacture of lower body garments. The chest girth dimension is subdivided setting by d ¼ 4 cm between two successive sizes according to the EN Standard (prEN 13402-3, 2004). This results to eight discrete sizes (n ¼ 8) in the range between 82 and 114 cm. Similarly, the neck girth dimension is subdivided according to d ¼ 1 cm resulting to ten discrete sizes (n ¼ 10) in the range between 34 and 44 cm. Finally, the waist girth is divided into sizes according to d ¼ 4 cm, which results to ten discrete sizes (n ¼ 10) in the range between 70 and 110 cm. 3.5 Stage 5: determination of secondary sizes for every primary size 3.5.1 Upper body garments (excluding shirts and underwear). For upper body garments (excluding shirts and underwear), the primary dimension is the chest girth and the secondary dimensions are the waist girth and the height. Both secondary dimensions lie within Mean 2 2SD and Mean þ 2SD. For example, for chest girth size 90-94, five waist girth sizes are developed (d ¼ 4) which correspond to individuals belonging to Body type

Table V. Classification of the population vs height

Very short , Mean 2 2SD Mean 2 2SD , Short , Mean 2 SD Mean 2 SD , Normal , Mean Mean , Tall , Mean þ SD Mean þ SD , Very tall , Mean þ 2SD Too much tall . Mean þ 2SD

Body type

Table VI. Classification of the target population according to the drop value

Very small Small Medium Full Large Extra large Very extra large

Height

Population

Percentage

, 164 164-171 171-178 178-185 185-192 192 .

240 2,040 4,568 3,978 1,143 211

2.0 16.7 37.5 32.7 9.4 1.7

Difference of chest – waist girth (cm)

Population

%

. 18 14-18 10-14 6-10 2-6 2 2 to 2 , 22

550 1,302 2,707 3,243 2,313 1,333 732

5 11 22 27 19 11 6

the Medium body type. This distribution is depicted with the 3D graph shown in Figure 3. Introducing the height as the second secondary dimension four more sizes are developed for every primary and first secondary size as it is shown in Table VII for chest girth size between 90 and 94 cm. These four height sizes correspond to the short, normal, tall and very tall body types. 3.5.2 Male shirts. The primary dimension for male shirts is neck girth and the secondary dimension is sleeve length. The secondary dimension lies within Mean 2 2SD and Mean þ 2SD. For example, for neck size 38, six sleeve length sizes are developed (d ¼ 2) which correspond to individuals belonging to the Medium body type. This distribution is depicted with the 3D graph shown in Figure 4. 3.5.3 Lower body male garments. For lower body male garments like the trousers, the primary dimension is the waist girth and the secondary dimension is the inside leg length. The secondary dimension lies within Mean 2 2SD and Mean þ 2SD. For example, for waist girth size 74-78, six inside leg length sizes are developed (d ¼ 3) which correspond to individuals belonging to the Medium body type. This distribution is depicted with the 3D graph shown in Figure 5.

Mass customization of garments 59

6 82-86

5

86-90 4

90-94

% 3

94-98

102-106

0

106-110

90-94

Total

106-110

98-102

90-94

82-86

102-106

94-98

Waist girth

Chest girth

Figure 3. 3D representation of the sizes produced according to the chest (primary dimension) and the waist girth (secondary dimension)

110-114 86-90

78-82

98-102

1

70-74

2

Chest girth

Waist girth

164-171

171-178

74-78 78-82 82-86 86-90 90-94

0.60 0.90 1.00 0.50 0.40 3.40

1.20 2.00 2.20 1.30 0.70 7.40

Height (per cent) 178-185 0.70 1.50 1.50 1.00 0.70 5.40

185-192

Total

0.10 0.30 0.40 0.30 0.10 1.20

2.60 4.70 5.10 3.10 1.90 17.40

Table VII. Sizes for upper body men garments (excluding shirts and underwear) and corresponding population percentages for chest girth size between 90 and 94 cm

IJCST 22,1

60 %

4.5

35

4.0

36

3.5

37

3.0

38

2.5

39

2.0

40 41

1.5

42

1.0

43

0.5

44

Sleeve length

39 37

4.0

70-74

3.5

74-78 78-82

3.0

82-86

2.5

86-90

% 2.0

90-94 94-98

1.0

98-102

0.5

102-106 94 98

86-90

78-82

70-74

Inside leg length

88-91

82-85

76-79

0.0

102-106

1.5

70-73

Figure 5. 3D representation of the sizes derived according to the waist girth (primary dimension) and the inside leg length (secondary dimension)

43

Neck girth

35

66-68

62-64

58-60

41

0.0 54-56

Figure 4. 3D representation of the sizes derived according to the neck (primary dimension) and sleeve length (secondary dimension)

106-110

Waist girth

4. Mass customization models for the Greek men population: application of stage 6 of the new methodology The mass customization models presented in section 2 are applied to the following categories of male garments: . shirts; . coats; and . trousers.

Mass production for preˆt-a`-porter garments correspond to conventional sizing systems. The rest sizing systems presented below correspond to the mass customization models 1-4. 4.1 Sizing systems for male shirts Ten medium sizes are proposed according to the mass production model of preˆt-a-porter shirts. The medium sizes correspond to sleeve length 60-62 cm (Table VIII): . According to mass customization model 1, 30 sizes are developed based on the sleeve length: ten for sleeve length 58-60 cm and ten for sleeve length 62-64 cm. . Mass customization model 2 is not applicable in this case because there is not a need to use a second secondary dimension. . Mass customization model 3 implies the development of 50 sizes: four more sizes based on the secondary dimension are developed for each medium size. Hence, ten more sizes are produced for sleeve length 56-58 cm, and ten for sleeve length 64-66 cm. . Mass customization model 4 implies the development of 59 sizes, which provides a comprehensive degree of mass customization for male shirts.

Mass customization of garments 61

4.2 Sizing systems for male coats Utilizing the chest and waist girth as well as height data of the target population, three mass customization models are developed for male coats. However, the obtained sizing systems can be regarded as general sizing systems for producing male upper garments excluding shirts and underwear. All results are summarized in Table IX, where for spacing reasons, we have omitted the sizes which correspond to chest girths between 94 and 110 cm: . According to mass production model, eight medium sizes based on the chest girth are produced. These regular sizes correspond to the waist girth and height of body type B (Medium). . According to mass customization model 1, 23 sizes are developed using the waist girth as secondary dimension. Seven sizes for body type A (small), eight for body type B (medium) and eight for body type C (full).

Sleeve length A B C D E F G H

55 57 59 61 63 65 67 69

54-56 56-58 58-60 60-62 62-64 64-66 66-68 68-70

0 35

1 36

2 37

0A 0B 0C 0D 0E 0F

1A 1B 1C 1D 1E 1F

2A 2B 2C 2D 2E 2F

3 38

Neck girth 4 5 39 40

6 41

7 42

8 43

9 44

3B 3C 3D 3E 3F 3G

4B 4C 4D 4E 4F 4G

6B 6C 6D 6E 6F 6G

7B 7C 7D 7E 7F 7G

8B 8C 8D 8E 8F 8G

9B 9C 9D 9E 9F

5B 5C 5D 5E 5F 5G

Table VIII. Sizing system for male shirts

IJCST 22,1 Chest girth

Body type

1 (82-86)

A-Small B-Medium C-Full D-Large E-Extra large A-Small B-Medium C-Full D-Large E-Extra large A-Small B-Medium C-Full D-Large E-Extra large

62 2 (86-90)

3 (90-94)

Table IX. Sizing system for upper body men garments excluding shirts and underwear

.. . 8 (110-114)

.

.

A-Small B-Medium C-Full D-Large E-Extra large

Height 2 (Medium) 3 (Tall) 171-178 178-185

Waist girth

1 (Short) 164-171

4 (Very tall) 185-192

70-74 74-78 78-82 82-86 70-74 74-78 78-82 82-86 86-90 74-78 78-82 82-86 86-90 90-94

1B1 1C1 1D1 1E1 2A1 2B1 2C1 2D1 2E1 3A1 3B1 3C1 3D1 3E1

1B2 1C2 1D2 1E2 2A2 2B2 2C2 2D2 2E2 3A2 3B2 3C2 3D2 3E2

1B3 1C3 1D3 1E3 2A3 2B3 2C3 2D3 2E3 3A3 3B3 3C3 3D3 3E3

1B4 1C4 1D4 1E4 2A4 2B4 2C4 2D4 2E4 3A4 3B4 3C4 3D4 3E4

94-98 98-102 102-106 106-110

8A1 8B1 8C1 8D1

8A2 8B2 8C2 8D2

8A3 8B3 8C3 8D3

8A4 8B4 8C4 8D4

According to mass customization model 2, 69 sizes are developed using the waist girth and height as secondary dimensions. Eight more sizes are produced for every primary size compared to the mass production model. According to mass customization models 3 and 4, 152 sizes are developed using the waist girth and height as secondary dimensions. About 24 more sizes are produced for every primary size compared with the mass production model. In this case, mass customization model 4 results to the same sizing system with the third one.

4.3 Sizing systems for male trousers Utilizing the waist girth and inside leg length, four mass customization models are developed for male trousers. All results are summarized in Table X: . According to mass production model, ten medium sizes based on the waist girth are produced. These medium sizes correspond to the inside leg length 79-82 cm. . According to mass customization model 1, 30 sizes are developed using the inside leg length as secondary dimension. About 20 more sizes than the mass production model are developed: ten for inside leg length 76-79 cm and ten for inside leg length 82-85 cm. . Mass customization model 2 is not applicable in this case since there is not a need to use a second secondary dimension. . According to mass customization model 3, 50 sizes are developed using the inside leg length. About 20 more sizes are developed compared to mass

A B C D E F G H

Inside leg length

72 75 78 81 83 86 89 92

70-73 73-76 76-79 79-82 82-85 85-88 88-91 91-93

0A 0B 0C 0D 0E 0F

0 72 70-74 1A 1B 1C 1D 1E 1F

1 76 74-78 2A 2B 2C 2D 2E 2F

2 80 78-82 3A 3B 3C 3D 3E 3F 3G

3 84 82-86 4A 4B 4C 4D 4E 4F 4G 4H

5A 5B 5C 5D 5E 5F 5G 5H

Waist girth 4 5 88 92 86-90 90-94 6A 6B 6C 6D 6E 6F 6G 6H

6 96 94-98 7A 7B 7C 7D 7E 7F 7G 7H

7 100 98-102

9 108 106-110 9A 9B 9C 9D 9E 9F 9G 9H

8 104 102-106 8A 8B 8C 8D 8E 8F 8G 8H

Mass customization of garments 63

Table X. Sizing system for lower body men garments

IJCST 22,1 .

64

customization model 2: ten for inside leg length 73-76 cm, and ten for inside leg length 85-88 cm. According to mass customization model 4, 73 sizes are developed. The produced sizing system satisfies 99.9 per cent of the target population.

4.4 Assessment of the developed sizing systems Table XI shows the results of applying the proposed satisfaction percentage index for the assessment of the developed mass customization models for the three aforementioned garment types. In the case of mass production (model 0), the total satisfaction percentage values are unacceptably low, particularly when the garment manufacturing constrains involve three distinct dimensions (e.g. coats). The satisfaction percentage is considerably increased with mass customization model 1, which expresses a first step towards mass customization for the case of shirts and trousers. Mass customization 2 is required for cloths like coats where three dimensions are utilized by the corresponding manufacturing constraints. This model provides a significant improvement to the cloths fitting compared to the previous one. Mass customization models 3 and 4 provide a considerable level of mass customization. The total satisfaction percentages of both models in the target population are very high, with the cost of producing a relatively large number of garment sizes. All results are shown in the graph of Figure 6. 5. Implementation Mass production strategies have driven apparel production for decades with a negative impact in design and fit of clothing. These strategies have categorized whole populations by a relatively small number of sizing systems and made it virtually impossible to meet the needs of those individuals who have special fitting requirements (Istook, 2002). The proposed methodology for the development of sizing systems combined with computer-aided and information technologies can enable the creation of garments, customized for fit, in a very quick and accurate manner. These customized garments can be inserted into normal production lines as an additional “size” and produced like every other garment of the same style. This means that successful companies with huge libraries of garment styles would be able to implement the proposed mass customization strategy with relatively little effort. Potential increase in production cost that would occur due to cutting a few garments at a time, rather than hundreds, could be offset with increases in sales and customer loyalty. The proposed method for mass customization can be implemented by using either manual measurements or automatic scanning as well as. In this paper, we applied the proposed method in sizing data taken by conventional manual methods which are usually subject to noise and inaccuracies. These shortcomings are not expected to appear in automatic scanning data which are high-accurate and filtered for noise removal. Finally, implementing a field research for the collection of anthropometric data are usually an expensive and time-consuming task which prerequisites the accomplishment of several criteria (Ujevic et al., 2006). Discussing on this task is, however, out of the scope of the present research. 6. Conclusions A new methodology for the mass customization of garments has been proposed in this paper. With the proposed method, it is possible to control the mass customization

0 1 2 3 4

10 30 – 50 59

24.9 67.4 – 90.2 92.4

0 1 2 3 4

8 23 69 152 152

9.6 24.0 53.2 83.3 83.3

0 1 2 3 4

10 30 – 50 73

23.4 55.9 – 80.3 93.9

Upper body garments Lower body garments Shirts Coats Trousers Model Number of sizes n Total satisfaction (%) Model Number of sizes n Satisfaction (%) Model Number of sizes n Total satisfaction (%)

Mass customization of garments 65

Table XI. The total satisfaction percentage of the target population according to developed mass customization models

66 Figure 6. The total satisfaction percentage of the target population with respect to the number of different garment sizes

100% Total satisfaction percentage aij

IJCST 22,1

80%

60%

40%

Shirts Coats

20%

Trousers

0% 0

50

100

150

200

Number of sizes n

degree according to four different models, which affect the corresponding sizing system of a specific garment type. In this way, a garments producer can choose the mass customization model that is closer to his manufacturing line and marketing pursues. This selection is facilitated through an assessment tool called “Total satisfaction percentage.” This statistical tool is in place to examine how much a sizing system “satisfies” the target population and it is harmonized with the European Standards of sizing systems development, which are going to be applied in the near future. The proposed methodology has been successfully applied for the development of mass customization models for male shirts, coats and trousers with respect to Greek men between 20 and 30 years old. This application indicates the limitations of conventional sizing systems and the advantages of using a “scalable” mass customization model. The methodology presented herein, can be applied to the development of mass customization models for other categories of garments and target population. Our future work includes the development of methods for the automatic garments grading with respect to the aforementioned mass customization models and practise. References Anderson, L.J., Brannon, E.L., Ulrich, P.V., Marshall, T., Staples, N., Grasso, M., Butenhoff, P. and Beninati, M. (1999), Discovering the Process of Mass Customization: A Paradigm Shift for Competitive Manufacturing, American Apparel Manufacturers Association, Auburn, AL, available at: www.auburn.edu/mcrp.html Ashdown, S.P. (1998), “An investigation of the structure of sizing systems: a comparison of three multidimensional optimized sizing systems generated from anthropometric data with the ASTM standard D5585-94”, International Journal of Garments Science and Technology, Vol. 10 No. 5, pp. 324-41. Ashdown, S.P. and Delong, M. (1995), “Perception testing of apparel ease variation”, Applied Ergonomics, Vol. 26 No. 1, pp. 47-54.

Barroso, M.P., Arezes, P.M., da Costa, L.G. and Miguel, A.S. (2005), “Anthropometric study of Portuguese workers”, International Journal of Industrial Ergonomics, Vol. 35, pp. 401-10. BS 7231 (1990), Part 1: Body Measurements of Boys and Girls from Birth to 16.9 Years, British Standards Institution, London. Cooklin, G. (1999), Pattern Grading for Women’s Clothes: The Technology of Sizing, Blackwell Science, Oxford, pp. 3-18. Davis, S.M. (1987), Future Perfect, Addison-Wesley, Reading, MA. DeLong, M., Ashdown, S., Butterfield, L. and Turnbladh, K.F. (1993), “Data specifications needed for apparel production using computers”, Garments Textile Research Journal, Vol. 11 No. 4, pp. 1-7. Fralix, M.T. (2000), “Mass customization using the internet”, Proceedings of the 80th World Conference of the Textile Institute, Manchester, 16-19 April. Gazzuolo, E., DeLong, M., Lohr, S., LaBat, K. and Bye, E. (1992), “Predicting garment pattern dimensions from photographic and anthropometric data”, Applied Ergonomics, Vol. 23, pp. 161-71. Goldsberry, E., Shim, S. and Reich, N. (1996), “Women 55 years and older: overall satisfaction and dissatisfaction with the fit of ready-to-wear: Part II”, Garments and Textile Research Journal, Vol. 14 No. 2, pp. 121-31. Gupta, D. and Gangadhar, B.R. (2004), “A statistical model for developing body size charts for garments”, International Journal of Garments Science and Technology, Vol. 16 No. 5, pp. 458-69. Hsu, C.-H. and Wang, M.-J.J. (2005), “Using decision tree-based data mining to establish a sizing system for the manufacture of garments”, International Journal of Advanced Manufacturing Technology, Vol. 26, pp. 669-74. Istook, C. (2002), “Enabling mass customization: computer-driven alteration methods”, International Journal of Clothing Science & Technology, Vol. 14 No. 1, pp. 61-76. Istook, C., Little, T., Hong, H. and Plumlee, T. (2003), “Automated garment development from body scan data”, NTC Project S00-NS15, National Textile Center Annual Report, November (formerly I00-S15). Jarosz, E. (1999), “Anthropometry of elderly women in Poland: dimensions for design”, International Journal of Industrial Ergonomics, Vol. 25, pp. 203-13. Kotha, S. (1995), “Mass customization: implementing the emerging paradigm for competitive advantage”, Strategic Management International, Vol. 16, pp. 21-42. LaBat, K.L. and Delong, M.R. (1990), “Body cathexis and satisfaction with fit of apparel”, Garments and Textile Research Journal, Vol. 8 No. 2, pp. 42-8. Lanenegger, R. and van Osch, R. (2002), “One size for Europe: a garment size system for Europe”, Avantex Frankfurt, 13 May. Loker, S., Ashdown, S. and Schoenfelder, K. (2005), “Size-specific analysis of body scan data to improve apparel fit”, Journal of Textile and Apparel, Technology and Management, Vol. 4 No. 3, pp. 1-15. McCulloch, C.E., Paal, B. and Ashdown, S.A. (1998), “An optimization approach to apparel sizing”, Journal of the Operational Research Society, Vol. 49, pp. 492-9. Pine, B.J. (1993), Mass Customization: The New Frontier in Business Competition, Harvard Business School Press, Boston, MA. prEN 13402-3 (2004), “Size designation of clothes – Part 3: measurements and intervals”, Final Draft.

Mass customization of garments 67

IJCST 22,1

68

Robertson, S.A. and Minter, A. (1996), “A study of some anthropometric characteristics of motorcycle riders”, Applied Ergonomics, Vol. 27 No. 4, pp. 223-9. Salusso-Deonier, C.J. (1982), “A method for classifying adult female body form variation in relation to the US standard for apparel sizing”, doctoral dissertation, University of Minnesota, Minneapolis, MN, available at: www.wsu.edu:8080/, salusso/BODY/s.html Salusso-Deonier, C.J. (1989), “Gaining a competitive edge with top quality sizing”, Quality Congress Transactions, Vol. 43, American Society of Quality Control, Toronto, pp. 371-6. Tryfos, P. (1986), “An integer programming approach to the apparel sizing problem”, Journal of the Operational Research Society, Vol. 37 No. 10, pp. 1001-6. Ujevic´, D., Rogale, D., Drenovac, M., Pezelj, D., Hrastinski, M., Narancˇic´, S.N., Mimica, Z. and Hrzˇenjak, R. (2006), “Croatian anthropometric system meeting the European Union”, International Journal of Clothing Science & Technology, Vol. 18 No. 3, pp. 200-18. Walter, L. (2002), “Will the “e-Tailor” become reality?”, Mass-Customization, Industrial Customisation, Industrial MtM and Personalised On-Line Shopping in the European Fashion Business – Project Results & Future Perspectives, Euratex, The EU Apparel Business Goes High-Tech, Brussels, 15 October. Walter, L. (2006), “The textile & clothing industry in Europe”, Textile Asia, Vol. 37 No. 4, pp. 40-6. Workman, J.E. (1991), “Body measurement specifications for fit models as a factor in garments size variation”, Garments and Textile Research Journal, Vol. 10 No. 1, pp. 31-6. Corresponding author Philip N. Azariadis can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

Recognition and re-visualization of woven fabric structures

Woven fabric structures

Pranut Potiyaraj Department of Materials Science, Faculty of Science, Center of Excellence in Textiles, Chulalongkorn University, Bangkok, Thailand and National Center of Excellence for Petroleum, Petrochemicals, and Advanced Materials, Chulalongkorn University, Bangkok, Thailand, and

79 Received 4 May 2009 Accepted 4 September 2009

Chutipak Subhakalin, Benchaphon Sawangharsub and Werasak Udomkichdecha Department of Materials Science, Faculty of Science, Center of Excellence in Textiles, Chulalongkorn University, Bangkok, Thailand Abstract Purpose – The purpose of this paper is to develop a computerized program that can recognize woven fabric structures and simultaneously use the obtained data to 3D re-visualize the corresponding woven fabric structures. Design/methodology/approach – A 2D bitmap image of woven fabric was initially acquired using an ordinary desktop flatbed scanner. Through several image-processing and analysis techniques as well as recognition algorithms, the weave pattern was then identified and stored in a digital format. The weave pattern data were then used to construct warp and weft yarn paths based on Peirce’s geometrical model. Findings – By combining relevant weave parameters, including yarn sizes, warp and weft densities, yarn colours as well as cross-sectional shapes, a 3D image of yarns assembled together as a woven fabric structure is produced and shown on a screen through the virtual reality modelling language browser. Originality/value – Woven fabric structures can now be recognised and simultaneously use the obtained data to 3D re-visualize the corresponding woven fabric structures. Keywords Pattern recognition, Image processing, Computer software, Fabric production processes Paper type Research paper

Introduction Recent advances in computer technology offer several automation processes in the weaving industry including structure recognition and characterization as well as 3D visualization of woven fabrics. Woven fabric structures greatly affect optical and mechanical properties of woven fabric. Accurate identification of the structural characteristics is required Sincere gratitude must be given to the Thailand Research Fund for funding this project. Also, this work would not have been possible without the support of National Center of Excellence for Petroleum, Petrochemicals and Advanced Materials, Chulalongkorn University.

International Journal of Clothing Science and Technology Vol. 22 No. 2/3, 2010 pp. 79-87 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011018577

IJCST 22,2/3

80

for quality control of the existing products, and is useful for developing new products. Traditionally, these characteristics, especially weave pattern, are manually determined by visual inspection. This manual operation is usually tedious, time consuming, and depends on the skills of the inspector. Some computerized methods for recognition of weave patterns from fabrics have been proposed, mostly based on image analysis techniques. Ohta et al. (1986) used a drum scanner to capture reflected images of woven fabrics. An image analysis was utilized to obtain the central lines of warps and wefts, and the fabric density was consequently determined. As described by Kinoshita et al. (1989) reflected and transmitted images of woven fabrics were captured using a charge-coupled device (CCD), and the obtained data were used to calculate the power spectra which were different for each type of weave. By acquiring reflected images from an image scanner, a similar technique utilizing fast Fourier transform to obtain power spectra was used to detect the appearance of reed marks (Sakaguchi et al., 2000). Using decision rules for identifying warp and weft floats based on geometric features of yarn distribution, the digitalized images captured from a microscope were analyzed to identify basic weave patterns. It was also reported that the determination of fabric counts using this technique show good agreement with manual measurement (Huang et al., 2000). Kang et al. (1999) used a CCD camera to capture fabric reflected and transmitted images perpendicular to the focal plane with halogen light mounted on the top of the fabrics. Both reflected and transmitted images were subjected to several image processing techniques. The results were combined in order to detect weave patterns. Visualization of woven fabrics has been extensively studied. The main challenge for applying 3D computer graphics to the simulation of woven fabric structures is to understand sufficiently the woven fabric structures (Hearle, 1994). A number of mathematical models have been developed in order to represent the thread paths in a woven fabric. One of the most important models is that of Peirce (1937) which assumes that yarn is incompressible, flexible and of circular cross-section. This geometrical model has been widely adopted and modified by many researchers (Lin, 1996). None of these models was believed to be the actual path, but all were reasonable approximations. The paths in real fabrics are complicated and depend upon yarn properties and forces, which cause the yarns to arrange themselves in positions of minimum energy, subject to external constraints and frictional effects (Lin and Newton, 1999). The structures can be represented in 2D fashion, taking the weave and some simple fabric properties such as warp and weft density into account. However, the 2D visualization cannot be totally satisfactory for today’s textile applications. Later, several programs were produced which resulted in graphics showing the thread path of fabrics in three dimensions, particularly that of Xu (1992). Her work made progress on the representations of varieties of single-layer weaves as well as a limited range of multi-layer fabrics. The employed method followed sine curves and straight lines to represent the yarn path within a fabric based on Peirce’s geometrical model for the plain weave and Love’s Peirce-like parameters based on Peirce’s geometrical model for non-plain weaves. Since then, other researchers have published papers that demonstrate solid-modelling techniques, for example, Keefe et al. (1992) and Keefe (1994a, b). Keefe’s work provides new possibilities for the simulation of woven fabrics, that is, the 3D models generated can be used further for mechanical analysis, and for engineering design and analysis of 3D woven fabrics. Lin (1996) also developed a program to generate and display various woven fabric structures in three-dimensions by computer graphics. The thread

paths displayed are generated by using cubic B-splines, the mathematics of which are well-developed, mainly for other areas of engineering design, especially in aerospace engineering. This work emphasized the true 3D solid model of limited typical weave structures. The program requires high-performance computers to process the simulation. Chen and his colleagues have developed a computerized simulation system for 3D woven fabrics based on their mathematical models (Chen et al., 1992). The structure has been parameterized and modelled mathematically (Chen et al., 1993a, b). The virtual reality modelling language (VRML) was first used for visualization of some woven structures based on the extrusion method in which the yarn cross-sectional image was swept along the predetermined yarn path (Chen and Potiyaraj, 1999). The structures were limited to angle-interlock and orthogonal structures. In their work, most of the weave parameters were also not taken into account. VRML is a computer language used for describing objects in a scene. VRML codes are simple ASCII text that can be parsed by a VRML interpreter (Ames et al., 1996). These interpreter programs are often called VRML Browser and are freely available as an add-on for several internet browsers. Lomov et al. (2007) successfully used VRML to represent both woven and knitted fabrics taking into account also mechanical parameters of the structures. In this research, image analysis techniques and algorithms for recognition of weave structures from reflected images of woven fabric were proposed. The weave patterns and parameters which were stored in digital format were, in turn, used for automatically generating 3D images of woven structures based on Peirce’s geometrical models. A Windows-based software system was developed to handle the seamless combination processes of weave pattern recognition and re-visualization of woven fabrics. Facilitating further development of fabric products, the system also allows various weave parameters to be adjusted in order to alter the appearance of the re-visualized 3D images, including yarn sizes, warp and weft densities, yarn colours as well as yarn cross-sectional shapes. Preparation and processing of fabric reflected images Two major types of woven fabrics were used as samples in this research, namely, plain and twill fabrics. They were scanned using an HP scanjet 5470c desktop flatbed scanner at the resolution of 600 dots per inch (dpi). It must be noted that the fabrics must be held properly so that warp and weft yarns were straight and perpendicular. Each square-inch scanned image was kept in the bitmap format. The scanned images of fabrics were converted into grayscale images. In order to amplify the distinction between free spaces and yarn-occupied spaces, the gray level at each pixel was reassigned by equalization the histogram. Equalization produces a flattened histogram with a more uniform gray level distribution so as to minimize the uneven distribution of gray levels of pixels caused by local illumination (Kang et al., 1999). An example of fabric before and after equalization is shown in Figure 1. Image analysis When scanning through a selected line in a prepared and processed fabric image, one encounters alternate bright and dark areas. A bright area is found when the yarn float is present, while a dark area is found when the yarn going down interlaces with the corresponding yarn or at the space not covered by warp or weft. Luminance at each pixel along a selected line in a fabric image can be calculated from RGB data according to

Woven fabric structures

81

IJCST 22,2/3

82

Figure 1. Scanned images

(a)

(b)

Notes: (a) A plain fabric; (b) a plain fabric after equalization

Grassmann’s law. A plot of luminance at each pixel along a selected line from a fabric image is shown in Figure 2. It can be seen that peaks are present in various widths and heights. Thus, the interpretation from this plot was not accurate enough. In order to smooth these peaks, the autocorrelation method was adopted using the following equations: C x;0 ¼

M X N X i

C 0;y ¼

Gi;j Gi2x;j

ð1Þ

Gi;j Gi;j2y

ð2Þ

j

M X N X i

j

where:

Figure 2. A luminance curve of a plain fabric image

Gi,j

is the luminance at coordinate.

M

is the maximum scanning point in the warp direction.

N

is the maximum scanning point in the weft direction.

Cx,0

is the autocorrelation value at points along the warp.

C0,y

is the autocorrelation value at points along the weft.

x and y

are the coordinates of pixel in the warp and weft direction, respectively.

400

400

350

350

300

300

250

250

200

200

150

150

100

100

50

50

0 R1

0

It was indicated that the autocorrelation technique was successfully adopted to determine structural repeat units in the fabric weave from the transmitted images of woven fabrics (Kang et al., 1999). Allowing a simple flatbed scanner to be used as an image capturing tool the same technique was applied to the reflected image, and a new algorithm to determine the weave structures was elaborated. When the luminance data were recalculated, a more uniform plot was obtained as shown in Figure 3. The width of each peak can be calculated by detecting the lowest point at each wave. This lowest point is the point where the slope of curve changes. This can be determined using the following equation: C x;0 ¼

C ðxþ1Þ;0 2 C ðx;0Þ 2x xþ1

ð3Þ

C 0; y ¼

C 0;ð yþ1Þ 2 C ð0; yÞ 2y yþ1

ð4Þ

Woven fabric structures

83

Scanning throughout the fabric image, the width of each peak in warp and weft directions was then combined and interpreted into a weave pattern. The spaces of yarns and the averages of the warp and weft yarns were also evaluated by calculating the position of each peak. Subsequently, warp and weft densities were determined by taking the reciprocal of the averaged spaces of warp and weft yarns (Huang et al., 2000). In order to represent a weave mathematically, a 2D binary matrix was used to keep the data of the weave pattern. An element value of the binary matrix is either 0 or 1. A value of 1 means the same as a mark on design paper indicating a warp-over-weft cross-over, and a value of 0, corresponds to a blank on design paper, meaning a weft-over-warp cross-over. The position of each element in the matrix is located by the co-ordinate (i, j), indicating the ith column from the left and the jth row from the bottom. Such a matrix is, hereafter, called a weave matrix. The determined weave patterns were digitally kept in this format. Simulation of 3D fabric structures There are several possible approaches which can be used for modelling the shape of yarns. As the most common method, yarn surfaces are approximately modelled from short cylinders or truncated cones with their axis along the direction of the yarn axis tangent. Lomov et al. (2007) created the yarn/ply surface using small simple plane elements, typically triangles. It was also discussed that a complicated model can be a 9,000

9,000

8,000

8,000

7,000

7,000

6,000

6,000

5,000

5,000

4,000

4,000

3,000

3,000

2,000

2,000

1,000

1,000 0

0 R1

Figure 3. An autocorrelated luminance curve of a plain fabric image

IJCST 22,2/3

84

tube of varying cross-sections which are created by moving the pre-defined cross-sectional shape along the yarn axis. The advantage of this method is that it can model a variety of dimensions of cross-sections along the yarn. This work employed this last-mentioned technique. Although VRML lacks the necessary primitives for cross-sectional shapes, several yarn cross-sectional shapes were employed in this work, namely, circular, ellipse, racetrack and lenticular. Peirce’s model was based on the assumption that a circular cross-sectional yarn in consideration is incompressible. In the actual situation, yarns are deformed so that two diameters, which correspond to yarn width and yarn thickness in the structures, must be considered. The extent which a yarn will be deformed is the flattening coefficient which is always less than 1. Peirce’s model dealt with circular cross-sectional yarn for which warp and weft yarn cross-sectional coordinates were calculated based on the equation of a circle. In case of an ellipse cross-section, E(xi,yi) and PX(xi,yi) were acquired using the equation of an ellipse. A racetrack shape is the combination of a half-circle with a rectangle in the middle. Thus, the co-ordinates were calculated based on the equation of circle. In the case of a lenticular cross-section image which is formed by two incomplete circles, the equation of a circle was adopted with only selected co-ordinates. Mathematical modelling of yarn paths in a single layer woven fabric structure was developed based on Peirce’s model. The yarn path can be divided into three sections, namely, the overfloat section where the warp is floated over the weft, the underfloat section where the weft is floated over the warp and the linking section where the former two sections are joined. Warp yarn path coordinates, Eðxi ; yi Þ; were determined according to equations (5) and (6). Weft yarn path coordinates, Pðxi ; yi Þ, were calculated similarly:
 1 ð5Þ xi ¼ xi21 þ ppc 8 < D4 jwi; j ¼ 1 ð6Þ yi ¼ 2D : 4 jwi; j – 1 where: ppc

is the weft density (picks per centimetre).

D

is the summation of warp thickness (d1) and weft thickness (d2).

In the 3D simulation technique, an object can be created using the extrusion method. This method involves sweeping a 2D cross-section along a line, which is called a spline, resulting in 3D images. In this work, when a yarn path was used as a spline, an image of this yarn was generated by sweeping the specified yarn cross-section along the spline. The integration of these yarns at appropriate positions in a 3D space gave the simulated 3D image of the structure. Programming implementation According to the techniques and models explained earlier, a Windows-based software system was developed using Visual Basic programming. The interface of the program is shown in Figure 4. When a scanned fabric image was supplied, image processing and

Woven fabric structures

85

Figure 4. The program interface

analysis were then performed. The obtained data were further analyzed in order to identify a weave pattern as well as warp and weft densities. The acquired information was then combined with other default parameters in order to re-visualize the fabric structures. The yarn cross-sectional image was then swept through the yarn path images using VRML extrusion node. The positions of these yarn cross-sectional images in a 3D space were also automatically calculated. The VRML file was then generated accordingly and the 3D images were displayed as shown in Figure 5. It was possible to adjust some parameters, e.g. warp and weft densities, yarn colour and yarn cross-sectional shape. These altered 3D images were visualized and this feature is useful for product development. The experimental results indicated that the program was able to recognize most of plain fabrics correctly. However, in the case of twill and satin fabrics, the correct scanning was around 90 percent since, in some cases, yarns in fabrics tended to offset from the straight and perpendicular lines. The colour information was discarded during the image-processing step. In addition, coloured patterns affect the accuracy of the weave identification. 3D images of woven structures were correctly simulated according to the data obtained from the weave-recognition step. The advantage of displaying 3D fabric images using VRML is that the fabric is able to be studied at a distance or in much closer detail. Thus, internal geometry of yarns in the structures can be visually studied. Although, in this research, mechanical deformation data were not taken into account, with the proposed modelling technique, if the yarn mechanical information is available, it is possible to visualize corresponding fabric structures.

IJCST 22,2/3

86

Figure 5. 3D re-visualization of a twill weave

Conclusions A novel computer program has been developed in order to perform two tasks simultaneously: recognition and re-visualization of woven structures. An image processing and image analysis technique, as well as recognition algorithm were elaborated in order to identify woven fabric structures. Woven fabric images in digital format were obtained by an ordinary desktop flatbed scanner. The fabric images were processed into greyscale and then equalized. Autocorrelation was performed at every pixel in the equalized images, and autocorrelation curves were then plotted. The distances between peaks of autocorrelation curves as well as the peak width were used to identify the weave pattern as well as warp and weft densities. It was found that the program can identify approximately 90 percent of fabrics, in particular the plain, twill and satin fabrics without coloured patterns. Subsequently, 3D simulation of the woven fabric structure was done by generating each warp and weft yarn images and assembling them into fabric images. Various yarn cross-section shapes, namely, circular, racetrack and lenticular, were employed. Yarn paths were computed according to several parameters based on Peirce’s geometrical model. The yarn cross-sectional co-ordinates were then swept through the yarn path co-ordinates using VRML extrusion node resulting in a 3D image of the woven fabric structure. References Ames, A.L., Nadeau, D.R. and Moreland, J.L. (1996), VRML 2.0 Source Book, 2nd ed., Wiley, Toronto. Chen, X. and Potiyaraj, P. (1999), “CAD/CAM for orthogonal and angle-interlock woven structures for industrial applications”, Textile Research Journal, Vol. 69 No. 9, pp. 648-55.

Chen, X., Knox, R.T., McKenna, D.F. and Mather, R.R. (1992), “A parametric modelling and simulation system for multi-layer fabric reinforcements”, Proceedings to the 5th International Conference on Fibre Reinforced Composites, Newcastle Upon Tyne, March 24-26, The Plastic and Rubber Institute, London, pp. 26/1-26/11. Chen, X., Knox, R.T., McKenna, D.F. and Mather, R.R. (1993a), “Composite reinforcements: de-idialised solid modelling and computer integrated manufacturing”, Proceedings of the Conference on Managing Integrated Manufacturing: Organization Strategy & Technology, Keele, Staffordshire, Keele University, Keele, September 22-24, Vol. 2, pp. 365-77. Chen, X., Knox, R.T., McKenna, D.F. and Mather, R.R. (1993b), “Solid modelling and integrated manufacturing of textile interlinking structures”, Proceedings of International Conference: Design to Manufacture in Modern Industry, Bled, Slovenia, June 7-9, 1993, Part 2, pp. 682-8. Hearle, J.W.S. (1994), “Textile for composites”, Textile Horizons, Vol. 14 No. 6, pp. 12-15. Huang, C.C., Lui, S.C. and Yu, W.H. (2000), “Woven fabric analysis by image processing Part I: identification of weave patterns”, Textile Research Journal, Vol. 70 No. 6, pp. 481-5. Kang, T.J., Kim, C.H. and Oh, K.W. (1999), “Automatic recognition of fabric weave patterns by digital image analysis”, Textile Research Journal, Vol. 69 No. 2, pp. 77-83. Keefe, M. (1994a), “Solid modelling applied to fibrous assemblies – Part I: twisted yarns”, Journal of Textile Institute, Vol. 85 No. 3, pp. 338-49. Keefe, M. (1994b), “Solid modelling applied to fibrous assemblies – Part II: woven fabric”, Journal of Textile Institute, Vol. 85 No. 3, pp. 350-8. Keefe, M., Edwards, D.C. and Yang, J. (1992), “Solid modelling of yarn and fiber assemblies”, Journal of Textile Institute, Vol. 83 No. 2, pp. 185-96. Kinoshita, M., Hashimoto, Y., Akiyama, R. and Uchiyama, S. (1989), “Determination of weave type in woven fabric by digital image processing”, Journal of the Textile Machinery Society of Japan, Vol. 35 No. 2, pp. 1-4. Lin, H.Y. (1996), “The simulation of woven fabric structure by 3D computer graphics”, PhD thesis, University of Manchester Institute of Science and Technology, Manchester. Lin, H.Y. and Newton, A. (1999), “Computer representation of woven fabrics by using B-splines”, Journal of Textile Institute, Vol. 90 No. 1, Part 1, pp. 59-72. Lomov, S.V., Mikolanda, T., Kosek, M. and Verpoest, I. (2007), “Model of internal geometry of textile fabrics: data structure and virtual reality implementation”, Journal of the Textile Institute, Vol. 98 No. 1, pp. 1-13. Ohta, K., Sakaue, K. and Tamura, H. (1986), “Pattern recognition of fabrics surfaces”, Journal of the Textile Machinery Society of Japan, Vol. 32 No. 1, pp. 7-10. Peirce, F.T. (1937), “The geometry of cloth structure”, Journal of Textile Institute Transaction, p. T45. Sakaguchi, A., Kim, H., Matsumoto, Y. and Toriumi, K. (2000), “Woven fabric quality evaluation using image analysis”, Textile Research Journal, Vol. 70 No. 11, pp. 950-6. Xu, Y.H. (1992), “Computer representation of woven fabric structure”, MSc thesis, University of Manchester Institute of Science and Technology, Manchester. Corresponding author Pranut Potiyaraj can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

Woven fabric structures

87

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

IJCST 22,2/3

Fit analysis using live and 3D scan models Elizabeth Bye University of Minnesota, St Paul, Minnesota, USA, and

88

Ellen McKinney Art Institute of Dallas, Plano, Texas, USA

Received 19 July 2007 Accepted 4 August 2009

Abstract Purpose – The purpose of this paper is to develop a “good fit” for garments for customer satisfaction, comfort, and functionality as well as a manufacturer’s success and reputation. Design/methodology/approach – This paper reviews and evaluates garments on a live fit model and makes recommendations for the acceptance or modification of the garment for production. As more manufacturing, product development, and designing responsibilities continue to take place globally, alternatives to the traditional fit analysis are under consideration. Findings – Fit analysis using live and three-dimensional scan models as an alternative to the traditional fit analysis are under consideration. Originality/value – This paper evaluates garments on a live fit model and makes recommendations for the acceptance or modification of the garment for production. Keywords Garment industry, Image scanners, Clothing, Design, Modelling Paper type Research paper

International Journal of Clothing Science and Technology Vol. 22 No. 2/3, 2010 pp. 88-100 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011018586

Review of literature Faster, secure communication technology, and options for imaging support alternatives to traditional fit analysis. The three-dimensional (3D) scanner captures an accurate representation of the body/garment relationship that minimizes visual distractions including color and texture (Ashdown et al., 2004). This interactive image of the scanned model can be rotated or enlarged and thus results in visual fit information that provides more critical detail than photographs (Douty, 1968) or videotapes (Kohn and Ashdown, 1998). Fit researchers have used scanned images and expert judges to evaluate the fit of pants (Ashdown et al., 2004) and cooling vests (Nam et al., 2005). Using one pose, a three-point scale was used to analyze fit at 13 body points in the pant fit analysis. In the vest study, a five-point scale was used to evaluate fit at 36 body points from three different poses. Ashdown et al. (2004) concluded that 3D scan models have the potential to substitute for live fit models when: . recording one single instance of fit; . creating a database of various fit models in a single size; . recording multiple poses; and . conducting a virtual fit analysis at any time or location. Nam et al. (2005) found using 3D scan images for fit analysis convenient and accurate because there were no constraints due to time, model availability, or fatigue. Specific issues with fit that were influenced by the thickness and irregularity of the surface

were difficult to evaluate. Broader issues with expert training and developing a reliable instrument remain. Kohn and Ashdown (1998) found a positive correlation between fit analysis using a video image and traditional fit analysis using a live model. Both models resulted in a reliable analysis of the garment/body relationship. Nam et al. (2005) found that some criteria were difficult for judges to evaluate from a 3D scan. Ashdown et al. (2004) noted that some dimensions of a live fit analysis cannot be addressed from a 3D scan. The goal of this study was to compare use of a live model versus a 3D scan model on judges’ ability to evaluate fit criteria and the reliability of fit analysis scores. Research design The results of using a live model were compared to the results of using a 3D scan of the same model during a fit analysis. An expert panel of six judges completed both live and 3D scan fit analyses of a dress and a pant slopers on the model. These garments were selected for the fit analysis because they are the foundations for many clothing styles. A total of 17 criteria for pant fit and 24 criteria for dress fit were rated on a five-point scale. Judges’ ability to assess a fit score for each garment and the fit score on each criterion were compared between the live and 3D scan models. Scores from each criterion for both dress and pant were compared to evaluate differences related to individual garments or interaction between a garment and the model variation. Judges were randomly assigned, so differences in scoring were evenly distributed between judges. Therefore, no analysis for observer effect was planned. Research hypotheses This study was designed to compare the results between live and 3D scan fit analysis. Null hypotheses were formulated to test ability to score and reliability of the scores. Criterion were tested individually and results reported by group. Ability to score To compare ability to score between the two model types, the following hypothesis was postulated: H1. There is no difference in ability to assess fit analysis scores between using a live model and 3D scan model; H0. mlive ¼ m3D-scan. Each garment was constructed and modeled by a different individual. To look for scoring differences or interactions between individual garments, two additional hypotheses were developed: H1a.

There is no difference in ability to assess fit analysis scores between the 19 garments that were evaluated; H0. mGarment1 ¼ mGarment2 · · · ¼ mGarment19 .

H1b.

H0. There is no interaction between the fit analysis model type and the garment evaluated.

Reliability of scores If a live fit analysis and a 3D scan fit analysis are equivalent, each method should result in the same scores for each criterion. To test this, the following hypothesis was postulated:

Fit analysis using live and 3D scan models 89

IJCST 22,2/3

H2. There is no difference in fit analysis scores for each criterion between using a live model and 3D scan model; H0. mlive ¼ m3D-scan. Fit analysis scores for each criterion were also compared among individual garments and for interaction of garments with model type: H2a.

There is no difference in fit analysis scores between the 19 garments that were evaluated; H0. mGarment1 ¼ mGarment2 · · · ¼ mGarment19 .

H2b.

H0. There is no interaction between model type and garment evaluated.

90

Data-collection procedures Internal Review Board approval was received before data-collection procedures were initiated. Garments A total of 19 students of a sophomore level clothing design class were paired with a partner to develop a custom fit dress sloper and pant sloper. Basic dress and pant patterns were drafted according to Armstrong (2006). The dress sloper (Figure 1) was composed of a basic bodice with front and back waist darts, bust darts, back shoulder darts, one-dart sleeves, and a skirt with two front and two back darts. A center-back zipper opening was used. The pant sloper (Figure 2) was a trouser style with two front darts, two back darts, a center back zipper, and waistband. All students used the same muslin to cut and sew the test garments.

Figure 1. Dress sloper

Fit analysis using live and 3D scan models 91

Figure 2. Pant sloper

Fit analysis Expert judges participated in both a live fit analysis and a 3D scan fit analysis. In this incomplete block design, four (of six possible) randomly selected judges analyzed the fit of each garment. Two live and two 3D scan analyses were completed for each of 19 dress and 19 pant slopers, resulting in four analyses of each garment. A total of 152 garment fit analyses were completed. For the live analysis, each student modeled her garments while two judges evaluated the fit. For the 3D scan analysis, each model was scanned in her garments (Figures 3 and 4) with a VITUS/Smart 3D Body Scanner from Human Solutions. Scans were saved as rotating movie files. Judges could play, pause, and replay the video as needed during the 3D scan fit analysis.

IJCST 22,2/3

92

Figure 3. Scan of a model in dress sloper

Figure 4. Scan of a model in pant sloper

A unique fit analysis sheet was developed for the dress and pant slopers based on published fit analysis instruments. Criteria groups for both the dress and pant slopers were: . Overall, alignment. “Is the sloper correctly aligned on the body?” . Dart placement. “Are darts pointing in the right direction and ending in the correct place?” . Looseness/tightness (L/T). “Check the following areas for looseness (gapping or bagginess) or tightness (strain). Comment on needed corrections”. Under each criterion group, individual criteria were listed for analysis. See Tables I and II for individual criteria by group. The following scale, with 1 indicating the lowest score and 5 indicating the highest score was used:

Model type Criteria OA Center front Center back Left-side seam Right-side seam Waist Shoulder seam Sleeve grain Hem Dart placement Front bust darts Front skirt waist darts Back shoulder darts Back bodice waist darts Back skirt waist darts Sleeve dart L/T Front neckline Back neckline Bust Bodice back Front waist Back waist Front hip Back hip Armscye Sleeve

Live

Scan

P

38 36 38 38 38 38 38 35

30 37 14 13 23 13 22 38

0.007 0.567 0.000 0.000 0.000 0.000 0.000 0.091

38 38 38 38 37 38

14 17 5 15 16 5

0.000 0.000 0.000 0.000 0.000 0.000

38 37 38 35 38 37 38 38 38 38

38 33 38 37 37 36 37 38 26 36

N/A 0.053 N/A 0.324 0.324 0.567 0.324 N/A 0.000 0.165

Note: Figures in italics indicate significance at the *0.05 level

(1) (2) (3) (4) (5)

unacceptable fit; poor fit; acceptable fit; good fit; and excellent fit.

Space for comments was provided for each criterion. Data analysis and results Data preparation Fit analysis sheets were reviewed for completeness. Responses with no score or with a comment such as, “unable to see details,” “can’t see,” or “can’t tell” were re-coded to indicate missing data. Coding non-response was important to account for the missing data when conducting a Type II comparison of means. Score data for each criterion was entered into SPSS 13.0 for Windows. A second variable was created for each criterion to indicate ability to score.

Fit analysis using live and 3D scan models 93

Table I. Difference in number of dress slopers scored without problem

IJCST 22,2/3

94

Table II. Difference in number of pant slopers scored without problem

Model type Criteria OA Inseams Left-side seam Right-side seam Front crotch seam Back crotch seam Dart placement Front darts Back darts L/T Waistband Stomach Front crotch Front full hip Front thigh Back crotch Back upper hip Back full hip Back thigh

Live

Scan

P

38 37 36 38 38

16 25 28 28 27

0.000 0.000 0.006 0.001 0.001

34 36

18 22

0.000 0.000

37 37 38 38 38 38 38 38 38

36 37 38 38 38 38 38 38 38

0.567 N/A N/A N/A N/A N/A N/A N/A N/A

Note: Figures in italics indicate significance at the *0.05 level

H1: ability to score between model types Descriptive statistics were run to analyze the number of criteria scored for each model type. A two-way ANOVA was conducted for each criterion to test H1: “There is no difference in ability to assess fit analysis scores between using a live model and 3D scan model”. Dress slopers. Significant differences in ability to score between model types were seen for 54 percent of criteria (Table I). For overall alignment (OA) criteria, significant differences were seen in 75 percent of criteria, except for center back and hem. The null hypothesis was rejected in favor of the alternative: mlive – m3D-scan. For dart placement criteria, 100 percent of criteria had significant differences; therefore, the null hypothesis was rejected. For dress L/T criteria, there was a significant difference for 10 percent of the criteria: the armscye. The null hypothesis was accepted for dress L/T criteria. Pant slopers. There were significant differences in ability to score by model type for 44 percent of pant sloper criteria. Significant differences were seen with 100 percent of the OA and 100 percent of the dart placement criteria (Table II). For these groups, the null hypothesis was rejected in favor of the alternative hypothesis: mlive – m3D-scan. For the L/T group, no significant differences were seen. H0: mlive ¼ m3D-scan was accepted. H1a: difference in ability to score between garments A two-way ANOVA was conducted to test H1a: “There is no difference in ability to assess fit analysis scores between the 19 garments that were evaluated.” There was no significant difference for any criteria in ability to assess fit analysis scores between the 19 garments for dresses or pants. Therefore, H0: mGarment1 ¼ mGarment2 · · · ¼ mGarment19 was accepted.

H1b: interaction between model types and garment A two-way ANOVA was conducted to test H1b: “There is no interaction between fit analysis model type and garment evaluated.” There was no significant difference in ability to score for 96 percent of dress criteria, except L/T of back neckline ( p ¼ 0.049) and 100 percent of pants criteria caused by interaction between model type and garment. Therefore, the H0: “There is no interaction between fit analysis model and garment evaluated” was accepted for both dresses and pants.

Fit analysis using live and 3D scan models 95

H2: difference in scores between model types Mean scores were calculated for each criterion by model type. A two-way ANOVA was conducted to test H2: “There is no difference in fit analysis scores for each criterion between using a live model and 3D scan model.” Mean scores for each criterion were compared for significant differences at the 0.05 level. Dress slopers. There was a significant difference between models for 29 percent of criteria (Table III). Significantly different scores were seen at hem alignment, back bodice waist dart and skirt dart placement, and back neckline, bust, front hip, and back hip L/T criteria. The null hypothesis was accepted for 71 percent of dress criteria. Model type Criteria OA Center front Center back Left-side seam Right-side seam Waist Shoulder seam Sleeve grain Hem Dart placement Front bust darts Front skirt waist darts Back shoulder darts Back bodice waist darts Back skirt waist darts Sleeve dart L/T Front neckline Back neckline Bust Bodice back Front waist Back waist Front hip Back hip Armscye Sleeve

Live

Scan

P

3.60 3.33 2.80 3.00 2.87 3.03 3.40 3.30

4.50 4.50 4.00 4.00 4.00 4.00 4.00 2.5

0.331 0.513 0.449 0.780 0.178 0.835 0.326 0.000

2.77 3.40 3.37 3.07 3.33 2.90

4.00 3.50 4.00 4.00 3.50 4.00

0.053 0.057 0.651 0.029 0.025 0.274

2.67 3.03 2.70 2.77 2.97 2.90 3.30 3.20 2.73 2.70

2.50 3.50 3.00 4.00 4.00 4.00 2.50 2.00 4.00 4.00

0.058 0.029 0.024 0.331 0.289 0.537 0.002 0.001 0.891 0.104

Note: Figures in italics indicate significance at the *0.05 level

Table III. Difference in mean score of dress sloper criteria

IJCST 22,2/3

Pant slopers. Significant differences in fit analysis scores between using a live model and 3D scan model were seen in 44 percent of criteria. The H0: mlive ¼ m3D-scan was accepted for 80 percent of OA criteria and 100 percent of dart placement criteria. A significant difference was seen in 67 percent of the L/T criteria between live model and 3D scan model (stomach, front crotch, front full hip, front thigh, back full hip, and back thigh). The null hypothesis was rejected for L/T criteria (Table IV).

96 H2a: difference in scores between garments Means were calculated for each criterion by model type. To test H2a: “There is no difference in fit analysis scores between the 19 garments that were evaluated”; a two-way ANOVA was conducted to compare means at the 0.05 level. Dress slopers. No significant differences in scores were seen for 58 percent of dress criteria. Significant score differences were seen between garments for 38 percent of OA criteria (right-side seam p ¼ 0.012, shoulder seam p ¼ 0.005, and hem p ¼ 0.000); 33 percent of dart placement criteria (back bodice waist darts p ¼ 0.013 and sleeve dart p ¼ 0.000); and 50 percent of L/T criteria (back neckline p ¼ 0.014, bust p ¼ 0.022, bodice back p ¼ 0.017, back waist p ¼ 0.024, and sleeve p ¼ 0.002). The null hypothesis was accepted. Pant slopers. No significant differences were seen for 69 percent of criteria; therefore, the H0: mGarment1 ¼ mGarment2· · · ¼ mGarment19 was accepted. Significant differences were seen for 20 percent of OA criteria (front crotch seam alignment p ¼ 0.005) and 44 percent of L/T criteria (front crotch p ¼ 0.015, front full hip p ¼ 0.001, back crotch p ¼ 0.010, and back full hip p ¼ 0.017).

Model type Criteria

Table IV. Difference in mean score of pant sloper criteria

OA Inseams Left-side seam Right-side seam Front crotch seam Back crotch seam Dart placement Front darts Back darts L/T Waistband Stomach Front crotch Front full hip Front thigh Back crotch Back upper hip Back full hip Back thigh

Live

Scan

P

3.57 3.13 3.09 3.48 3.04

3.07 3.29 2.86 3.36 3.36

0.069 0.224 0.068 0.042 0.806

3.52 3.09

3.29 3.21

0.148 0.582

3.61 3.35 3.35 3.35 3.61 2.96 2.57 3.22 3.35

3.57 3.07 2.71 2.57 3.21 2.86 2.64 2.57 2.5

0.810 0.015 0.000 0.000 0.001 0.063 0.517 0.001 0.000

Note: Figures in italics indicate significance at the *0.05 level

H2b: interaction between model type and garment Means were calculated for each criterion for each garment by model type. To test H2b: “There is no interaction between model type and garment evaluated”; a two-way ANOVA was conducted to compare means at the 0.05 level. Dress slopers. No significant differences were seen for 96 percent of dress sloper criteria, The null hypothesis was accepted. Interaction between fit analysis model and garment evaluated was seen at hem alignment ( p ¼ 0.011). Pant slopers. No significant differences were seen for 81 percent of pant sloper criteria. The null hypothesis was accepted. Differences in scores were seen by interaction between garment and model type for front crotch seam alignment ( p ¼ 0.046); and front crotch ( p ¼ 0.026) and back crotch ( p ¼ 0.035) L/T.

Fit analysis using live and 3D scan models 97

Summary of data analysis Results of data analysis for all hypotheses are summarized in Tables V and VI. Discussion Ability to assess fit with live or 3D scan model There were missing data on many garment fit analysis sheets completed with 3D scan models. Problems were seen mainly in the OA and dart placement criteria groups. For dress slopers, 98 percent OA and 100 percent dart placement were scored with live models and only 63 percent OA and 32 percent dart placement were scored with 3D scan models. For pant slopers, results were similar, with 98 percent of OA and 92 percent of dart placement criteria scored with live models, but only 65 and 54 percent, respectively,

Criteria group All criteria OA Dart placement L/T

H1. Difference between live and scan model (%) Dresses Pants 54 75 100 10

44 100 100 0

H1a. Difference between garments (%) Dresses Pants 0 0 0 0

0 0 0 0

H1b. Interaction between model and garment (%) Dresses Pants 4 0 0 11

0 0 0 0

Note: Percentage indicates the percentage of fit criteria in that criteria group with significant differences at the *0.05 level

Criteria group All criteria OA Dart placement L/T

H2. Difference between live and scan model (%) Dresses Pants 29 13 33 40

44 20 0 67

H2a. Difference between garments (%) Dresses Pants 42 38 33 50

31 20 0 44

Table V. Summary of significant differences in ability to assess fit analysis scores

H2b. Interaction between model and garment (%) Dresses Pants 4 13 0 0

19 20 0 23

Note: Percentage indicates the percentage of fit criteria in that criteria group with significant differences at the *0.05 level

Table VI. Summary of significant differences in fit analysis scores

IJCST 22,2/3

with 3D scan models. Significant differences were seen in ability to score between models on six out of eight OA criteria (75 percent), except center back or hem. These criteria may be more visible on a scan because a ridge is caused by the zipper at center back and the hem is a strong edge. Scoring of criteria that required identification of seam lines and body landmarks was more difficult on a scan. These results are in agreement with the Ashdown et al. (2004) findings that seam lines are difficult to identify in 3D scans.

98 Differences in fit score evaluated with live or 3D scan model Overall, differences in fit scores were not significant between live and scan models. On the dress, hem alignment had a significantly different score and the mean live score was higher. This may be because a tilted hemline is more easily observed as the 3D scan is rotated to view all sides, resulting in a lower fit score. With the pants, front crotch seam alignment was significantly different in mean score between model type, between garments, and garment/model interaction. Significant differences were seen in the mean criteria score, between mean garment scores, and by garment model interaction in the L/T thigh, crotch, and full hip criteria (Table VII). Two out of three pant sloper L/T scores were higher with the live model than with the 3D scan model. This could be because pulls and wrinkles are more noticeable in a 3D scan than with a live model, resulting in a lower fit score. With a live model, surface color can be used positively to create visual illusions of a better fit, while the 3D scan image emphasizes the physical structure of the fit. These results indicate a need for awareness that fit scores may be lower when using 3D scan. Interactions between ability to score and fit score evaluated In the OA and dart placement criteria groups, where ability to evaluate with 3D scan models was poor, the scores were significantly different between model types. In the L/T criteria group where ability to evaluate with 3D scan models was good, significant differences were seen in scores between model types on 67 percent of pant sloper criteria and 40 percent of dress sloper criteria. The judges felt they could evaluate looseness and tightness on a scan, but actually scored many criteria differently. These results can guide training for evaluation of 3D scans. All judges had experience with live fit analysis; however, using the 3D scan model was a new or unfamiliar experience. Judges approached both live and scan models positively and technical help was available during the scan evaluation. There was no time limit to judging with either method.

Table VII. P-values for pant slopers OA and L/T criteria with significant differences

Full hip Front Back Crotch Front Back Thigh Front Back

Higher live score OA L/T

Differences between garments OA L/T

0.000 0.001

0.001 0.017

0.042

0.000 0.001 0.000

0.005

0.015 0.010

Garment model interaction OA L/T

0.046

0.026 0.035

Conclusions and implications While 3D scans offer the convenience of evaluating garment fit from any time or location, there are some concerns about ability to score and reliability of scores for specific fit criteria. All seam line alignment scores were significantly lower with 3D scan models than live models. Judges had significantly lower ability to score dart placement with 3D scan models than with live models. When analyzing seam and dart alignment with body landmarks is an important part of the fit analysis, judges may have difficulty with assessment. Neck There were no differences in scoring front neck, but back neck scores with 3D scan models were higher, and thus unreliable. Shoulder Judging shoulder fit is challenging because shoulder seam alignment and shoulder dart had significant problems in ability to score with 3D scan models. Sleeve/armscye The sleeve and armscye criteria, including sleeve grain, sleeve dart placement, and armscye L/T, were significantly unable to be scored with 3D scan models. Only sleeve L/T was reliable between live models and with 3D scan models. Bust There is significantly poor ability to score front bust darts from a scan and significantly different L/T scores by scan. Waist L/T at waist for dress slopers and pant slopers were evaluated without significant difference between model types. Hip Evaluation of hip and thigh criteria was not reliable. All hip and thigh L/T scores were lower with 3D scan models, except for pant sloper back upper hip. Hems Hems were able to be scored with 3D scan models as well as with live models, but the scores with 3D scan models were significantly lower. Overall, results indicate a need for awareness that fit scores may be lower when using 3D scan and that training is needed for evaluation of 3D scans. For criteria where fit can be evaluated, it may be possible to train judges to score 3D scan models as reliably as live models. Future advancements in 3D scanners, such as color, may improve the ability to see seams and darts, and thereby improve their ability to be evaluated. With current technology, there are significant differences between fit analyses with live or 3D scan models that must be considered before implementing them in research or industry settings.

Fit analysis using live and 3D scan models 99

IJCST 22,2/3

100

References Armstrong, H.J. (2006), Patternmaking for Fashion Design, 4th ed., Pearson Prentice-Hall, Upper Saddle River, NJ. Ashdown, S.P., Loker, S., Schoenfelder, K. and Lyman-Clarke, L. (2004), “Using 3D scans for fit analysis”, Journal of Textile and Apparel, Technology, and Management, Vol. 4 No. 1, pp. 1-12. Douty, H.I. (1968), “‘Visual somatometry’ in health-related research”, Journal of the Alabama Academy of Science, Vol. 39, pp. 21-34. Kohn, I.L. and Ashdown, S.P. (1998), “Using video capture and image analysis to quantify apparel fit”, Textile Research Journal, Vol. 68 No. 1, pp. 17-26. Nam, J., Branson, D.H., Cao, H., Jin, B., Peksoz, S., Farr, C. and Ashdown, S. (2005), “Fit analysis of liquid cooled vest prototypes using 3D body scanning technology”, Journal of Textile and Apparel, Technology and Management, Vol. 4 No. 3, pp. 1-13. Further reading Bye, E. and LaBat, K. (2005), “An analysis of apparel industry fit sessions”, Journal of Textile and Apparel, Technology and Management, Vol. 4 No. 3, pp. 1-5. Schofield, N.A., Ashdown, S.P., Hethorn, J., LaBat, K. and Salusso, C.J. (2006), “Improving pant fit for women 55 and older through an exploration of two pant shapes”, Clothing & Textiles Research Journal, Vol. 24 No. 2, pp. 147-60. Corresponding author Elizabeth Bye can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

Automatic basic garment pattern generation using three-dimensional measurements Choong Hyo Kim, In Hwan Sul and Chang Kyu Park Department of Textile Engineering, Konkuk University, Seoul, South Korea, and

Automatic basic garment pattern generation 101 Received 28 May 2009 Accepted 4 October 2009

Sungmin Kim Faculty of Applied Chemical Engineering, Chonnam National University, Gwangju, South Korea Abstract Purpose – The purpose of this paper is to illustrate the generation of basic garment pattern using three-dimensional body measurement data. Design/methodology/approach – A pre-defined garment model is deformed using free-form deformation method and the model is flattened to generate flat patterns. Findings – The paper finds that individual basic garment patterns are automatically generated and verified to be well fit on human subjects. Research limitations/implications – The current approach is to focus on the generation of basic bodice patterns; however, other patterns can also be generated by this method by preparing more garment models. Practical implications – This method can reduce the time required to design basic patterns as well as enhance their fitness. Originality/value – The automatic generation of individually fitted garment pattern is one of the most important steps in future garment production process. Keywords Deformation, Garment industry, Textile technology, Modelling, Computer aided design Paper type Research paper

Introduction Recently, there have been many studies on the application of information technology to the production of customer-oriented fashion goods such as made-to-measure (MTM) garments. MTM garments are made based on three-dimensionally measured individual human body data through digital processes such as automatic pattern generation and virtual drape simulation. For automatic pattern generation, although a commercialized system has not yet been developed, there have been many studies of the development of garment patterns directly from three-dimensional (3D) body data. For example, optimum-fit patterns for each person were generated by drawing pattern outlines and darts directly on the surface of a parametrically deformable garment model (McCartney et al., 2000; Kim and Park, 2003). Most such 3D pattern generation methods were based on the data acquired by non-contact-type body scanners. However, it is difficult to flatten these garment models into patterns, so such methods are not suitable for practical application (Kang and Kim, 2000a, b; Kim and Park, 2003).

International Journal of Clothing Science and Technology Vol. 22 No. 2/3, 2010 pp. 101-113 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011018595

IJCST 22,2/3

102

In this study, a contact-type coordinate measurement system was used to reconstruct the body surface model with a topologically regular mesh structure. Parametric modeling was used to deform the body model into the target shape and size. For this, a free-form deformation (FFD) method was used instead of cross-section-based or morphing methods to deform the shape of the body model (Sederberg and Parry, 1986; Coquillart, 1998). The body model was projected into flat garment patterns using multiple darts, and the fit of the actual garments made from those patterns was verified on three subjects by self-sensory tests. Methodology 1. Preparation of body model 1.1 Contact-type coordinate measurement system. Generating a virtual 3D model using a non-contact measurement system includes the reconstruction of a surface model from so-called point cloud data. However, the resulting model tends to have a noisy surface and requires much post-processing to be converted into a smooth and topologically well-structured model, because it is composed of irregular triangular meshes. Therefore, non-contact-type measurement systems are not suitable for garment modeling, because smooth lines and regular mesh structure are very important when projecting the body model onto flat patterns (Gonzalez and Woods, 2003; Russ, 2007). In this study, we obtained a body model with a topologically regular surface mesh structure using a contact-type coordinate measurement system by measuring the absolute coordinate of each point on the lines drawn on a body model that are required for garment design. 1.2 Reconstruction of body model. Measured coordinate data must be converted into a parametrically deformable surface model to be used in garment pattern generation. The surface model consists of a number of triangles, and as the index of each point on a triangle has been determined previously, measured coordinate data can be converted into a 3D surface model with triangular meshes as shown in Figure 1. 2. Deformation of body model using a FFD method 2.1 Free-form deformation. FFD is a method to deform the shape of a 3D model that was developed by Sederberg and Parry (1986). This method has been widely used because the shape of a complex 3D model can be deformed easily using only a few control points, regardless of the modeling method used (Sederberg and Parry, 1986). 2.1.1 Formation of FFD lattice. An FFD lattice is a hexahedral set of control points used for the deformation of an enclosed 3D model. Control points in the lattice are usually located by B-spline interpolation as described below (Farin, 2002). An FFD lattice is defined in the Cartesian coordinate system. First, a hexahedral lattice is formed to enclose completely the 3D model being deformed. Then arbitrary numbers of control points are located along each side of the lattice at regular intervals. The distance between two adjacent control points can then be adjusted by non-uniform rational B-spline interpolation to map the proper knot vectors and control points. 2.1.2 Model deformation. Control points on the lattice are closely related to the points on the enclosed model, and the changes in their positions are directly reflected on the shape of the model. Continuity is guaranteed inside the lattice so that the deformation is smooth everywhere on the surface of the enclosed model. In addition, there are no restrictions on the movement of each lattice control point.

Automatic basic garment pattern generation 103

Figure 1. Reconstruction of 3D body model

2.2 Definition of FFD lattice on body model. 2.2.1 Formation of FFD lattice. One of the most frequently used body model deformation methods is the cross-section-based method, in which the body model is usually generated by connecting and skinning a series of cross-sectional shapes. Some representative cross-sections such as neck, bust, waist, and hip are selected, and their girths are adjusted to become those of the target body model. Girths of intermediate cross-sections between two representative cross-sections can be interpolated by a simple equation (Kang and Kim, 2000a, b). Although this method is very simple, the continuity of the model surface is not guaranteed, especially when the deformation becomes very large. In this study, the FFD method was used to deform the body model so that a continuous and differentiable surface model could be obtained by adjusting a few control points as shown in Figure 2. 2.2.2 Definition of FFD operators. Although the shape of an enclosed body model can be modified by manipulating the lattice control points, it is very difficult to adjust multiple points simultaneously to change the shape of the model along the three principal axes. Therefore, some operator functions must be defined to manipulate multiple control points. The body model is aligned as shown in Figure 2. For the definition of operators, control points with the same x coordinates are grouped as “sagittal” sections, and those with the same z coordinates are grouped as “frontal” sections, while control points with the same y coordinates are grouped as “layer” sections. Operator functions move or rotate multiple control points on specific sections according to the supplied parameter values, and the model deformation process can be managed in a simple and consistent way. The definitions of landmarks and measurement items required for the design of women’s bodice patterns are shown in Figure 3. As the landmarks were already drawn on the body model and measured by the coordinate measurement system, the geodesic distance between two landmarks can be calculated by considering the indices of

IJCST 22,2/3

104

y

x

Figure 2. Deformation of a 3D body model using the FFD method

z

3

1 2 4

h d e 5

6 a

b

c

f

7

8

Figure 3. Definition of body points and sizes for bodice formation

g

Body points for bodice deformation

Body sizes for bodice deformation

1. Cervicale 2. Anterior neck 3. Lateral neck 4. Acromion 5. Mesosternal 6. Nipple 7. Anterior waist 8. Posterior waist

a.Waist front length b. Waist backlength c. Bust point-bust point d. Neck base circumference e. Chest circumference f. Bust circumference g. Waist circumference h. Shoulder angle

two points on the body model. To change the size of the body model accurately, it is important to know the relationship between the parameters supplied to the operators and the actual changes in model geometry caused by the operators. Assuming that the changes in measurement items are independent of each other, the relationship between the input and output of an operator can be obtained by regression analysis. For example, a linear regression equation can be obtained by supplying a parameter value ranging from 250 to 50 mm to an operator and observing the resulting size change in the body model. Once the regression equation is established, the parameter value required to invoke a certain size change on the body model can be calculated by the inverse of that equation. According to the experiments made in this study, the actual changes made to the body model by each operator had good correspondence with the initially intended values, with R 2 values greater than 0.98. 3. Flat pattern generation 3.1 Definition of darts. Darts are necessary to form 3D garments from two-dimensional (2D) patterns and vice versa. Darts can have various widths, lengths, and shapes such as linear or curvilinear. There can be more than one dart in a pattern and usually the location of each dart is pre-determined according to the overall style of the garment. Potential darts and their positions for a bodice are shown in Figure 4. Darts on the body model can be defined as follows. First, a dart is marked on the actual body model using marking tape. Then the indices of the start and end points of the dart, and the end points of each segment intersecting with the dart line, are recorded. Finally, the shortest path for the dart can be obtained by finding and connecting all the “best” intersection points of each segment. The best intersection point of a segment can be located by finding the t-value that minimizes the total distance from the start point to the end point through the intersection point, where the t-value ranges from 0 to 1. Therefore, the shortest path for a dart can be determined by finding all the t-values, as shown in Figure 5. Once the dart path is determined, the triangular elements on the surface model must be restructured, as shown in Figure 5. If the incising line passes through one of the three points of a triangle, the triangle is divided into two triangles, while the triangle is divided into three triangles if the line intersects with two sides of the triangle as shown in Figure 5. 3.2 Projection of the body model. When projecting a 3D body model onto a 2D plane to form 2D garment patterns, some distortions are unavoidable, as shown in Figure 6. This distortion can be reduced by equalizing the lengths of each side on the triangular elements in 3D and 2D iteratively. A stopping criterion for this iterative process is defined in equation (1):

3D
N X 3
2D X
Lij 2 Lij
C Strain ¼ 100 £ ð1Þ
ð%Þ;


L2D ij i¼1 j¼1 where N is the number of triangles, and Lij is the 2D and 3D length of the j-th side of the i-th triangle. The equalization process is stopped when the CStrain value becomes smaller than a pre-defined threshold value. Finally, the 3D body model is projected into flat patterns. 3.3 Generation of pattern outlines. As the projected patterns consist of triangular elements, it is necessary to generate a closed outline for subsequent pattern manipulation in garment production processes. An outline can be determined by finding every border

Automatic basic garment pattern generation 105

IJCST 22,2/3

b

a

3

d

2

4

106

1

5 c

6 7 11

8 9

Front

Figure 4. Definitions of potential darts for a bodice

1. Center front neck point 2. Neck line 3. Shoulder line 4. Shoulder point 5. Arm hole 6. Under arm

10

Back 7. Low under arm 8. Waist 9. Center front waist 10. Center front line 11. Bust point

a. Shoulder line b. Neck line c. Waist d. Arm hole

edge that has only one neighboring triangle and connecting the end points of all those border edges in the correct order, as shown in Figure 7(b). When the included angle at a point on that outline differs from 1808 by more than a pre-defined threshold value, the point can be regarded as a “corner” point. The outline can then be approximated using connected multiple smooth B-spline curves segmented by those corner points, as shown in Figure 7(c) (Farin, 2002). Cutting lines can be generated by duplicating and locating each segment at a certain offset called the ease from the original outline as shown in Figure 7(d). Experimental 1. Measurement of body model 1.1 Preparation of body model. In this study, a bodice model was chosen for automatic pattern generation, because it has the most complex shape among garment models

n = Last node n–1

Automatic basic garment pattern generation

t = 0~1

n–2

107 t = 0~1 0 = First node

(a)

(b) Notes: (a) Shortest path search algorithm for dart generation; (b) generated dart

Figure 5. Schematic diagram of dart generation

Figure 6. Schematic diagram of 2D projection

including neck, armhole, bust, and waist lines. For this reason, the design of an accurate bodice pattern is the most difficult process in garment design, and therefore bodice modeling is suitable for the verification of the performance of a newly developed method.

IJCST 22,2/3

108

Boundary edge

Common edge

Figure 7. Schematic diagram of sewing line generation

(a)

(b)

(c)

(d)

Notes: (a) Path search algorithm; (b) extraction of closed paths; (c) B-spline approximation; (d) cutting line generation

1.2 Measurement device. A contact-type MicroScribe G2 desktop digitizing system from Immersion Corporation was used to measure the surface geometry of a bodice model, as shown in Figure 8. Rhinoceros 3.0 modeling software was used to convert the measured data into an ASCII-format data file. 1.3 Definition of surface mesh structure. A regular mesh structure was drawn on the model using triangles and quadrilaterals using 1.0 mm wide marking tape. The lengths of sides ranged from 1 to 3 cm according to the complexity of the local surface shape. The coordinates of each mesh point were measured and each point was numbered for use in the formation of the triangular mesh structure. The index assigned to each point was also used in the measurements as well as in the definition of darts.

Automatic basic garment pattern generation 109

Figure 8. Multiple joint-type 3D coordinate measurement system

2. Pattern generation and trial test 2.1 Preparation of subjects. In this study, three human subjects were chosen to verify the pattern generation process using a self-sensory test with real garments. The bust girths of subjects ranged from 82 to 85 cm and waist girths from 67 to 70 cm. The basic measurement results of subjects’ upper bodices are shown in Table I. 2.2 Deformation of body model and generation of patterns. A five-layer FFD lattice was defined around the reconstructed bodice model. The layers were numbered from 0 to 4 from the bottom. Layer 1 is located at the under-bust level, 2 at the bust, and 3 at the front neck. Body deformation rules were established and calibrated by FFD operators using the actual measurement data. The body model was deformed to fit each subject and projected into flat patterns using multiple combinations of potential darts. In this study, the combination of front shoulder line, front waist, back waist, and back armhole dart seemed to be the best one. Projected patterns were plotted and cut by Mimaki CG-100AP apparel cutting plotter with cutting and sewing lines. 2.3 Trial test. Plotted patterns were sewn into real garments using white cotton fabric with low extensibility. A zipper was added on the back for wearability and a constant ease was added for each pattern for minimum comfort of subjects. A private fitting room with a full-body mirror was prepared so subjects could assess the fit of the garments.

Items Neck girth Waist girth Bust girth Upper bust girth Back length Front length Note: Unit: mm

1

Subjects 2

3

380 700 845 850 390 340

370 670 820 820 385 330

370 680 835 840 395 345

Table I. Upper body sizes of subjects

IJCST 22,2/3

2.4 Self-sensory test for fitting evaluation. A self-sensory questionnaire test was performed with 12 questions on fitting evaluation. Subjects stood relaxed with the garment on and answered each question using a satisfaction index ranging from 1 to 5. The results of the self-sensory test were analyzed by the SPSS statistical analysis software package using t-tests and one-way ANOVA tests.

110

Results and discussion The garment generated from the initially measured model is shown in Figure 9(a). A garment generated for a human subjected is shown in Figure 9(b).

(a)

Figure 9. Try-on test

(b) Notes: (a) Dummy model; (b) human subject

The results of the self-sensory test are shown in Table II. As shown in the results, Subject 2 felt most uncomfortable with the location of the neck circumference and the overall appearance (Q12) for appearance evaluation. In the comfort evaluation, Subject 2 felt uncomfortable with the neck circumference (Q5), shoulder width (Q6), front armhole width (Q7), front center length (Q9), and waist girth (Q11). It is thought that this discomfort was caused by the inappropriate location of darts after the large deformation of the original model, as Subject 2 had the smallest bust and waist sizes among the subjects. However, all the subjects felt approximately the same degree of comfort with the location of the other girths (Q2, Q3, and Q4) and the comfort on the back (Q8, Q10). Considering the overall results, Subject 2 differed from Subjects 1 and 3 ( p , 0.001). Subjects 1 and 3 felt comfortable with the garment, but the discomfort of Subject 2 indicates that a comprehensive experiment on the definition of the body model deformation rule seems to be necessary for the method to be applicable to various subjects. The results of the virtual try-on are shown in Figure 10. The development of a drape simulator is another research topic in our laboratory, and it could become an efficient tool for garment design if the pattern generation system developed in this study were to be integrated into the drape simulation system.

Automatic basic garment pattern generation 111

Conclusion The 3D pattern generation method is an innovative method by which optimum-fitting patterns can be obtained easily without trial-and-error-based traditional grading methods. We developed an automatic pattern generation system. A body surface model was reconstructed from 3D human body data measured by a contact-type coordinate measurement system. With this method, a surface model with a regular mesh structure could be generated, and its topological information could be used for efficient flat-pattern development. An FFD-based method was used to change the size and shape of the body model to match the bodies of human subjects. Finally, the body model was projected into flat garment patterns incorporating multiple darts. The fit of

1 Self-sensory question 1. Location of neck circumference 2. Location of upper bust circumference 3. Location of bust circumference 4. Location of waist circumference 5. Neck circumference 6. Shoulder length 7. Front body width at armpit level 8. Back body width at armpit level 9. Front center line length 10. Back center line length 11. Waist circumference 12. Overall appearance Average

M b

5.0 4.7 4.7 5.0 5.0b 4.3b 4.0b 5.0 5.0b 4.3 4.7b 4.3b 4.7b

SD 0.00 0.58 0.58 0.00 0.00 0.58 0.00 0.00 0.00 0.58 0.58 0.58 0.09

Subjects 2 M SD a

3.6 3.0 3.3 4.0 3.3a 2.3a 2.3a 4.3 3.3a 4.0 4.0a 2.7a 3.4a

0.58 1.00 0.58 1.00 0.58 0.58 0.58 0.58 0.58 0.00 0.00 0.58 0.05

3 M a

4.0 4.0 4.3 4.7 4.3b 3.7b 2.7a 4.7 4.7b 4.3 5.0b 4.7b 4.3b

SD

p-value

0.00 0.00 0.58 0.58 0.58 0.58 0.58 0.58 0.58 0.58 0.00 0.58 0.29

0.007 * * 0.058 0.068 0.252 0.014 * 0.014 * 0.011 * * 0.296 0.011 * 0.630 0.027 * 0.011 * 0.000 * * *

Notes: Significant at: *p , 0.05, * *p , 0.01, and * * *p , 0.001, respectively; superscripts were applied where a significant difference was observed after SNK test (a , b)

Table II. The results of the self-sensory test

IJCST 22,2/3

112

Figure 10. Virtual try-on test using 3D drape simulator

garments made from those patterns was verified on three human subjects by self-sensory tests that showed a good correspondence. In this method, changes made on the garment model such as dart insertion are instantaneously reflected on the resulting patterns, so designers can make customer-oriented MTM garments easily. Further research will be made on the practical application of this method on various garments through the construction of a database for model deformation rules. References Coquillart, S. (1998), “Extended free-form deformation: a sculpturing tool for 3D geometric modeling”, ACM SIGGRAPH Computer Graphics, Vol. 24 No. 4, pp. 187-96. Farin, G. (2002), Curves and Surfaces for CAGD: A Practical Guide, 5th ed., Morgan Kaufmann, San Francisco, CA.

Gonzalez, R.C. and Woods, R.E. (2003), Digital Image Processing, 2nd ed., Prentice-Hall, New York, NY. Kang, T.J. and Kim, S. (2000a), “Development of three-dimensional apparel CAD system”, International Journal of Clothing Science & Technology, Vol. 12 No. 1, pp. 39-49. Kang, T.J. and Kim, S. (2000b), “Optimized garment pattern generation based on three-dimensional anthropometric measurement”, International Journal of Clothing Science & Technology, Vol. 12 No. 4, pp. 240-54. Kim, S. and Park, C.K. (2003), “Fast garment drape simulation using geometrically constrained particle system”, Fibers and Polymers, Vol. 4 No. 4, pp. 169-75. McCartney, J., Hinds, B.K., Seow, B.L. and Gong, D. (2000), “An energy based model for the flattening of woven fabrics”, Journal of Materials Processing and Technology, Vol. 107, pp. 312-18. Russ, J.C. (2007), The Image Processing Handbook, 5th ed., CRC Press, New York, NY. Sederberg, T.W. and Parry, S.R. (1986), “Free-form deformation of solid geometric models”, ACM SIGGRAPH Computer Graphics, Vol. 20 No. 4, pp. 151-60. Further reading Kim, S. and Kang, T.J. (2002), “Garment pattern generation from body scan data”, Computer-Aided Design, Vol. 35, pp. 611-8. Corresponding author Sungmin Kim can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

Automatic basic garment pattern generation 113

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

IJCST 22,2/3

Reconstruction of individualized dress forms using parameterized silhouettes

114

Xinrong Hu

Received 29 June 2009 Accepted 11 October 2009

Department of Computer Science, Wuhan University of Science and Engineering, Wuhan, China and School of Human Ecology, University of Texas at Austin, Austin, Texas, USA, and

Bugao Xu School of Human Ecology, University of Texas at Austin, Austin, Texas, USA Abstract Purpose – The purpose of this paper is to develop a fast parameterized modeling approach to generate individualized dress forms for realistic human bodies. Design/methodology/approach – An individualized dress form is created by deriving a new set of fitting functions from a number of key existing dressing parameters and pre-defined templates. The fitting functions only contain simple shapes of circular and/or elliptical arcs, which can be modified computationally based on a few personal dressing data. Findings – This paper reaffirms that individual body shape can be adequately described by a number of critical cross-section silhouettes, and a personalized dress form can be constructed based on key dressing parameters and templates. Originality/value – The fitting functions and relevant dressing data for specific cross-sectional silhouettes are determined, permitting a user to create personalized dress forms only by inputting a simple set of dressing parameters. Keywords Computer aided design, Fashion design, Modelling Paper type Research paper

International Journal of Clothing Science and Technology Vol. 22 No. 2/3, 2010 pp. 114-126 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011018603

1. Introduction Human body modeling has been one of the challenging tasks encountered by computer graphics researchers and fashion designers. Many parameterized modeling techniques have been applied for creating body models (Rose et al., 1998). In general, these techniques can be classified into two categories: one that models body surface from a sample of scanned data (Lewis et al., 2002; Blanz and Vetter, 1999; Kry et al., 2002; Sloan et al., 2001; Allen et al., 2002; DeCarlo et al., 1998; Dekker et al., 1999; Ju et al., 2000; Yu and Xu, 2008) and the second that models body surface from selected key feature parameters (Seo and Magnenat-Thalmann, 2003). Lewis et al. (2002) proposed “pose space deformation,” approaching the problem of geometric model deformation by using the scanned elastic surface of varying postures and blending them during the animation. Blanz and Vetter (1999) presented a “morphable face model” for manipulating an existing model according to changes in certain facial attributes. New faces were modeled by forming linear combinations of

the prototypes that were collected from 200 scanned face models. Manual assignment of attributes was used to define shape and texture vectors that, when added to or subtracted from a face, manipulated a specific attribute. Recently, Kry et al. (2002) proposed another extension modeling technique based on principal component analysis, allowing for optimal reduction of the data and thus expediting the modeling. Sloan et al. (2001) applied radial basis function for blending example facial models and the arm. Allen et al. (2002) presented another example-based method for creating realistic skeleton-driven deformation. DeCarlo et al. (1998) reduced the problem of generating simple face geometries into the problem of generating sets of anthropometrical measurements by using various modeling techniques. However, the modeling process needed minutes of calculation to produce a face model corresponding to the given measurement set. Dekker et al. (1999) used a series of anatomical assumptions in order to optimize, clean and segment data from a whole body range scanner to generate quad patch representations of human bodies and built applications for the clothing industry. Ju et al. (2000) introduced an approach to automatically segment the scan model to conform it to an animatable model. Yu and Xu (2008) presented an effective algorithm for reconstruction of the human body with the data from the two-view body scanner. Seo and Magnenat-Thalmann (2003) introduced a parameterized body modeling technique using key sizing parameters. They employed a two-phase algorithm to conduct global and local mapping, and used radial basis interpolation to form the body shape. Their proposed scheme took unorganized scanned data from realistic human body to generate appropriate shape and proportion of the body by forming deformation functions from the input parameters. A static body model is formed by altering the control vertices of a number of key template silhouettes with the user’s dressing parameters. The premise for this modeling concept is that the topology of the model is a known priori and shared by other resulting models. For dressing purposes, a strict fidelity model is not necessary for each user. An individualized dress form often suffices the need for fast visualization in virtual clothing. This simplification can tremendously reduce the needed measurements and make the modeling nearly real-time, which is critical for implementing online virtual try-on. 2. Parameterized silhouettes Since the head, hands, and feet of a person do not have direct effects on his/her dressing body appearance, most dress forms include only the torso and legs. The proposed modeling scheme for individualized dress forms will use regular dressing parameters, such as stature, shoulder width, bust girth, waist girth, hip girth, and the other correlative dressing sizes, to calculate the parameters necessary for constructing simplified silhouettes. To meet the requirements of the dressing, we can reconstruct a dress form by selecting some critical cross-sectional silhouettes as shown in Figure 1. The selection principle is that these silhouettes not only define the basic body shape needed for the dressing, but also can be reconstructed from the dressing parameters quickly. In order to achieve these objectives, we sampled some people to form the set of human body dressing parameters that can be used to model the template model shape. 2.1 Shoulder silhouette The critical parameters that determine the shoulder silhouette are shoulder breadth, depth and the height from heel to shoulder. It is also important to find the relationships

Reconstruction of individualized dress forms 115

IJCST 22,2/3

Shoulder Bust

116

Waist

Figure 1. Key cross-sectional silhouettes of a human body

Hip

Thigh

between these parameters and stature. The actual shoulder silhouettes for a male and a female are shown in Figure 2. These shoulder shapes may be represented with a simple shape shown in Figure 3. The simple shape consists of four connected geometrical arcs, two circular arcs and two elliptical arcs that are divided by the two blue lines. The key technique for re-constructing this shoulder shape is to solve for 20 selected vertices that determine two elliptical and two circular arcs. Seven vertices are extracted from each circular arc and three vertices from each elliptical arc. To solve for all these vertices, the vertices connecting a circle and an ellipse should be solved first, that is, to solve u1 as shown in Figure 3. The coordinates of a vertex on the ellipse arc can be represented: x ¼ C s*W s*cosðuÞ;

Figure 2. Actual shoulder shapes

y ¼ H s;

(a)

z ¼ C s*Ds*sinðuÞ þ Ds

(b)

Notes: (a) Male; (b) female

Ws

Ds

θ3

Figure 3. Fitted shoulder silhouette

θ2

θ1

ð1Þ

Here, u is inclined angle between the line that links the vertex on the ellipse arc and ellipse center and the horizontal line; Ws is shoulder breadth; Ds is shoulder depth; Hs is shoulder height from heel to shoulder; Cs is the weighted coefficient of the long axis and the short axis. The coordinates of vertex on circle arc can be represented: x < 0:5* ðW s 2 Ds*cosðuÞÞ;

y ¼ H s;

z < 0:5* Ds*sinðuÞ

ð2Þ

Reconstruction of individualized dress forms 117

Here, u is inclined angle between the line that links the vertex on the circle arc and circle center and the horizontal line. Then the shoulder shape silhouette is modeled by linking these vertices. 2.2 Bust silhouette Figure 4 shows two bust silhouettes. The critical parameters that determine the bust silhouette are bust girth and the height from heel to bust. For the dressing purpose, the bust silhouette can be modeled by the shape shown in Figure 4(c). Since the bust shapes of a male and a female are different, the depth of the simulating curve should be adjustable to reflect the difference. We propose to use four tangent ellipses to construct the desired simulating curve as shown in Figure 4(c). The four ellipses are labeled as A, B, C, and D, respectively. The medium bold curve in Figure 5 can be seen as the simulated curve of the bust silhouette, the shape of which can be manipulated as follows: . the depth of the front bust can be changed by adjusting the abscissa axes of ellipses A, B and D; . the smoothness of the back bust can be changed by adjusting the abscissa axes of ellipses A, B and C; . the bust width can be changed by adjusting the abscissa axes of ellipses A and B; and . the bust depth can be changed by adjusting the ordinate axes of ellipses A and B. It is important to locate the four tangent vertices generated by ellipses C or D and A or B. Let us assume that: . W and H represent the half abscissa axis and half ordinate axis of ellipses A or B, and they determine the bust width and depth;

(a)

(b)

Notes: (a) Male; (b) female; (c) simulated

(c)

Figure 4. Bust shapes

IJCST 22,2/3

C

CAL

AL

H

118

W

A

B

2R

D

DAL

Figure 5. Tangent ellipses and correlative parameters .

.

.

.

CW and CH represent the half abscissa axis and half ordinate axis of ellipse C, and they determine the width and depth of the back bust; DW and DH represent the half abscissa axis and half ordinate axis of ellipse D, and they determine the width and depth of the front bust; HYC is the distance between the original point and the center of ellipse C, and HYD is the distance between the original point and the center of ellipse D; and R is the half distance between the centers of ellipses A and B.

The formulas of ellipses C and B are: x2 ð y 2 HYCÞ2 ðx 2 RÞ2 y 2 þ ¼ 1:0; and þ 2 ¼ 1:0: CW2 CH2 W2 H From equations (3) and (4), we have:

›ðx 2 =CW2 þ ð y 2 HYCÞ2 =CH2 Þ ›ððx 2 RÞ2 =W 2 Þ þ y 2 =H 2 Þ ¼ : ›y ›y The coordinates of the tangent vertices between ellipses C and B are: 8 R > > ; if R ¼ W > > < 2:0 x¼ > ðR * CW2 2 W * CW * RÞ > > ; if R – W > : ðCW2 2 W 2 Þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi H * W 2 2 ðx 2 RÞ2 y¼ W

ð3Þ

ð4Þ

ð5Þ

We can also calculate: rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi x2 HYC ¼ y þ 1:0 2 * CH; CW2

ð6Þ

and two angles: 

CAL ¼ arctan j

y R2x

 j ;



DAL ¼ arctan j

y R2x

 j

ð7Þ

Thus, when HB (the height from the heel to the middle bust) is known, the tangent vertices between two of ellipses A, B, C or D can be calculated with these parameters as follows. The tangent vertices of ellipses A and C: x ¼ W · cosðCALÞ 2 R;

y ¼ H B;

z ¼ H · sinðCALÞ:

Reconstruction of individualized dress forms 119

ð8Þ

The tangent vertices of ellipses B and C: x ¼ 2W · cosðCALÞ þ R;

y ¼ H B;

z ¼ H · sinðCALÞ:

ð9Þ

z ¼ H · sinðDALÞ:

ð10Þ

The tangent vertices of ellipses A and D: x ¼ W · cosðDALÞ 2 R;

y ¼ H B;

The tangent vertices of ellipses A and C: x ¼ 2W · cosðDALÞ þ R;

y ¼ H B;

z ¼ H · sinðDALÞ:

ð11Þ

2.3 Waist silhouette The critical parameters that determine the waist shape are the waist girth and height from heel to waist. The actual waist shapes are shown as in Figure 6, and can be modeled with two merged super-ellipses (Xiao, 2006). The general formula of a super-ellipse is: xn yn þ n¼1 n W H

ð12Þ

Here, W and H are the half abscissa axis and the half ordinate axis of the ellipse, respectively. If a vertex on the super-ellipse is represented with polar coordinates, that is: x ¼ r ðaÞ cosðaÞ;

y ¼ r ðaÞ sinðaÞ

Then, the super-ellipse can be written as:     r ðaÞ cosðaÞ n r ðaÞ sinðaÞ n þ ¼ 1: W H

(a) Notes: (a) Male; (b) female

(b)

ð13Þ

ð14Þ

Figure 6. Actual waist shape

IJCST 22,2/3

120

Hence: rðaÞn ¼

1 : ðcosðaÞ=W Þn þ ðsinðaÞ=H Þn

ð15Þ

Figure 7 shown super-ellipses with different n-values. We choose the super-ellipse with n being three to represent the waist silhouette. These two super-ellipses used to simulate the waist silhouette can be represented as: ( x ¼ ðW þ c * sinðaÞÞ * cosðaÞ ; ð16Þ y ¼ ðH þ d * cosðaÞÞ * sinðaÞ W and H are the waist width and depth, and c and d are the coefficients that are used to adjust W and H. The waist girth can be computed with trapezoid element integral method. The vertices on the front super-ellipses are: x ¼ ðW 2 WR · sinðaÞÞ · cosðaÞ;

y ¼ H w;

z ¼ ðH þ FHR · cosðaÞÞ · sinðaÞ

ð17Þ

The vertices on the back super ellipses are: x ¼ ðW 2 WR† sinðaÞÞ · cosðaÞ;

y ¼ H w;

z ¼ ðH þ BHR · cosðaÞÞ · sinðaÞ ð18Þ

In equations (17) and (18), WR is the weighted coefficient of horizontal axis and Hw is the height from the heel to the waist. FHR is the weighted coefficient of vertical axis on the front super-ellipses and BHR is the weighted coefficient of vertical axis on the back super-ellipses. The bust shape silhouette can be modeled by linking these vertices. 2.4 Hip and thigh silhouette Figure 8 shown the silhouettes of two actual hip shapes. The critical parameters that determine the hip shape are the hip girth and height from heel to hip. 3 2 1

π 2

y H a

x (α) π

O

α

W x

Figure 7. Super-ellipses with different n

4 5 3π 2

The real hip shape can be modeled with the shape shown in Figure 9. Similar to waist silhouette model, the simulated hip silhouette is irregular, and it can also be modeled with two super-ellipses. The two super-ellipses are merged to construct the simulated hip silhouette curve. In fact, the hip girth size can be made up with a line and a curve as shown in Figure 9. The line size depends on the two topmost vertices. The coordinates of the two vertices can be determined from simple mathematical relationships. Let x denote the sine value of the included angle a1 formed by the topmost vertex and the center of the hip. Then: pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2W þ W 2 þ 8 · WR2 : ð19Þ sin a1 ¼ 4 · WR Similarly, y is related with the cosine value of the included angle a and: pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2H þ H 2 þ 8 · BHR2 : cos b1 ¼ 4 · BHR

Reconstruction of individualized dress forms 121

ð20Þ

The coordinates of a vertex of the simulated hip silhouette are as follows: x ¼ ðW þ WR · sinða1ÞÞ · cosða1Þ;

y ¼ ðH þ BHR · cosðb1ÞÞ · sinðb1Þ:

ð21Þ

In equation (21), W and H are the half horizontal and vertical axes of the super-ellipses, WR is the weighted coefficient of horizontal axis. In equations (22) and (23), FHR is the weighted coefficient of vertical axis on the front super-ellipses, and BHR is the weighted coefficient of vertical axis on the back super-ellipses.

(a)

Figure 8. Actual hip shape

(b)

Notes: (a) Male; (b) female β1

α1

α2 β2

Figure 9. Simulated hip silhouette

IJCST 22,2/3

The vertices on the front super-ellipses are: x ¼ ðW 2 WR · sinða1ÞÞ · cosða1Þ; y ¼ H H;

ð22Þ

z ¼ ðH þ FHR · cosðb1ÞÞ · sinðb1Þ

122

The vertices on the back super-ellipses are: x ¼ ðW 2 WR · sinða2ÞÞ · cosða2Þ; y ¼ H H;

ð23Þ

z ¼ ðH þ BHR · cosðb2ÞÞ · sinðb2Þ In equations (22) and (23), HH is the height from the heel to the hip. The bust shape silhouette can be modeled by linking these vertices and solving the hip girth. We can model the thigh root silhouettes (shown in Figure 10) with two super-ellipses in a similar way described above. 3. Human body shape modeling 3.1 Template model The template model of a human body is determined by the dressing parameters and is composed of triangular meshes, as shown in Figure 11. A coarse template model is made of 11 cross-section silhouettes and 16 vertices on each silhouette. In order to increase the detail of the surface fitting, each triangular patch can be subdivided into four triangles. Each newly generated vertex is mapped to the modeled surface and a displacement vector is recalculated. The refined template model improves the resolution with Loop (1987) subdivision. All cross sections are hierarchically linked so that the body shape can be preserved in any transformation (scaling, rotation, and translation). 3.2 Feature parameters and critical cross sections General template models can be generated using the standard male and female size parameters specified in the Chinese Body Size Criterion (General Administration of Quality Supervision , Inspection and Quarantine of China, 1989). Body configuration can be defined by the cross sections and the parameters that form the approximate silhouettes of these cross sections. By applying the user’s personal body parameters to the general template model, one can obtain a new model describing the personal body shape efficiently. The feature parameters to be used should be those familiar to the

Figure 10. The thigh root silhouette

(a) Notes: (a) Male; (b) female

(b)

Reconstruction of individualized dress forms 123

(a)

(b)

Notes: (a) Male; (b) female

user. In addition to gender and age range, we include the parameters, such as stature, shoulder width, shoulder depth, bust girth, waist girth, hip girth, thigh girth and the height at these landmarks, in the modeling. 3.3 Runtime results The Wolf dress forms of sizes eight and ten (Wolf Dress Forms, 2009) were scanned with our body scanner (Xu et al., 2002) and reconstructed with the methods presented above after their dressing parameters were supplied (Table I). A total of 13 regular dressing parameters were used to reconstruct these two standard mannequins. The experimental results comparing individualized body models with the actual dressing forms are shown in Figure 12. It can be easily seen that our system faithfully reproduces models that are consistent with the input parameters familiar to the user. 4. Summary In this paper, we presented a geometric solution for modeling dress forms based on key dressing parameters. The modeling utilized only a few cross-sectional silhouettes that are critical to defining individual body shape, and the pre-defined body templates. The silhouettes at important body landmarks were constructed by using simple shapes, such as circle, ellipse, and super-ellipse arcs, and the dressing parameters. The benefits of using this dress form modeling approach are that a realistic and individualized body model can be produced quickly based on straightforward measurements of the body and the pre-defined template model. In the near future, we plan to utilize more accurate dressing parameters obtained from the 3D body scanner, and to extend this framework to head, hands, and feet modeling with skin color and texture rendering.

Figure 11. Template models with different resolutions

37 38.5

8 10

Table I. Parameters of the dress forms (cm)

Shoulder width

13 13.5

Shoulder depth 135 137.5

Shoulder height 87 91

Bust girth 64 66.5

Waist girth 92 94

Hip girth 32.5 34.5

Thigh length 45 46.5

Calf length

124 126

Armpit height

118.5 121.5

Bust height

101.5 103.5

Waist height

82 85

Hip height

124

Size

75.5 76

Thigh height

IJCST 22,2/3

Reconstruction of individualized dress forms 125

(a)

(b)

(c)

(d)

Notes: (a) and (c) scanned; (b) and (d) simulated

References Allen, B., Curless, B. and Popovic, Z. (2002), “Articulated body deformation from range scan data”, Proceedings of the 2002 SIGGRAPH, Addison-Wesley, New York, NY, pp. 612-19. Blanz, B. and Vetter, T. (1999), “A morphable model for the synthesis of 3D faces”, Proceedings of the SIGGRAPH’99, Addison-Wesley, New York, NY, pp. 187-94. DeCarlo, D., Metaxas, D. and Stone, M. (1998), “An anthropometric face model using variational techniques”, Proceedings of the SIGGRAPH’98, Addison-Wesley, New York, NY, pp. 67-74. Dekker, L., Douros, I., Buxton, B.F. and Treleaven, P. (1999), “Building symbolic information for 3D human body modeling from range data”, Proceedings of the Second International Conference on 3-D Digital Imaging and Modeling, IEEE Computer Society, Washington, DC, pp. 388-97. General Administration of Quality Supervision, Inspection and Quarantine of China (1988), GB-10000-88 Human Dimensions of Chinese Adults, Chinese Standard Press, Beijing. Ju, X., Werghi, N. and Siebert, J.P. (2000), “Automatic segmentation of 3D human body scans”, Proceedings of the IASTED International Conference on Computer Graphics and Imaging 2000 (CGIM 2000), Las Vegas, NV, USA. Kry, P.G., James, D.L. and Pai, D.K. (2002), “EigenSkin: real time large deformation character skinning in graphics hardware”, ACM SIGGRAPH Symposium on Computer Animation, San Antonio, TX, pp. 153-9. Lewis, J.P., Cordner, M. and Fong, N. (2002), “Pose space deformations: a unified approach to shape interpolation and skeleton-driven deformation”, Proceedings of the 2000 SIGGRAPH, Addison-Wesley, New York, NY, pp. 165-72. Loop, C. (1987), “Smooth subdivision surface based on triangles”, Master’s thesis, Department of Mathematics, University of Utah, Salt Lake City, UT.

Figure 12. Dress forms of sizes eight and ten

IJCST 22,2/3

126

Rose, C., Cohen, M. and Bodenheimer, B. (1998), “Verbs and adverbs: motion interpolation using RBF”, IEEE Computer Graphics and Applications, Vol. 18 No. 5, pp. 32-40. Seo, H. and Magnenat-Thalmann, N. (2003), “An automatic modeling of human bodies from sizing parameters”, ACM SIGGRAPH Symposium on Interactive 3D Graphics, ACM, New York, NY, pp. 19-26. Sloan, P.P., Rose, C. and Cohen, M. (2001), “Shape by example”, ACM Symposium on Interactive 3D Graphics, Washington, DC, pp. 135-43. Wolf Dress Forms (2009), available at: www.wolfform.com, June. Xiao, L. (2006), “The Theory and Implement of Super-ellipse Curve Code System”, Economic Management Press, Beijing. Xu, B., Huang, Y. and Yu, W. (2002), “A 3D body scanning system for apparel mass customization”, Optical Engineering, Vol. 41 No. 7, pp. 1475-9. Yu, W. and Xu, B. (2008), “Surface reconstruction from two-view body scanner data”, Textile Research Journal, Vol. 5 No. 5, pp. 1-10. Corresponding author Xinrong Hu can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

Style previewing in 3D using name-based sewing rules

Style previewing in 3D

In Hwan Sul College of Engineering, i-Fashion Technology Center, Konkuk University, Seoul, South Korea

127 Received 18 February 2009 Accepted 28 May 2009

Abstract Purpose – The purpose of this paper is to combine patterns from different garment sets and preview garment styles in 3D apparel design by giving sewing names to patterns and sewing edges. Design/methodology/approach – A new rule for 3D garment sewing is made. Unlike conventional vertex number-based method, patterns and their edges are given specific names. If two edges have a same edge name, they make a sewing line. Thus, patterns from different garments can be combined and draped with this method. Numbers of boundary mesh nodes were controlled using B-Spline to combine sewing edges of different lengths. Findings – It is found that by only assigning names to patterns and sewing edges, garment style can be previewed by substituting patterns. Originality/value – Styles and details of garments can be previewed in 3D by mixing patterns of different garment sets like in 2D technical flat sketching. Even patterns with different edge lengths can be combined by controlling the pattern meshes using B-Spline. Keywords Customization, Computer aided design, Garment industry, Fashion design Paper type Research paper

1. Introduction The development of computer hardware and software enabled 3D apparel computeraided design (CAD) system to visualize virtual garment from 2D pattern data effectively. The same technology is used also for animations and special effects in movies. The 3D virtual garment can be transferred via internet and such virtualization technology can promote business-to-business or business-to-customer industry. The most eminent benefit from using 3D apparel CAD is that we can see the garment design without making it in reality. But such an advantage is limited to a single garment set. The 3D CAD operation is similar to real garment design/sewing/stitching so only one style of garment (including pattern grading) is generated each time. The conventional sewing operation generates pairs of mesh nodes to sew. The list of sewing node pairs is used as sewing constraints in cloth simulation, thus combining patterns into one in 3D during drape simulation. But such sewing information is specific to each garment. Thus, patterns from different garment sets cannot be mixed. It is because the sewing information is based on the boundary mesh node ID’s. Meanwhile, when the garment designer draws technical flats in 2D, the designer can select any kind of collars, cuffs, and arm styles from previously stored database and generate a new design. Until now there was no way to implement this feature in 3D garment CAD system. Therefore, we propose new sewing rules based on edge names for 3D apparel CAD. The idea is to give standard names to every edge and record the sewing information with respect to the edge names, not to the mesh node ID’s. The advantage is that once

International Journal of Clothing Science and Technology Vol. 22 No. 2/3, 2010 pp. 127-144 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011018612

IJCST 22,2/3

the garments are given names, their detail patterns can be replaced with patterns from other garments. That is, straight collar on a man’s shirts can be substituted by wing collar or mandarin collar virtually without doing the 3D CAD work again. We defined this feature as 3D style previewing in that the user can preview various combinations of garment patterns just as in 2D technical flats.

128

2. Previous work CAD is widely used for many engineering products. Small mechanical parts, cars, or buildings are modeled and simulated via computer-based tools such as AutoCADw, Abaqusw and so on. Such computer-based simulation reduces the cost for refining final product from initial design. For apparels, the computer-assisted design is only limited to 2D. Pattern shapes (Eckert and Helmut, 1999) or technical flats (Ji et al., 2002) can be drawn using splines efficiently. For 3D garment design, Luo and Yuen (2005) devised a method to modify 3D garment simulation result from 2D pattern design. They used particle-based governing equation so that the modified patterns can their equilibrium shape. They used a female one-piece and slacks for example and showed modification result of neck shape. But their work focused on modification of a single piece of pattern and not on the replacement of patterns or conjunction of multiple patterns. They team also proposed a method to construct 3D garment mesh from 2D sketches (Wang et al., 2003). The sketch-based drawing method seems to be a candidate for next-generation design method in virtual environment, but currently it is hard to depict details of patterns in 3D. So our approach is not to draw 3D details directly, but to utilize conventional 2D patterns. The apparel industry already has database of garment patterns with size and style variations. Our name-based sewing rule can be standard nomenclature for combining patterns from different garment sets. Any combination of garment patterns, such as arm pattern from men’s shirt and bodice pattern from female’s shirt, can be sewn and simulated in 3D. We describe the methodology in the next chapter. 3. Naming rules for arbitrary pattern conjunction 3.1 Pattern names and grouping Garment parts are generally composed of multiple patterns. We sorted them out to pattern groups. For typical basic garments, the patterns can be categorized into several groups. Figure 4 shows the four kinds of basic female shirt patterns used in this investigation. We divided each pattern set into seven groups, such as collar (designated as “COLLAR”), bodice front (“FRONT”), bodice back (“BACK”), right arm (“ARM1”), left arm (“ARM2”), right cuff (“CUFF1”), and left cuff (“CUFF2”). There are no limits for specifying the number of groups, but the categorization into seven groups seemed appropriate in this case. The patterns of the same shaded rectangular regions in Figure 7 belong to the same group. The pattern groups become the basic units for style previewing. We used simple strings for the names of patterns and edges. The name strings were composed of two or three words, separated by underscore character. Each word means group name, ID, and type, respectively. For example, pattern name “FRONT_1” means the first pattern of front bodice group and edge name “EAST_1_INV” means an edge located in the eastern side with inverse type. The type string is a description for optional details for edges. Patterns of a same group have name with same

suffix, i.e. front bodice patterns of Figure 6(b) has name strings of “FRONT_1”, “FRONT_2”, . . . , “FRONT_7”, while the ID’s do not have to be continuous integers. Once all the patterns are given name, the next job is to give names for each edge. We classified the edges into two categories, such as inter- and intra-group edges. 3.2 Inter-group sewing edges Inter-group edges were defined as edge pair whose edges are from different groups. Figure 1 shows an example. It is necessary that inter-group edges should have same name throughout all the garment sets, because inter-group edges actually combine patterns from different sets. For example, right arm pattern and bodice pattern should have edges with the same name, “ArmholeFR” as shown in Figure 1. In this way, if any other right arm pattern has an edge name “ArmholeFR”, such as patterns “ARM_1” in Figure 6(b)-(d), it can replace the default one in Figure 1.

Style previewing in 3D

129

3.3 Intra-group sewing edge names Intra-group edges are defined as edges sewn to each other inside a same group. Intra-group names are not critical to style previewing and thus they can have any naming rules. But our test result showed that use of North-East-West-South terminology (abbreviated as “NEWS”) was helpful. As the DXF patterns are deployed on 2D plane, they have generally have four- to eight-neighbor intra-group patterns (Figure 2(a)). For example, patterns shown in Figure 2(b) are located in parallel horizontally. We called this as “West-East” connection and the left hand side edge is given name of “WEST_1” and the right hand side edge is given name of “EAST_1”. The same method applies to “North-South” (Figure 2(c)) or “northeast-southwest” connections. This approach has an advantage that the user does not have to worry about creating different names for edges. Additionally, dart edges had name “DARTA”, “DARTB” and so on. And patterns such as arm pattern that were sewn to themselves in a cylindrical shape had edge named “SELF”. Only edges those cannot be described with previous types had to be given extra names such as “ETC_1” or else,

Ar m ho le FR

ArmR_1

A rm ho le FR

Arm 1

Front_1

Front

Figure 1. Example of inter-group sewing connection between group “ARM1” and “FRONT”

IJCST 22,2/3

Eas

West

130

t

North

South

(b) East-west type sewing

South

North

South North

Figure 2. Example of intra-group sewing

(c) South-north type sewing

West

East

West

East

(a) Example of four-direction based intra-group sewing type

but there was no need to use extra intra-group names in the four template garments used in this study. 4. Implementation 4.1 User input The whole procedure is similar to conventional 3D apparel CAD except that the edge names are needed. If the pattern data in DXF format is ready, the user should arrange the patterns on the plane and input the pattern/edge names. The relative positions of patterns determine the East-West or North-South edge names so they should be fixed once they are aligned on the plane. The following procedures went on in an automatic way. A windows program in C þ þ language was written to implement them.

Style previewing in 3D

131

4.2 Boundary mesh node generation using B-Spline When replacing a pattern group with another in style previewing, the new patterns generally have different edge lengths. But we want edges with even different arc lengths can also be sewn. This can be easily done by setting the boundary mesh nodes equal between two edges even if their total arc length is different. This may result in seam puckering in 3D drape simulation, but the same phenomenon will occur in reality, too. Thus, all the input DXF patterns (Figure 3(a)) were converted to multiple numbers of B-Splines before mesh generation (Figure 3(b)). The actual number of splines was determined by number of extremal curvature points and existence of sewing edge.

(a) Original DXF point data

(b) Piecewise B-Spline interpolation

(c) Variation of number of boundary nodes and mesh density

Figure 3. Boundary mesh node control using B-Spline interpolation

IJCST 22,2/3

132

If DXF pattern line segments had high curvature change or sewing edge in it, they were dissected into multiple splines. After all the pattern edges were interpolated by B-Splines, the mesh density was controlled (Figure 3(c)) so that the sewing edge couples have the same number of boundary mesh nodes. 4.3 Mesh generation Triangular mesh was used by Shewchuk’s (2002) conforming Delaunay triangulation. The boundary mesh node ID’s were recorded for later sewing operation. 4.4 Group identification Based on the pattern names, patterns with names of a same first-word (e.g. “Front_”) were grouped. Figure 7 shows an example of automatic grouping result of the template sets. 4.5 Intra-group edge searching In each group, intra-group edges were found and recorded. 4.6 Inter-group edge searching Among multiple groups, edges of same name were searched. As it is a rule that there should be always one pair of inter-group edge name, the searching is not time consuming by ignoring already found intra-group edges. 4.7 Assigning sewing information After all the intra- and inter-group edge pairs are found, it is time to convert them to sewing information. Sewing information for particle method means the mesh node ID’s of two points to sew. Using the boundary node ID’s found in Section 4.2, the sewing information can be filled. 4.8 Drape simulation Now all the necessary information, such as mesh information and sewing information, is given. We used particle-based method for drape simulation (Shewchuk, 2002) using matrix symmetry for fast calculation (Baraff and Witkin, 1998). The material property of the patterns was set to that of cotton. It was critical to check the cloth self-collisions because there were frequent penetration around collars. K-DOP based hierarchical method was used (Oh et al., 2006) for cloth self-contact detection. The simulation speed was real time (30 frames/s) for 1,000 mesh nodes including collision detection time. 5. Results and discussion 5.1 Test for template garment sets Before trying style previewing, simple female shirts were used to check the validity of the algorithm. Four kinds of basic designs with different details were used as template garment sets from open database of Korean sewing technology institute (Figure 4). Figure 5 shows the technical flats. They had different styles of arm length, collar, and bodice. The four sets were designated as set A, B, C and D, respectively. Set A had long arms, straight collar of one-ply pattern and the bodice had three patterns. Set B had mandarin style two-ply collar, long arms and the bodice was composed of 13 patterns.

Style previewing in 3D

133

(a) Set A

(c) Set C

(a) Set A

(c) Set C

(b) Set B

(d) Set D

Figure 4. The template DXF pattern sets

(b) Set B

(d) Set D

Figure 5. Technical flats of the template sets

IJCST 22,2/3

134

Set C had short arm, two-ply straight collar and 11-pattern bodice. Set D had round collar, short arm, and four-pattern bodice with darts. Pattern data were imported in DXF format as shown in Figure 4. Buttons, button holes and some details were ignored for simplicity’s sake. After manually assigning name information for four garment sets, respectively, (Figure 6), the groups were found automatically (Figure 7). From the drape simulation result (Figure 8), it was verified that our naming-based sewing method works well for each template garment.

Self CuffL

L

FL Waist

Dart Dart

East West

Waist F

_1

R

FR

Arm

hol

eB

Ar m ho le FR

le

Dart Dart

ho

Front_2

Front_1

(c) Set C (d) Set D Notes: Bold characters: pattern name; inverted: intra-group sewing names; the rest: inter-group sewing names

Self

CuffL_1

ArmholeBL

ArmL_1 South North ArmL_2 Self

CuffR

Ar m ho le FR _2

R

m

West Front_4 East West Front_5 East West Front_6 WaistFL

Self

NeckBR NeckBL Self L Sho lderF ulde rFR Collar_2 Shou ArmR_1 Ar mh South ole FR Collar_1 North _1 Sho Self ArmR_2 ulde rFR rFL Self ulde Sho NeckFR NeckFL Self

CuffR_1

CuffL_1

CuffL CuffL

kF

Front_7

ec

Front_3

N

Ar

WaistFR Front_1 East West East Front_2

Back_2

Self

CuffL

CuffL_1

Back_5

East West Back_6 WaistFL

East West Back_4 WaistFL

Dart Dart

WaistFR

rm

A

BL

le

ho

rm

A

East

West

CuffR

CuffR

ArmholeBR

CuffR_1

_2

Back_1

Self CuffL

Cuff R

CuffR

CuffR_1

CuffL

CuffL CuffL_1

2 BL _ le

East

West

BR

Figure 6. Examples of sewing names for the four template garment sets

Self

ho

le

Front_3 WaistFL

ho Self

Back_3

WaistFR Back_1 East West Back_2

BL le A WaistFL

East

West

m

Front_1

WaistFR

Ar

Armh _1 oleBL oleBR Self _1 Sou th Armh Ar Sou th mh Nor th Self Nor th ole BR Front_4 Front_2 R L F kF L Shou lderF Nec Neck houlderF ArmR_1 R Collar_2 S ArmL_1 South R North Ar mh leF ho ole Collar_1 m FL Self Ar R Shou lderF Self Self lderF Shou L FL k c 2 _ Ne ArmholeFL_1 oleFR h rm A N. N. _2 .E. S.E W. N.E. . S.E W. N .W. leFL . S.W S mho . Ar

Self Self

Self

ArmL_1

(b) Set B

Dart Dart

R ckF

FR Ne

Ar m ho le FR WaistFR Armh oleBR Front_1 ArmholeFR_1 _1 East _2 A F le rmho o R West leBR Armh _2 Front_2 3 A _ rm R holeB East East oleF R_3 West Armh West Ne Front_3 ckF East NeckBR R West Front_4 East West FL NeckBL Front_5 Neck L_1 East East Armh holeB oleFL _1 Arm West Front_6 West Armh _2 East oleFL oleBL h rm _2 A West _3 Front_7 Armh oleBL oleFL Armh _3 WaistFL

WaistFL ArmholeBL

NeckBL

NeckBL

NeckFR NeckFL

NeckBR

le ho m

Ar

FL

mh

ole

ole

mh

(a) Set A

BL

ole

mh

Ar

FL

Self

L

Ar L

WaistFR

BR

ArmR_1

Front_2

Front_1

Collar_2 South North Collar_1

ole

le

Self

mh

ho

Self

Ar

rm

ckF

lderF

Self

Self

Self

A

FL le ho

rm

Ar

Self ArmL_1

Shou

Ne

R

lderF

Shou

Self

A

FR

CuffR

Shou Collar_1

ArmR_1

Self

L

lderB

R

BR

le

ho

lderB

ho

Shou

m

Self

NeckBR

ArmholeBR

Ar

Self

Self CuffR_1 Cuff R

Back_1

rm

WaistFR

5.2 Style previewing test As the four template garments has seven distinctive groups, there can be 47 ¼ 16,384 possible choices of mixing the pattern groups. Among them, we chose four mixing types for test purpose. 5.2.1 Type nos 1b-1d. Starting from the garment set A, the left and right arms/cuffs were replaced with those of set B (designated as type no. 1b), C (type no. 1c) and

Collar (B) Arm2 (B)

Arm1 (B)

Cuff2 (B)

Cuff1 (B)

Collar (A)

135

Arm2 (A)

Front (A)

Front (B)

(a) Set A

(b) Set B Back (D)

Front (C)

(c) Set C

Arm2 (C)

Cuff1 (D)

Collar (C) Arm1 (C)

Cuff2 (C)

Back (C)

Cuff2 (D)

Arm1 (A)

Cuff1 (C)

Style previewing in 3D

Back (B)

Cuff2 (A)

Cuff1 (A)

Back (A)

Collar (D) Arm2 (D)

Arm1 (D)

Front (D)

(d) Set D

D (type no. 1d). Figure 9 shows the concept of 3D style previewing for this example. Note that the user does not have to input edge names again, which were already input in the template sets in Figure 6. The user has only to select which groups to use for a new garment set. As shown in Figures 10-12, the new garment sets had different arm patterns. 5.2.2 Type nos 2c and 2d. In type no. 2, set B was used as a base and the collar patterns were replaced with those of set C (type no. 2c) and D (type no. 2d). Figures 13 and 14 show the result. It is evident that the new garment sets have different collar patterns, while maintaining the original bodice patterns. 5.2.3 Type no. 3. In this type, set B was used as a base and the right arm/cuff patterns were replaced by those from set C. Also the left arm and left cuff were replaced by set D. Figure 15 shows the result. There seemed some puckering around shoulders because the length of new collar was shorter than that of the original one. 5.2.4 Type no. 4. Type no. 4 garment set was composed of front/back bodice of set C, right arm/cuff of set A, collar of set B and left arm/cuff pattern of set D. Figure 16 is the result.

Figure 7. Pattern DXF data and their automatic grouping result

IJCST 22,2/3

136 (a) Set A

(b) Set B

(c) Set C

(d) Set D

Figure 8. Drape simulation result of the four template garment sets

Figure 9. Conceptual view of 3D style previewing for mix type no. 1

It is notable that the user’s job is to only fill pattern names and edge names in Figure 6. The rest are done automatically. We implemented a computer algorithm to convert name-based sewing information into vertex ID’s for that. The drape simulation used time step of 33 ms for about 100 iteration loops. The drape simulation took about one minute in AMD Athlon 64 X2 3 Ghz PC with 2 GB RAM and Geforce 7900 GT graphic card. The sewing and meshing process took also less than a minute. But this could be skipped if all the mesh densities were set to an equal value. This means real-time pattern substitution can be possible. (The color figures and movies are available at: http://snowman0.com/StylePreview3D/).

Style previewing in 3D

Collar (A) Arm1 (B)

Arm2 (B)

Cuff2 (B)

Cuff1 (B)

Back (A)

137

Front (A)

(a) DXF pattern configuration

(b) Sewing result

(c) Simulation result

6. Conclusions A new name-based sewing rule for 3D garment patterns were proposed. Female shirts of four kinds of basic design were used for verification, but the same method can be applied to other kind of garments, such as slacks, blouse, one-piece dress, men’s suit and so on. The user’s work is only to prepare pattern data and assigning names

Figure 10. Result of mix type no. 1b

IJCST 22,2/3 Back (A)

Cuff2 (C)

Cuff1 (C)

138

Collar (A)

Arm1 (C)

Arm2 (C)

Front (A)

(a) DXF pattern configuration

(b) Sewing result

(c) Sewing and simulation result

Figure 11. Result of mix type no. 1c

Style previewing in 3D Back (A)

Cuff2 (D)

Cuff1 (D)

139

Collar (A) Arm1 (D)

Arm2 (D)

Front (A)

(a) DXF pattern configuration

(b) Sewing result

(c) Simulation result

Figure 12. Result of mix type no. 1d

IJCST 22,2/3

Cuff1 (B)

Collar (C) Arm1 (B)

Arm2 (B)

Front (B)

(a) DXF pattern configuration

(b) Sewing result

(c) Simulation result

Figure 13. Result of mix type no. 2c

Cuff2 (B)

Back (B)

140

Style previewing in 3D Back (B)

Arm1 (B)

Collar (D)

Arm2 (B)

Cuff2 (B)

Cuff1 (B)

141

Front (B)

(a) DXF pattern configuration

(b) Sewing result

(c) Simulation result

Figure 14. Result of mix type no. 2d

IJCST 22,2/3 Back (B)

Cuff2 (D)

Cuff1 (C)

142 Collar (B) Arm1 (C)

Arm2 (D)

Front (B)

(a) DXF pattern configuration

(b) Sewing result

(c) Simulation result

Figure 15. Result of mix type no. 3

Style previewing in 3D Back (C)

Collar (B) Arm1 (A)

Arm2 (D)

Cuff2 (D)

Cuff1 (A)

143

Front (C)

(a) DXF pattern configuration

(b) Sewing result

(c) Simulation result

Figure 16. Result of mix type no. 4

IJCST 22,2/3

144

to patterns and edges. We believe that our method can be used as a standard rule for 3D garment sewing description, because our method can be used to arbitrary garment data. References Baraff, D. and Witkin, A. (1998), “Large steps in cloth simulation”, Proceedings of the Annual Conference Series on Computer Graphics, Vol. 32, pp. 43-54. Eckert, C.M. and Helmut, E.B. (1999), “A garment design system using constrained Bezier curves”, International Journal of Clothing Science & Technology, Vol. 12 No. 2, pp. 134-43. Ji, Y.A., An, J.S., Lim, K.S. and Lee, D.H. (2002), “An introduction to a garment technical drawing system and its DB construction methodology”, International Journal of Clothing Science & Technology, Vol. 14 Nos 3/4, pp. 247-50. Luo, Z.G. and Yuen, M.M.F. (2005), “Reactive 2D/3D garment pattern design modification”, Computer-Aided Design, Vol. 37 No. 6, pp. 523-630. Oh, S., Ahn, J. and Wohn, K. (2006), “Low damped cloth simulation”, The Visual Computer: International Journal of Computer Graphics, Vol. 22, pp. 70-9. Shewchuk, J.R. (2002), “Delaunay refinement algorithms for triangular mesh generation”, Computational Geometry: Theory and Applications, Vol. 22 Nos 1/3, pp. 21-74. Wang, C.C.L., Wang, Y. and Yuen, M.M.F. (2003), “Feature based 3D garment design through 2D sketches”, Computer-Aided Design, Vol. 35 No. 7, pp. 659-72. Further reading Sul, I.H. (2010), “Fast cloth drape simulation using voxel based indexing and collision matrix”, International Journal of Clothing Science & Technology, Vol. 22 Nos 2/3 (in press). Corresponding author In Hwan Sul can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

Fast cloth collision detection using collision matrix

Fast cloth collision detection

In Hwan Sul College of Engineering, i-Fashion Technology Center, Konkuk University, Seoul, South Korea

145 Received 25 July 2008 Accepted 12 June 2009

Abstract Purpose – The purpose of this paper is to develop a new and simple methodology for fabric collision detection and response. Design/methodology/approach – A 3D triangle-to-triangle collision problem was converted to simple 2D point-in-triangle problem using pre-computed 4 £ 4 transformation matrices. The object space was partitioned using voxels to find easily collision pair triangles. k-DOP was used to find inter-pattern collisions. Findings – Complex 3D collision detection problem is solved by simple matrix operations. Voxel-based space partitioning and k-DOP-based hierarchical methods are successfully applied to garment simulation. Originality/value – This paper shows that the collision matrix method can cover from triangleto-point to triangle-to-triangle collision with mathematical validity and can be simply implemented in garment simulation. Keywords Cloth, Simulation, Fabric production processes, Textile technology Paper type Research paper

1. Introduction Garment is 3D product which is sewn from 2D fabrics (patterns). To predict the final try-on shape of the garment, both material property of the clothes and bodice shape are needed. Material property of clothes determines the deformation of the garment while bodice shape constrains the final shape of the garment. For the modeling of fabric deformation, continuum approaches such as finite element method (FEM) or particle-based method were applied successfully. Kang and Yu (1995) adapted explicit FEM for fabric deformation. Etzmuss et al. (2003) also showed realistic virtual try-on simulation of a male suit. Although FEM is a reliable method for mechanical analysis, particle-based method was more widely used because calculation speed is an important factor for graphics and apparel design purpose, particle-based method seems to be more proper method for repetitive drape simulation such as garment designing. Modeling of fabric deformation has long been studied and it is not main concern of this paper. The other major factor that determines the garment product shape is collision reaction between garment and the body. Collision detection also becomes a rate determining step when the garment has multiple layers of fabrics. Collision detection among multiple bodies is an important theme in computer graphics area because no natural phenomena can be described without collision reactions. There have been many researches for accurate and fast collision detection. The simplest approach for detecting collision between triangular mesh elements is to check whether the two triangles overlap or not. GJK (van den Bergen, 1999),

International Journal of Clothing Science and Technology Vol. 22 No. 2/3, 2010 pp. 145-160 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011018621

IJCST 22,2/3

146

V-Clip (Mirtich, 1998) and I-Collide (Cohen et al., 1995) were the famous tools to compute the distance between objects. But comparing every triangle pair in each time step needs a massive calculation time and the burden becomes bigger when the number of mesh triangles increase. So wrapping the whole object into simple bounding objects such as boxes or spheres and collisions between bounding objects can be used for preliminary test to skip actual test. They are called as bounding box (Volino and Magnenat-Thalmann, 2000) or bounding sphere according to their shape. If the object mesh has complex shape, then the hierarchy of the bounding volumes can be used instead. Such hierarchy-based method showed notable time decrease. However, the actual triangle-to-triangle pair test is inevitable when the preliminary bounding volume test fails. This is the usual case in garment drape simulation where the cloth always contacts the body surface. Therefore, it is best to decrease the complexity of the triangle pair collision test to increase the overall drape simulation speed. Roughly, tens of iterations are needed for an expert apparel computer-aided design operator to initially align 2D pattern on proper 3D space even when the patterns are completely prepared. Therefore, fast drape simulation is the first necessary condition for fast garment pattern designing. This paper approximated the triangle pair collision test with simple matrix operations. The other time-consuming step in collision detection is to find possible collision pairs. Above-mentioned hierarchy method can also be used for finding possible collision pairs so we used k-DOP for hierarchical representation of meshes. Moreover, this paper also used voxelization of the space. The object 3D space was voxelized into multiple cells and each voxel recorded cloth or body mesh element ID’s at each frame. Voxel-based method was already applied for collision detection (Zhang and Yuen, 2000) but they voxelized the object itself so the voxels could not represent the original object smoothly due to memory limit. We voxelized the space into space roughly and used them indirectly as a storage to record the nearby mesh element ID’s. The detailed formulations are described in the following chapters and the results are displayed. 2. 3D voxelization 2.1 Voxel-based collision detection Not only actual triangular overlapping test, but also finding possible collision pair of triangles are rate determining step of drape simulation. Several approaches to find quickly possible pairs were done and the simplest approach is using voxel. The advantage of voxel is that the voxel index can be easily calculated from the coordinates. Figure 1 shows finding voxel ID from x-, y-, and z-coordinates. Once the voxel ID’s of reference triangle and target triangle are known, it can be known if they are possible collision pairs by checking whether they share common voxel ID. However, the disadvantage of voxel is aliasing and memory limit. As the voxel has box axis-aligned shape and the body elements have curved shape, there must be a voxelization error coming from aliasing. Smaller size of voxel can reduce the error, but using 1,0243 voxels need 109 space of memory. Without any further encoding procedure, simple voxelization cannot replace actual collision detection but it can be used only as an preprocessing method. Therefore, this paper used low density of voxels and the voxels are used only for registering element ID’s. As the cloth or body element deforms or moves, their position

Fast cloth collision detection

Voxel space

Voxel index j

Vi–1 j,k

Vi, j,k

→

Cloth node R = (x, y, z)

147 Vi–1, j–1,k

Vi,j–1,k

Voxel index i Voxel index k

index_i = int | (x–x_min) / VoxelSize | index_ j = int | (y–y_min) / VoxelSize | index_ k = int | (z–z_min) / VoxelSize |

Vi,j,k = index_i *n VoxelY *n VoxelZ +index_ j*nVoxelZ +index_k

is recorded in each voxels, respectively. Then, elements within same voxel ID become possible collision pairs. This method facilitates finding the possible pairs than finding nearly elements randomly. 2.2 3D voxelization of triangular prism Finding voxel ID of a vertex is easy as shown in the Figure 1. To find voxel ID’s of a triangular face, 2D rasterization algorithm was adopted. Bresenham (1965) algorithm is a method used for finding pixel area of a given polygon. This paper applied and expanded this method to 3D. As shown in Figure 2(a), edges of triangles are first voxelized. And then with respect to z-axis, overall voxels are separated into layers and each layer is applied Bresenham algorithm. Then, 3D voxel shape of the original triangle is acquired by accumulating the layers. But voxelizaing only triangle face can lead to errors so that nearby triangle pair can have different voxel ID’s. To avoid the error, this paper presents voxelization of triangular prism which is formed by upper and lower triangles. The upper and lower triangle is found by mesh enlarging and shrinking (Figure 2(b)). The coefficient of enlarging and shrinking was that of Taubin’s (1995) mesh smoothing algorithm so that the average volume is the same with the original mesh. Use of Taubin’s algorithm guarantees that each voxel only contains only one element ID, because Taubin’s method produces no volume change and no element entangling occurs. 2.3 Finding collision pair from voxel ID As each triangle element has multiple number of voxel ID’s, they should have a variable to keep voxel ID’s. To test collision of a triangle T, each voxels are opened and target triangle ID’s are acquired from the voxels. And then triangle-to-triangle test is

Figure 1. Finding voxel index from coordinates

IJCST 22,2/3

148 (a) 2D rasterization using Bresenham algorithm

Inflated layer

Original triangle

Figure 2. Voxelization scheme

Deflated layer (b) 3D voxelization of triangular prism

done with the target triangle. If collision has occurred, collision test of triangle T is finished. Otherwise another voxel or target triangle is compared until all the voxels are tested. This may seem redundant but actually voxels has only one element ID when the two objects are distant apart. Otherwise when the two objects are closer together, collision test end in the first voxel. Therefore, voxel-based collision pair finding is suitable for garment drape simulation. The simulation result shows the validity. 3. Transformation matrix for collision detection 3.1 Triangle-to-triangle collision detection Collision types between two triangles can be classified to three cases, which are node infiltration, edge infiltration, and face penetration (Figure 3). Node infiltration is the case in which a vertex lies under the reference triangle. Edge infiltration is similar to node infiltration but the intersection between edge and the reference triangle lies inside the reference triangle. Face penetration occurs when two triangles shear a common edge. To deal with three cases, both reference triangle and target triangle are rotated and translated so that the reference triangle lies on the xy-plane with one vertex clamped at the origin. Then the problem becomes a 2D point-in-triangle problem. Therefore, 3D triangle-to-triangle overlapping problem can be solved by two operators such as MapToXYPlane ( ) and PointInTriangle ( ). Assume ri, ui, fi are spherical coordinates of normal vector ni of reference triangle i. and pi0, pi1 and pi2 are the three vertices of triangle i. Then, MapToXYPlane() operation can be represented by a matrix operation Mi: p0 ¼ M i ð p 2 pi0 Þ þ pi0

→ →

p' ( p'x, p'y , p'z)

Z

p ( px, py , pz)

→

n'=(0,0,1) = (r' =1, 0,

Rotation and translation

→ n ( nx, ny, nz) = (r = 1, q, f)

π ) 2

Fast cloth collision detection

Y

149

→

p' ( p'x , p'y ,0)

Z Y

X

X

(a) Point infiltration →

p→( px, py, pz)

p' ( p'x, p'y,0) →

n' (0,0,1)

Rotation and translation

→

n

Y

Z →

→

q ( qx, qy, qz)

Y

q' ( q'x , q'y , q'z)

X

X (b) Edge penetration →

→

r'

→

p ( px, py , pz)

p'

r→( rx, ry , rz)

→

n' (0,0,1)

n→

Rotation and translation

Y

Z →

q ( qx, qy, qz)

Y

Figure 3. Three cases of triangular collisions

→

q'

X

X (c) Face penetration

2 6 M i ¼ M i ðu; fÞ ¼ 6 4 2 6 ¼6 4

cos fi

0 2sin fi

0

1

sin fi

0

32

cos ui

76 76 2sin ui 54 cos fi 0 0

cos fi cos ui

sin ui cos fi

2sin fi

2sin ui

cos ui

0

cos ui sin fi

0

sin fi

sin ui cos ui 0

0

3

7 07 5 1

3 7 7 5

And then conventional PointInTriangle( ) operation comes to check target vertex lies in the moved triangle.

IJCST 22,2/3

150

3.2 Precomputed collision matrix Transforming target triangle with respect to the reference triangle at every simulation frame is redundant when the target triangle belongs to bodice which does not move during drape simulation. In that case, transformation matrix Mi can be pre-calculated only at the first time. But still PointInTriangle() should be checked to know whether the target triangle lies inside the reference triangle or not. This can be omitted by adding additional transformation to Mi so that target triangle always have the three vertices at (0, 0), (1, 0) and (0, 1). By applying Mi, the first vertex always lies on the origin while the other two vertices do not. To match the other two points at desired positions, respectively, total five matrix operations were done. The 4 £ 4 matrix instead of 3 £ 3 of Mi was used to implement translation transformation. Figure 4 shows the graphical illustration of the procedure: . Step 1 (Translate): 3 2 1 0 0 2pi0x 7 6 6 0 1 0 2p 7 6 i0y 7 7 6 M Trans3D ¼ 6 7 6 0 0 1 2pi0z 7 7 6 5 4 0 0 0 1

.

.

M Trans3D translates the whole triangle by the displacement of pi0 from the origin. Step 2 (Rotation): 3 2 cos fi cos ui sin ui cos fi 2sin fi 0 7 6 6 2sin u cos ui 0 07 7 6 i 7 6 M Rot3D ¼ 6 7 6 cos ui sin fi 0 sin fi 0 7 7 6 5 4 0 0 0 1 M Rot3D rotates the triangle i so that the new normal vector becomes (0, 0, 1). Step 3 (Rotation2D): 2 3 sin ai 0 0 cos ai 6 7 6 2sin a cos a 0 0 7 6 7 i i 6 7 M Rot2D ¼ 6 7 6 0 7 0 1 0 6 7 4 5 0 0 0 1 where ai is the angle between x-axis and the edge p00i0 p00i1 . M Rot2D rotates the triangle so that edge p00i0 p00i1 lies on the þ x-axis.

pi0 n→( nx, ny, nz) =(r = 1, q, f) pi1

Z X

Fast cloth collision detection

Z M Trans3D

p'i0

Y

pi2

151

Y

X p'i1

(a) Matrix M Trans3D

p'i2

Z Z p'i0

Y M Rot3D n→( nx, ny, nz) = (r = 1, q, f)

X

p'i1

p'i2 (b) Matrix M Rot3D

→

n' (0,0,1) π = (r' = 1,0, 2 )

p''i0

Y

p''i1

X

Y

Y qi2

p''i2 M Rot2D p''i0

X

αi p''i1

qi0

qi1

X

(c) Matrix M Rot2D Y

Y qi2

q'i2

gi

qi0

MShear2D qi1

X

q'i0

q'i1

X

(d) Matrix MShear2D Y

Y 1

1 q'i2 MScale2D

q'i0

1 q'i1

X

X 0

(e) Matrix MScale2D

1

Figure 4. Procedure of triangle transformation

IJCST 22,2/3

.

Step 4 (Shear2D): 2 M Shear2D

152

.

1 gi

0

1

0

0

1

0

0

6 60 6 ¼6 60 4 0

0

3

7 07 7 7 07 5 1

where gi is the shear angle of edge qi0 qi2 with respect to the y-axis. M Rot2D deforms the triangle so that point qi2 lies on the y-axis. Step 5 (Scale2D): 2 3 sx 0 0 0 6 7 6 0 sy 0 0 7 6 7 M Scale2D ¼ 6 7 6 0 0 1 07 4 5 0 0 0 1 where sx ¼ 1=q0i1 ; sy ¼ 1=q0i2 . M Scale2D shrinks or deflates the triangle so that it matches (0, 0) (1, 0) and (0, 1). So the final matrix would be combination of the above transformations: P i ¼ M Scale2D · M Shear2D · M Rot2D · M Rot3D · M Trans3D

It is apparent that matrix Pi transforms any original triangle i to fixed triangle T0 ¼ {(0, 0), (1, 0), (0, 1)}. Then, the point-in-triangle test can be simply done by checking any point p lies inside T0. This is much easier than 3D triangle-to-triangle test where the triangle has arbitrary coordinates. Although Pi contains shearing and scaling deformation and they are not affine transform, they do not affect collision geometry because all the space is deformed linearly. The actual collision was classified as three types. The first is point penetration type, where a point lies inside the triangular column whose cross-section is the same with the reference triangle Ti. The second is edge penetration type, where an edge e crosses the reference triangle, i.e. e > T i – 0 and e > T i , T i . The last is face penetration type, where a target triangle T1 meets with reference triangle, i.e. T 1 > T i – 0 and T 1 > T i , T i . It is easy to check point infiltration type by seeing if the point lies insides the area formed by y ¼ 0, x ¼ 0 and y ¼ 2 x þ 1 curves. Point infiltration type alone can be used for simple collision test instead of all the three tests, but in that case convex mesh elements have a dead angle. Therefore, three types of tests should be used simultaneously. 3.3 Approximation for transformation Pi When the collision checking is only for the body-to-cloth and the body does not have motion, preparation of Pi can be done only one time. But when the body has motion or the checking is cloth-to-cloth, the coefficients, ai, gi, sx and sy cannot be known until the vertices are pi1 and pi2 transformed by M Rot3D · M Trans3D . It is not desirable to calculate

Pi at every frame. If we assume that the shear angle of edge pi0 pi1 and pi0 pi2 does not change during simulation and the initial values can be used as constants. Then, the approximated matrix Q i can be simply acquired as: Q i ¼ M Scale2D · M Shear2D · M Rot2D · M Rot3D · M Trans3D 2 3 Q i12 Q i13 Q i14 Q i11 6 7 6 7 Q i22 sy sin a sin f Q i24 7 6 Q i21 6 7 ¼6 7 6 cos u sin f sin u sin f cos f Q i34 7 6 7 4 5 0 0 0 1 where: Q i11 Q i12 Q i13 Q i14 Q i21 Q i22 Q i24 Q i34

¼ sx cos u cos fðcos a 2 g sin aÞ 2 sx ðg cos a þ sin aÞsin u ¼ sx ðcos aðg cos u þ cos f sin uÞ þ sin aðcos u 2 g cos f sin uÞÞ ¼ sx ð2cos a þ g sin aÞsin f ¼ 2sx ðcos að pi0x cos uðg þ cos fÞ þ pi0x ð2g þ cos fÞsin u 2 pi0z sin fÞ2 sin að pi0x cos uð21 þ g cos fÞ þ pi0x ð1 þ g cos fÞsin u 2 pi0z g sin fÞÞ ¼ 2sy ðcos u cos f sin a þ cos a sin uÞ ¼ sy ðcos u cos a 2 cos f sin a sin uÞ ¼ sy ð2pi0x cos aðcos u 2 sin uÞ þ sin að pi0x cos u cos f þ pi0x cos f sin u2 pi0z sin fÞÞ ¼ 2pi0x ðcos u þ sin uÞsin f 2 pi0z cos f

Qi’s can be prepared preliminarily and can be updated by inserting normal vector information such as ui’s and fi’s. 4. Finding inter- and intra-pattern collision pairs 4.1 Finding collision pairs using k-DOP tree Using space dividing scheme including voxel-based method is a good choice for culling possible collision pairs of non-deformable objects such as body mesh data. Since body data are pre-made and can be modified to a minimum number of mesh elements so that the size and number of voxels can be minimized. It can be also adapted to the case of inter- or intra-pattern collisions, but the voxel size should be more fine and the voxel information should be updated in every frame because the patterns deforms and moves (Teschner et al., 2004). So we used a better collision culling scheme using k-DOP tree (Klosowski et al., 1996). k-DOP tree is an extended method of axis-aligned bounding box (AABB) to a user defined number of coordinate axes. It should also be updated in each and every frame but it can be easily updated just the way as the AABB’s are updated. We constructed one k-DOP tree for each garments pattern meshes and the bottommost leaf nodes contained the actual k-DOP information of mesh triangle elements. If the garment patterns are far away, the root nodes of k-DOP trees do not collapse and the collision detection can be skipped. If the patterns collides at all, the child nodes among the k-DOP trees are compared and the colliding triangle element pairs can be known instantly.

Fast cloth collision detection 153

IJCST 22,2/3

154

Figure 5. Examples of body mesh voxelization

5. Results and discussion 5.1 Body mesh voxelization Figure 5 shows the example of body mesh voxelization. Both the arbitrary object (Figure 5(a)-(d)) and the human body data (Figure 5(e)-(f)) can be transferred to voxel space with the desired voxel sizes. Of course, there is trade-off between voxel size and efficiency. The smaller voxel size becomes, the voxels contains less elements and the possible collision pair of triangle elements reduces which results into faster collision detection speed. The bigger the voxel becomes, the smaller memory is needed for keeping the voxel information. Even though the respective voxel contains very small amount of data (ID’s of triangle elements which resides in the voxel), the total number of voxels can exceed one million if we voxelized the 1 m3 space with 1 cm interval. But if we

(a) Original Stanford bunny mesh data

(b) Voxels of 2 cm intervals

(c) 0.8 cm

(d) 0.4 cm

(e) Female avatar mesh

(f ) 4 cm voxelization

can reduce the number of mesh elements of the body data, we can use bigger size of voxels. In this paper, 10 cm voxel was used for the body mesh shown in Figure 5(e), which has 7,034 triangular elements (excluding the hair, shoes, and garments mesh in Figure 5(f)). 5.2 Inter-pattern collision detection using k-DOP tree To resolve the collision between garment patterns, each pattern was given with k-DOP tree structure and the collision among k-DOP nodes were used as a preliminary test to find possible collision pairs. Figure 6 shows a real time free fall simulation of 24 pieces

Fast cloth collision detection 155

(a) Bird-eye view of initial state

(b) Top view of the final stacking

(c) Lateral view

Figure 6. Real time stacking simulation of 24 colored papers on a marble plate

IJCST 22,2/3

of 4 £ 4 cm papers (or patterns). Each paper was given a different color to easily identify the penetration. The figure shows no inter-pattern collisions occur even though the simulation is real-time. Figure 7 shows the several k-DOP nodes in different levels. A 18-DOP (which has degree of freedom of nine axes) was used but simply their

156

(a) Simulation result

(b) Root nodes of k-DOP trees

(c) 2nd level nodes

Figure 7. k-DOP’s of different hierarchy levels

(d) 6th level nodes Note: k-DOP's are roughly shown as AABB's for the sake of convenience

AABB’s were drawn for simplicity’s sake because the actual 18-DOP is difficult to draw in real time (Klosowski et al., 1996). Once the possible collision pairs are known by k-DOP tree test, the actual collision detection between two triangles are done by using collision matrix Mi (where i is the index of the lower lying triangle element). With this strategy, the self-collision (collision inside each pattern) can be also searched within the tree. But including self-collision takes results into slower simulation speed and self-collision occurs in specific cases such as dress, shirt collar or skirt. But in those cases, possible self-collisions can be avoided by assigning multiple number of k-DOP trees to each patterns. In this investigation, only simple patterns which definitely have no self-collision were used. 5.3 Garment try-on collision test To verify the exactness of the collision matrix-based test, virtual garment try-on simulation was done. The garment pattern data was imported from DXF file of real garment patterns and the textures were from graphical artworks. Semi-implicit particle-based method of Baraff and Witkin (1998) was used for cloth simulation with time step of 20 ms. Simple shirt, pants, and vest patterns were used and they were triangulated so that the edges have approximately have 1 cm length. The patterns were initially aligned around the body mesh (Figure 8(a)) and three types of tests were done. Figure 8(b) shows the shirts above pants case. Figure 8(c) is the reverse case. In both cases, upper patterns are covering the lower patterns successfully. If the inter-pattern collision occurs, the outer patterns were lifted to collision-free position with respect to the lower pattern normal vector. Also, the inner patterns were given pressing force from the outer patterns. Figure 8(d) shows the vest is pressing and preventing the shirt pattern from falling-off. Figure 9 shows the collision status with red spheres (cloth-to-body) and blue cones (cloth-to-cloth). 5.4 Calculation time Table I summarizes the simulation time of colored papers and try-on trials. Cloth meshes of about thousand vertices can be simulated in real time (15 frames/sec in this case of stacking colored paper). Complex cloth and body meshes increases the total simulation time, but they increase linearly. 6. Conclusions Various techniques exist for collision detection algorithm but most of them are focused on objects moving independently. In the case of garment simulation, the body and cloth is under constant collision state and effective collision culling algorithm is important. We used voxel-based space dividing scheme for fast body-to-cloth collision detection and k-DOP-based hierarchical method for cloth pattern-to-pattern collision detection. The actual collisions between two triangles were checked by multiplying simple 4 £ 4 collision matrices to vertex coordinates. Using the collision matrix, the 3D collision test problem is simplified to 2D point in triangle test. The garment trial simulation tells that the collision matrix-based method can be almost real time for small number of vertices. Even for the large number of mesh vertices, the results showed reliability.

Fast cloth collision detection 157

IJCST 22,2/3

158

Figure 8. Garment collision detection using k-DOP tree and collision matrix

(a) Initial position of patterns

(b) Shirt above pants

(c) Pants above shirt

(d) Vest above shirt and pants

Fast cloth collision detection 159

Figure 9. Collision status

Notes: Red sphere: cloth-to-body penetration; blue cone: cloth pattern-to-pattern contact

Cloth mesh data Body mesh data Drape simulation time (ms/frame) No. of No. of No. of No. of No. of Semi-implicit Collision patterns vertices elements vertices elements integration detection Total Color paper stacking Pants and shirt Vest, shirt, and pants

24

1,440

2,040

242

478

33.8

30.2

64.0

10

5,587

9,202

7,034

13,850

188.0

125.0

313.0

18

7,065

11,533

7,034

13,850

236.0

147.0

383.0

Note: Including rendering

References Baraff, D. and Witkin, A. (1998), “A large steps in cloth simulation”, Proceedings of the Annual Conference Series on Computer Graphics, pp. 43-54. Bresenham, J.E. (1965), “Algorithm for computer control of a digital plotter”, IBM Systems Journal, Vol. 4 No. 1, pp. 25-30. Cohen, J.D., Lin, M.C., Manocha, D. and Ponamgi, M.K. (1995), “I-COLLIDE: an interactive and exact collision detection system for large-scale environments”, Proceedings of the ACM Interactive 3D Graphics Conference, pp. 189-96. Etzmuss, O., Keckeisen, M. and Strasser, W. (2003), “A fast finite element solution for cloth modeling”, Proceedings of the 11th Pacific Conference on Computer Graphics and Applications, pp. 244-51.

Table I. Speed of simulation and collision detection

IJCST 22,2/3

160

Kang, T.J. and Yu, W.R. (1995), “Drape simulation of woven fabric by using the finite-element method”, Journal of Textile Institute, Vol. 84 No. 4, pp. 247-57. Klosowski, J.T., Held, M., Mitchell, J.S.B. and Sowizral, H. (1996), “Efficient collision detection using bounding volume hierarchies of k-DOPs”, IEEE Transactions on Visualization and Computer Graphics, Vol. 4 No. 1, pp. 21-36. Mirtich, B. (1998), “V-clip: fast and robust polyhedral collision detection”, ACM Transactions on Graphics, Vol. 17 No. 3, pp. 177-208. Taubin, G. (1995), “Curve and surface smoothing without shrinkage”, 5th International Conference on Computer Vision, pp. 852-7. Teschner, M., Kimmerle, S., Heidelberger, B., Zachmann, G., Raghupathi, L., Fuhrmann, A., Cani, M.-P., Faure, F., Magnenat-Thalmann, N., Strasser, W. and Volino, P. (2004), “Collision detection for deformable objects”, Proceedings of the Eurographics, pp. 119-35. van den Bergen, G. (1999), “A fast and robust GJK implementation for collision detection of convex objects”, Journal of Graphics Tools, Vol. 4 No. 2, pp. 7-26. Volino, P. and Magnenat-Thalmann, N. (2000), Virtual Clothing: Theory and Practice, Springer, Berlin. Zhang, D. and Yuen, M.M.F. (2000), “Collision detection for clothed human animation”, Proceedings of the 8th Pacific Conference on Computer Graphics and Applications, pp. 328-37. Corresponding author In Hwan Sul can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

Study of heat transfer through layers of textiles using finite element method Yuchai Sun College of Textiles, Dong Hua University, Shanghai, China and Hebei University of Science and Technology, Shijiazhuang, China

Xiaogang Chen

Study of heat transfer

161 Received 7 January 2009 Accepted 12 September 2009

School of Materials, The University of Manchester, Manchester, UK

Zhonghao Cheng Hebei University of Science and Technology, Shijiazhuang, China, and

Xunwei Feng College of Textiles, Dong Hua University, Shanghai, China Abstract Purpose – The purpose of this paper is to present the results of a study on heat transfer through a textile assembly consisting of fabric and air layers based on a theoretical model capable of dealing with conductive, convective and radioactive heat transfer. Design/methodology/approach – Quantificational results were given out by the aid of finite element (FE) analysis software MSC MARC Mentat. Findings – Significant findings through this paper include the change in heat flux against time and the transit temperature distribution at the cross-section of the fabric assembly. The size of the air gaps has a significant influence on the heat transfer. The balance heat flux drops by 40 per cent when the air gap increases from 2 to 10 mm. The influence of the air gap tends to become smaller as the air gap is further increased. The number of fabric layers in the textile assembly has a noted influence, more so when the ambient temperature is lower. Comparisons between the theoretical and tested results show a good agreement. Originality/value – This paper has established a new method for clothing comfort study by making use of a general purpose FE method software package. Keywords Finite element analysis, Heat transfer, Flux, Textile making-up processes Paper type Research paper

1. Introduction Heat transfer through a textile assembly or a fabric system is a complex process, involving conduction, radiation and convection. The combined heat transfer across the fabric system, consisting of fabric and air layers, is not simply the sum of what each mechanism would do in the absence of the others. The three heat-transfer mechanisms work together to determine the characteristics of the overall heat-transfer process. At different conditions, these heat-transfer mechanisms are of different importance. It has long been realised that heat transfer through textile assemblies involves multiple mechanisms. Peirce and Rees (1946) pointed out in 1946 that at the outer surface of the clothing exposed to the air, heat is lost by means of both convection and radiation.

International Journal of Clothing Science and Technology Vol. 22 No. 2/3, 2010 pp. 161-173 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011018630

IJCST 22,2/3

162

However, their work stopped short in establishing a model to describe the combined influence from the two. Farnworth (1983) presented a theoretical treatment of heat transfer through a bed of fibres considering conduction and radiation. It was illustrated that the theoretical results fitted well with the experimental data from testing some fibrous insulating materials. He reported that no detectable convective heat transfer took place inside the fibre bed. He assumed that the fibre bed is placed between two plates and thus the boundary conditions were not included in his work. In a more recent study, Mohammade et al. (2003a, b) presented a theoretical equation of the combined thermal conductive, convective, and radioactive heat flow through heterogeneous multi-layer fibrous materials. They used the effective thermal conductivities of the samples obtained by an experimental method, along with Fricke’s formula (Fricke, 1924; Fricke et al., 1990) for thermal conduction, to calculate the radiation thermal conductivity at high temperatures. Again, they did not take into account the boundary conditions at the surface of textile material. For clothing textiles, heat transfer is a complicated transient process. Generated form the body, heat transfers through the air gap between skin and fabric, then through the fabric system, to the outer surface of the fabric system. At the outer-surface (OS), heat is lost into the environment by heat exchanging with the surrounding air. During this process, conduction, convection and radiation are all involved, may be to different extent, in determining the total heat loss. The establishment of a theoretical model to describe this process is important for the understanding of heat-transfer process and therefore important for the design of fabric of different requirement of heat-transfer properties. This present study aims to describe the whole process of heat transfer from the very beginning when fabric contacts with skin to the dynamic heat transfer balance, and attempts to set up a theoretical model which combines heat conduction through the textile assembly with convection and radiation as the boundary conditions at the outer surface of the assembly. Finite element (FE) simulation is used to solve the heat-transfer problem as defined earlier, where changes in the structural parameter of the fabric assembly and in ambient temperature during the heat-transfer process are taken into consideration. From the methodology point of view, this research also aims to identify and evaluate a new technique through the use of MARC Mentat, a commercial FE package, to characterize the heat-transfer features through the textile assembly, including details of changes in heat flux against time, the transit temperature distribution at the cross-section of the fabric assembly as well as how ambient temperature, size of air gaps, number of fabric layers and air layers between fabrics affect thermal properties of the system by using FE method (FEM). 2. Theories of heat transfer To understand the thermal properties of the textile system, it is necessary to assess the contributions of the various heat-transfer mechanisms that may be operative. These mechanisms are conduction, convection and thermal radiation for dry heat transfer. 2.1 Conduction Fibres and air intermingle together in any textile yarns and fabrics. In another word, the fabrics are neither homogeneous nor isotropic. However, with the preposition that the average heat-transfer properties of fabrics are to be measured and calculated through the

theoretical and practical work, it is reasonable to assume that a fabric is a homogeneous and isotropic material in heat transfer. In addition, since thickness dimension of a fabric is substantially smaller than the fabric width and length dimensions in normal clothing situations, it is also feasible to consider the heat transfer through a fabric is a one-dimensional problem. Under such assumptions, the transient heat-transfer process through the insulating material is described as (Yang and Tao, 1999):

›T l ›2 T ¼ · 2 ›t cr › x

163 ð1Þ

where: T ¼ temperature (8K); t

¼ time (s);

l

¼ conductivity (W m2 1 K2 1);

r

¼ mass density (kg m2 1);

c

¼ specific heat (W S kg2 1 K2 1); and

x

¼ direction of heat transfer.

2.2 Convection As one of the basic heat-transfer mechanisms, convection involves the transport of energy by means of the motion of the heat-transfer medium, in this case the air surrounding the human body. When cold air moves past a warm body, it sweeps away warm air adjacent to the body and replaces it with cold air. It has been found that there is no convection inside clothing insulation even with a very low density (Peirce and Rees, 1946). For this reason, this paper considers convective heat transfer only at the outer surface of the textile assembly. In the FE analysis, the convective heat transfer will be set as a boundary condition. The heat flux due to convection can be expressed as follows (Incropera and DeWitt, 2002): q ¼ hðT G 2 T 1 Þ

ð2Þ

where: q

¼ heat flux (W m2 2);

h

¼ film coefficient (W m2 2 K2 1);

TG ¼ out surface temperature of the fabric (8K); and T1 ¼ temperature of the ambient atmosphere (8K). 2.3 Radiation The heat loss carried out by radiation from a clad human body to the environment is a situation where the clad human body as the heat source is enveloped by the environment. In this case, the heat flux by radiation at the outer surface of the textile assembly is governed by the following equation (Incropera and DeWitt, 2002): 4 Þ q ¼ s · 1ðT G4 2 T 1

Study of heat transfer

ð3Þ

IJCST 22,2/3

where:

s ¼ Stefan-Boltzmann constant, which is 5.6703 £ 102 8 W m2 2 K2 4; and 1 ¼ emissivity of the surface.

164

2.4 Combined convective and radiative boundary condition There is a temperature difference between the outer surface of the textile assembly that clad the human body and the environment, and this temperature difference causes heat transfer between the textile assembly and the environment in the form of convection and radiation. In this research, convection and radiation together are taken as a boundary condition at the outer surface of the textile assembly, which couples together with conduction to determine the total heat-transfer process through the textile assembly. According to Fourier’s law and energy conservation, the relationship between the temperature gradient in the thickness direction of the textile assembly and combined convective and radiative heat transfer can be expressed by equation (4) (Farnworth, 1983; Yang and Tao, 1999; Incropera and DeWitt, 2002):

›T 4 ¼ hðT G 2 T 1 Þ þ s · 1ðT G4 2 T 1 Þ ð4Þ ›n where n indicates the distance in the thickness direction. Equations (1) and (4) are the theoretical background for dry heat transfer through textile assemblies. 2l

3. Mesh generation and boundary conditions 3.1 Mesh generation Mesh definition is the process of converting a physical problem into discrete geometric entities for the purpose of analysis (Chen, 2002; Rao, 1982). What is most important about heat-transfer process is the overall temperature distribution across the textile assembly and the heat flow through it. Under the assumption that the solid fibre and the air are mixed evenly together in a fabric, what is important for mesh generation is the precision of the calculation and the normality of the elements shape. In this research, quad elements are used during the FE simulation. 3.2 Boundary conditions In the case of heat transfer through a fabric assembly, convection and radiation contribute jointly to the heat loss at the outer side of the textile assembly and they are therefore used in setting up the boundary conditions for FE analysis. The film coefficient is an important parameter heat transfer by convection. According to Rapp, the film coefficient for nature convection is (McIntyre, 1980): h ¼ 4ðW m22 K21 Þ The emissivity of a surface describes how effective it is at radiating energy compared with a black body. For textile fabrics, it is reasonable to assume the emissivity 1 ¼ 0.9 (McIntyre, 1980). 4. Theoretical results by FEM In the simplest terms, the discipline of heat transfer is concerned with only two factors, i.e. temperature, and the flow of heat. Temperature represents the amount of thermal

energy available, whereas heat flow represents the movement of thermal energy from one location to another. When textile fabrics contact with the skin of human body, heat transfers through the textile assembly because of the temperature difference between skin and the outside environment. After sometime, the transfer process reaches a dynamic balance when the temperature difference and the heat flow through the textile system become constant. For such a situation, this research uses the balance heat flux to describe the heat-transfer property for the fabric assembly. At the same time, the transient maximum heat loss from the skin when the fabric first touches the human body is also studied towards the understanding of transient cool feeling when clothes is just put on. During the research, different ambient temperature, different layer arrangement and different air intervals between the fabric layers are all taken consideration.

Study of heat transfer

165

4.1 Parameters of the textile assembly Conduction heat transfer through the textile assembly is affected by the material properties of the fabric and by the construction of the textile assembly which contains fabric layers and air intervals. In this study, cotton fabrics are used as the fabric layer in the textile assembly. The material parameters for air and the cotton fabric are shown in Table I (Yao and Zhou, 1990; Rohsenow and Hartnett, 1973; Raznjevic, 1976). 4.2 Description of the fabric and air system Figure 1 shows schematic illustrations of models used for the textile assembly, where two types of models are involves. The first type is to simulate the tight-fit of a piece of clothes on the body where the fabric layer touches the skin. The other type of models assumes that there is a air gap between the inner-fabric layer and the skin. In all cases, between any two layers of fabrics in the textile assembly, there is a layer of air. 4.3 Theoretical results 4.3.1 Influence of size of air gap. The size of the air gap between human body and fabric is affected by the profile of the body, the body movement and the fitness of the clothes. It is assumed in this analysis that the air gap between the skin and the fabric is the constant for the location considered, though in practice this gap can never be of the same size. Different sizes of the air gap between skin and the inner fabric are simulated to see its effect on the heat-transfer properties. Figure 2 shows the influence of the air Mass density (kg m2 3)

Specific heat (W kg2 1 K2 1)

Conductivity (W m2 1 K2 1)

1.17 234

1,027 1,217

0.026 0.071

Air Cotton fabric

Table I. Parameters of air and the fabrics

A layer of fabric A layer of air Skin

Tight-fit

1 layer system

2 layers

3 layers

Figure 1. Illustration of the textile assembly models

IJCST 22,2/3 Outer-surface temperature y/°C

166

34 32 30 28 26 24 22 0

2

4 6 Air gap x/mm (a) OS temperature

8

10

0

2

4 6 Air gap x/mm (b) Balance heat flux

8

10

0

2

Balance heat flux y/W.m–2

120 100 80 60 40

Maximum heat flux from skin y/W.m–2

20

Figure 2. Influence of the size of air gap between skin and inner fabric

1,600 1,400 1,200 1,000 800 600 400 200 0 4 6 8 Air gap x/mm (c) Maximum heat flux from skin

10

gap on the OS temperature, balance heat flux, and the maximum heat flux from skin. The ambient temperature is 208C and the skin temperature is 33.58C. It is clear from Figure 2 that the air gap between skin and the inner fabric have significant influence on heat transfer. Balance heat flux is the energy flow form the body through the textile assembly to the environment when heat-transfer process achieves its balance situation. Balance heat flux has been used to describe the warmth-keeping property of clothes. When the size of air gap increases, balance heat flux is decreased most rapidly when air gap increase form 0 to 1 mm and the balance heat flux reduced 26 per cent per mm. As the air gap increases from 1 to 2 mm, the balance heat flux is reduced 20 per cent per mm. Further increase in the air gap shows a gradual slowing down of the balance heat flux. When air gap increases from 9 to 10 mm, the reduction in the balance heat flux becomes a mere 7.8 per cent per mm. A similar pattern is shown for the OS temperature against the change in the size of air gap. The maximum heat flux from the skin relates to the transient cool feeling when putting on a piece of clothes. The maximum heat flux is reduced by 80 per cent when air gap is increased form 0 to 1 mm. When the air gap changes from 3 to 10 mm, the reduction rate of the maximum heat flux changes from 10 to 2 per cent per mm. All these suggest that an air gap between the skin and the inner fabric helps the warmth-keeping as well as reducing the transient cool feeling. So, it is necessary to create an air gap between the skin and the underclothes. Increasing the roughness of the fabric is one way for air gap creation. 4.3.2 Influence of fabric layers. Layered clothes are considered in this analysis. The assumption adopted here is that the size of air gap between the inner-fabric layer and the skin and that between any adjacent layers are both 1 mm, though in reality, air gaps are never neatly defined as this. The four models of textile assembly shown in Figure 1 are used. The models are subject to different environmental or ambient temperatures, ranging from 10 to 308C. In Figure 3, the “Fabric” curves refer to the tight-fit model where a single-layer of fabric touches the skin tightly and there is no air gap in between. The results reveals, as shown in Figure 3(a), that in all cases the OS temperature drops as the number of layers of the textile assembly increases, indicating that addition of more layers of fabrics with air gaps in the assembly is an effective measure to reduce heat transfer. It is more effective when the environment temperature is low. It also shows that the environment temperature affects the OS temperature positively proportionally. This is an indication that the increase in environment temperature reduces the temperature difference and hence reduces the heat transfer from the body through the textile assembly to the OS of the textile assembly. This is supported by the curves shown in Figure 3(b) where the balance heat flux decreases as the environment temperature goes up. It is of interest to see that when the environment temperature increases to a certain value all curves in Figure 3 will eventually meet at a single point, respectively. It can be speculated from Figure 3(a) that this temperature is actually the skin temperature of the human body, which is 33.58C. Balance heat flux through the textile assembly and the maximum heat flux from the skin at this point both become zero as can be seen in Figure 3(b) and (c). These would suggest that at such an environment temperature, the composition of the textile assembly will not cause any difference in heat transfer. At such a situation, the heat source, i.e. the human body, and the textile assembly have become one new heat source. This of course,

Study of heat transfer

167

Outer-surface temperature y/°C

IJCST 22,2/3

168

33 32 31 30 29 28 27 26 25 24 23 22 21

Fabric 1Layer 2Layers 3Layers

10 12 14 16 18 20 22 24 26 28 30 32 Environment temperature x/°C

Maximum heat flux from skin y/W.m–2

Balance heat flux y/W.m–2

(a) OS temperature

Figure 3. Influence of fabrics layer at different environment temperature

200 Fabric 180 1Layer 2Layers 160 3Layers 140 120 100 80 60 40 20 0 10 12 14 16 18 20 22 24 26 28 30 32 Environment temperature x/°C (b) Balance heat flux

2,500 2,000

Fabric 1Layer 2Layers 3Layers

1,500 1,000 500 0 10 12 14 16 18 20 22 24 26 28 30 32 Environment temperature x/°C (c) Maximum heat flux from skin

is a situation that is not easily achieved in practice as this balance can be interfered by many factors such as the body movement and the constantly changing ambient condition. It is also seen in Figure 3(c) that the maximum heat flux from skin reduces more rapidly for the tight-fit model than for the three-layered models whose curves overlap in Figure 3(c). For all models, the increase in ambient temperature reduces the transient cool feeling of the textile assembly and the number of fabric layers in the textile assembly becomes insignificant. 4.3.3 Temperature and heat flux distribution with the one-layer model. It is of interest to see how temperature and heat flux propagate from the heat source through the textile assembly. In this study, the one-layer fabric model is used with the air gap between the skin and the fabric being 10 mm. The ambient temperature is set to be 208C. The temperature distribution across the 10 mm air gap and at the inner-surface and OS of the fabric against time is shown in Figure 4. It is evident that at any given time the temperature gradient exists, with higher temperature closer to the skin and vice versa. The second phenomenon is that the temperature of air closer to the skin increases much more quickly than that further form the skin. For instance, the air temperature 2 mm away from skin gets closer to its balance in a matter of seconds, whereas the air temperature 8 mm away from the skin takes about 50 s to reach its balance. Figure 5 shows the heat flux distribution against time. On the skin when the air gap between the skin and fabric is 0 mm, the heat flux peaks within the first few seconds creating a sharp cool sensation after the clothes are just put on. The heat flux from the skin becomes smaller when the fabric is further away from the skin. The dynamic process for heat flux dies down eventually to reach the heat flux balance. It is not a surprise to see that the OS of the fabric responded most slowly to reach the heat flux balance.

Study of heat transfer

169

5. Experimental analysis To test the theoretical predictions, the OS temperature of the fabric was measured under different ambient temperature, variable fabric layers, and different sizes of air 32 2 mm 4 mm 6 mm 8 mm I.S O.S

31 30

Temperature y/°C

29 28 27 26 25 24 23 22 21 20 0

50

100 Time x/S

150

200

250

Figure 4. Temperature distribution

Heat flux y/W.m–2

IJCST 22,2/3

170

Figure 5. Heat flux distribution

75 70 65 60 55 50 45 40 35 30 25 20 15 10 5 0

0 mm 2 mm 4 mm 6 mm 8 mm I.S O.S

0

10

20

30 Time x/S

40

50

60

gap between the inner layer of fabric and the skin. The testing apparatus mainly consists of two parts, a heating plate providing a fixed temperature of 33.58C to stimulate the skin, and a data acquisition system with a temperature sensor to test and record the OS temperatures of the textile assembly. All tests were carried out in an artificial climatic chamber where the relative humidity was 65 per cent and the temperatures were made changeable from 12 to 308C. Figure 6 shows the composition of the testing apparatus, where 1 is the temperature sensor, 2 is the outer-fabric layer, and 3 is the textile assembly holder. Table II enlists the comparison of OS temperatures under three different situations between theoretical and experimental approaches. When examining the influence of the ambient temperature, a single layer fabric was used where the air gap between the fabric and the skin was 1 mm. In the case of fabric layers, while the ambient temperature was set to be 208C, the air gap between adjacent fabric layers and between the inner-fabric layer and the skin was chosen to be 1 mm. When considering the

1

RS232 RS232

89C52

D/A

Heating control

PC Heating plate

Figure 6. Illustration of the testing apparatus

Temperature measurement

A/D

PID

2

3

influence of the air gap, the single layer fabric model was used again with ambient temperature setting to 208C. Both experimental and theoretical data indicate that the ambient temperature affect the heat transfer from the skin to the OS of the clothing fabric. At a lower ambient temperature, the OS temperature is lower than it is subjected to a higher ambient temperature, leading to a larger temperature gradient which facilitates higher rate of heat transfer. In terms of the number of layers of fabrics involved in the textile assembly, addition of fabric, and air gap, layers provides better insulation against heat loss. Within a reasonable range, increase in air gap size leads to reduced OS temperature, indicating less heat loss. Devising a fabric assembly with controllable air gap sizes would enable the same piece of garment be used under different climatic environment. It is evident from Table II that the theoretical and experimental data agree well in all cases. This gives confidence in using the general purpose software MARC Mentat, for analysis on heat transfer for clothing comfort applications. This is of significance as it reduces the necessity of developing specialised software tool for clothing comfort. OS temperature was also measured against time. In this experiment, the single layer fabric model with 10-mm air gap between the fabric and the skin was used, and the ambient temperature was set to 208C. Figure 7 shows the change of OS temperature

Approach

Ambient temperature (one-layer fabric) 158C 258C

Theory Experiment

27.1 26.9

Fabric layers (ambient temperature 208C) 1 2 3

30.1 29.7

29.4 28.9

27.2 27.0

Study of heat transfer

171

Air gap (ambient temperature 208C) 5 mm 10 mm

25.8 26.0

24.7 24.9

22.9 23.1

Table II. OS temperature (8C)

23.5

Outer-surface temperature y/°C

23.0 22.5 22.0

Experimental result Theoretical result

21.5 21.0 20.5 20.0 0

50

100

150

200 Time x/S

250

300

350

400

Figure 7. OS temperature changes against time

IJCST 22,2/3

172

against time, obtained both experimentally and theoretically. It can be seen that a similar trend was demonstrated between the theoretical and experimental results, with the theoretical curve showing sharper temperature increase within the first 50 s. The reason for this is likely that the temperature sensor takes time to respond to the temperature changes. In both cases, the OS temperature stabilises between 22.8 and 23.08C. The experimental curve also demonstrated a fluctuation because the heating plate was set to work intermittently in simulating the skin temperature. 6. Conclusions Heat transfer through a layered textile assembly is studied using the FE approach. After analysing the nature of heat transfer from a human body through a textile assembly, the study presented a theoretical model that takes all three forms of heat transfer into consideration. While the heat conduction was regarded as the main mechanism for heat transfer, convection and radiation were involved as boundary conditions in solving the problem. On the other hand, four models of textile assemblies were created in the study, including the tight-fit, and three-layered models which allow an air gap between adjacent fabric layers and between the skin and the inner-layer fabric. Based on these models, simulations were carried out using a general purpose commercial FE package, MARC Mentat, and the relationships between structural parameter of the textile assembly and the thermal properties have been established. The establishment of theoretical models and the use of FE tool have enabled quantitative description of heat transfer with accuracy as verified by the experimental results. The size of the air gaps has a significant influence on the heat transfer. The balance heat flux drops by 40 per cent when the air gap increases from 2 to 10 mm. The influence of the air gap tends to become smaller as the air gap is further increased. The number of fabric layers in the textile assembly has a noted influence, more so when the ambient temperature is lower. Another aspect of work that has been reported is the heat flux from the skin when the skin touches the fabric under different situations, leading to further understanding of transient cooling sensation. This paper also reported on the setting up of a simple apparatus for testing surface temperature of fabrics, which is capable in dealing textile assemblies with different sizes of air gaps. It has shown that the theoretical results agree significantly well with the experimental results. This has demonstrated a new route for thermal analysis relating to clothing comfort. References Chen, H.H. (2002), MARC Finite Element Course, China Machine Press, Beijing. Farnworth, B. (1983), “Mechanisms of heat transfer through clothing insulation”, Textile Research Journal, Vol. 53, pp. 717-25. Fricke, H. (1924), “A mathematical treatment of the electric conductivity of disperse systems: the electric conductivity of a suspension of homogeneous spheroids”, Phys. Rev., Vol. 24. Fricke, J., Buttner, D., Caps, R., Gross, J. and Nilsson, O. (1990), “Solid conductivity of loaded fibrous insulation, insulation materials, testing, and applications”, in McElroy, D.L. and Kimpflen, J.F. (Eds), ASTM STP 1030, American Society for Testing and Materials, Philadelphia, PA, pp. 66-78. Incropera, F.P. and DeWitt, D.P. (2002), Fundamentals of Heat Transfer, 5th ed., Wiley, Somerset, NJ.

McIntyre, D.A. (1980), Indoor Climate, Applied Science, London. Mohammade, M. and Banks-Lee, P. (2003a), “Determining effective thermal conductivity of multilayered nonwoven fabric”, Textile Research Journal, Vol. 73, pp. 802-8. Mohammade, M. and Banks-Lee, P. (2003b), “Determining radiative heat transfer through heterogeneous multilayer nonwoven materials”, Textile Research Journal, Vol. 73, pp. 896-900. Peirce, F.T. and Rees, W.H. (1946), “The transmission of heat through textile fabrics, Part II”, Journal of the Textile Institute, Vol. 37, pp. T181-T204. Rao, S.S. (1982), The Finite Element Method in Engineering, Pergamon Press, Oxford. Raznjevic, K. (1976), Handbook of Thermodynamic Tables and Charts, Hemisphere, Washington, DC. Rohsenow, W.M. and Hartnett, J.P. (1973), Handbook of Heat Transfer, McGRAQ-HALL Book, New York, NY. Yang, S.M. and Tao, W.Q. (1999), Heat Transfer, Higher Education Press, Beijing. Yao, M. and Zhou, J.F. (1990), Textile Materials, China Textile Press, Beijing. Corresponding author Xiaogang Chen can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

Study of heat transfer

173

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

IJCST 22,2/3

174 Received 10 November 2008 Accepted 24 April 2009

Comparative analysis of hand properties and compositions of trace elements in linen fabrics produced in different regions Takako Inoue, Kengo Ishihara and Kyoden Yasumoto School of Life Studies, Sugiyama Jogakuen University, Nagoya, Japan, and

Masako Niwa Nara Women’s University, Nara, Japan Abstract Purpose – The purpose of this paper is to examine ladies’ linen fabrics produced in different regions – Japan, Italy, and Poland – to ascertain differences in mechanical, thermal, and air permeability properties. Design/methodology/approach – The paper investigates mechanical properties, air permeability, and thermal conductivity. The silhouettes of Polish, Italian, and Japanese linen fabrics are different. The thermal conductivities of the Polish linen fabrics are high. The levels of 72 elements were analyzed and remarkable differences were observed in the levels of 16 elements, including Li, Al, Si, Ti, Cr, Ni, Rb, and Y, Ag, among Polish, Italian linen fabrics, and linen fabrics made in Japan. Another ten elements were detected at some level in either the samples of Polish linen fabrics or linen fabrics made in Japan. Findings – There are differences among the Polish, Italian, and linen fabrics made in Japan, but the differences are not remarkable. Research limitations/implications – This paper is a wide world regional study of linen characterisation. Practical implications – Another ten elements are detected at some level in either the samples of Polish linen fabrics or linen fabrics made in Japan. There are differences among the Polish, Italian, and linen fabrics made in Japan, but the differences are not remarkable. Originality/value – The paper presents useful Measurement instrumentation, analysis and characterisation of linen fabrics from different regions of the world. Keywords Textiles, Mechanical properties of materials, Permeability, Thermal conductivity Paper type Research paper

International Journal of Clothing Science and Technology Vol. 22 No. 2/3, 2010 pp. 174-186 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011018649

Introduction We had clarified that the Polish linen fabrics and linen fabrics made in Japan (typical Hari type ladies’ garment fabrics produced in different districts) differed in terms of mechanical properties and thermal conductivity (Kawabata et al., 2000). Linen is an annual grass of the flax family, produced primarily in comparatively cold areas including northern France, Belgium, Russia, East European countries, and China (Japan Linen, Ramie & Jute Spinners’ Association, 2010). Early in the Showa era, The authors would like to thank Akira Ishikawa of Unix, Inc. for providing the linen fabrics made in Japan.

Russia was the country with the most flax-producing land, followed by Latvia, and then Poland ( Mori, 1949). Linen was imported to Japan from Belgium, China and others as fiber, from Italy, Poland and others as single yarn, and from China, Italy and others as woven fabric (Japan Linen, Ramie & Jute Spinners’ Association, 2010). We investigated the mechanical properties, hand evaluation, thermal conductivity, air permeability and the contents of trace elements of 100 percent linen fabrics. The samples used were 19 fabrics provided by the Polish National Research Laboratory, as well as an additional 16 fabrics made in Japan, and three fabrics made in Italy but collected in Japan. There is a paper reporting the results of a study, which lasted almost 30 years, of effects of climatic and soil conditions in Poland on five varieties of fiber flax (Stanislaw et al., 1997). In this paper, the effects of environmental conditions on the growth, development and yield of the following varieties of fiber flax were estimated: Fortuna, Minerwa, Svapo, Waza, and Nike. The environmental parameters were the soil composition, the type and pH of the soil, the climatic conditions, the time of agronomic operations, and the agronomy level. In this paper, we researched the differences in the properties of 100 percent ladies’ linen fabrics which were produced in the different districts. The levels of trace elements in these fabrics were analyzed to investigate whether significant correlations exist among trace element levels and the mechanical properties of 100 percent linen fabrics which are used to make suits, jackets, trousers, skirts, blouses, one-piece dresses, etc. Experiments Samples In total, 19 sample fabrics made in Poland from fiber grown in Poland were provided by the Polish National Fiber Research Laboratory. The sample fabrics made in Japan were eight fabrics provided by company A, four fabrics by company B, and four fabrics by another company. The growing districts for the linen fabrics from company A were Ireland and North France; the yarn was processed in Italy and woven, dyed, and finished in Japan. The growing districts for the linen fabrics from company B are not known, but the fabrics were woven, dyed, and finished in Japan in the same way as those from company A. The three sample fabrics made in Italy were woven, dyed and finished in Italy. Fabric analysis The mechanical properties of all of the linen sample fabrics were measured with the KES-FB-AUTO system (Kawabata et al., 1990). The tensile properties and compression properties were measured under standard conditions and high-sensitivity conditions applied to thin fabrics (Matsudaira et al., 1984). The measurement items and conditions are shown in the Appendix. The air resistance, RA(kPa s/m) of the sample fabrics was measured with a KES-F8 air permeability tester and the thermal conductivities, l(W/mK), of the sample fabrics were measured with a KES-F7 thermo labo. Trace element levels in fabrics The sample fabrics were weighed into a quartz digestion reaction container. After nitric acid (Wako Pure Chemical Industries, Ltd, Osaka, Japan) was added for trace element measurement, wet digestion was performed using microwave digestion apparatus (Multiwave and Perkin Elmer). The sample fabrics were weighed into a quartz digestion reaction container. After nitric acid (Wako Pure Chemical Industries, Ltd, Osaka, Japan)

Trace elements in linen fabrics

175

IJCST 22,2/3

176

was added for trace element measurement, wet digestion was performed using microwave digestion apparatus (Multiwave and Perkin Elmer). The clear solution was transferred to 50 ml polypropylene volumetric tubes, resulting in a 1:10 dilution; 1,000 g of a 0.1 mg/g yttrium (Y) standard solution in 5 percent HNO3 was added as an internal standard to create a concentration of 10 ng/g. Trace element concentrations were determined by an ICP-DRC-MS. The mass spectrometer settings and plasma conditions were optimized with a solution of 10 mg indium/l and the instrument operating conditions were as follows: ratio-frequency power, 1,100 W; plasma gas flow, 15 l/min; auxiliary gas flow, 1.2 l/min; nebulizer gas flow, 0.91 l/min. Data collection was performed from mass numbers 6-15, 19-39, 42-210 and 230-240. The signal intensities in each mass number of the element were compared to the signal intensities of 10 mg/ml of Mg, Rh and Pb, and the concentrations of each element were calculated using the Total Quant Method mode (ELAN Ver. 2.4, PerkinElmer Sciex Instruments). Results and discussion Mechanical properties The global chart (Inoue and Niwa, 2003) shown in Figure 1 was prepared to identify the region of the mechanical properties of ladies’ garment fabrics for different intended uses. The scale of the horizontal axis is normalized by the mean value and the standard deviation of the mechanical property of the 280 global ladies’ garment fabrics (Inoue and Niwa, 2003). The mean values and standard deviations are shown for the Polish linen fabrics and linen fabrics made in Japan. The mechanical properties of Italian linen fabrics were plotted individually. The mechanical properties of Polish linen fabrics showed high bending properties B, 2HB, low EM, LT, WT, and high tensile resilience RT. The shear stiffness G, shear hysteresis 2HG and 2HG5 of the Polish linen fabrics were low, and those of the Italian linen fabrics were the lowest. Italian linen fabrics have a low shearing stiffness. As for the surface properties, Polish linen fabrics have high SMD, but MIU was low and the surface was smooth. Of the samples used in this study, the thickness and weight of the Polish linen fabrics were higher than those of the linen fabrics made in Japan. Silhouette and hand evaluation Figure 2 shows a discrimination graph for the optimum silhouette design of ladies’ linen fabrics. The fabrics were discriminated using the mechanical properties of the three optimum silhouette designs (Niwa et al., 1998) – Drape-type has a beautiful drape silhouette, Hari-type has an anti-Drape silhouette which forms a suitable space between the body and clothing, and tailored-type has a silhouette which covers the body with a beautiful, dimensional shape. Niwa and others obtained a silhouette discrimination equation (Niwa et al., 1998) which discriminates optimum silhouette design using tensile properties, bending properties, shearing properties and weight per unit area of fabric. We used this equation. Ladies’ linen fabrics are located broadly within the Hari and tailored silhouettes. The Japanese company B linen fabrics falling at the the center of Hari-type to the outside of Hari-type show strong Hari-type characteristics. The other four fabrics made in Japan have Hari-type silhouettes only. The primary hand values KOSHI, SHARI, HARI, and FUKURAMI of ladies’ linen fabrics were calculated from the mechanical properties of linen fabrics using the KN301S equation (Kawabata, 1980), and then the mean values were plotted on a criteria chart for

Tensile

–3σ

–2σ

EM1

0.2

0.3

0.4 0.5

EM2

0.2

0.3

0.5

EM

0.2

LT

0.3

0.4

0.4

WT

0.7

0.005

2HB1

0.005 0.005

0.01

0.005

0.005

20

0.9

0.2

1.0

0.01

1.1 3

0.5 1

0.5 0.1

0.05

4

1

0.5

0.1

2 2

3

1

0.5

0.1

0.05

177

100 0.3

0.1

0.05

50 20

2

0.05

0.01

30

10

5

0.1

0.01

0.001

B

20

80

0.05

0.001

2HB2

4

60

0.01

B2

10

5

0.5

40

B1

10

5

0.8

0.1

RT

Bending

0.6

4 3

2

1

4

Trace elements in linen fabrics

(N = 280) 3σ

2σ

3 3

2

1

0.5

σ

2

1

0.5

0.05

(X-M)/s 0

–σ

2

1

0.5

1

2HB Shear

0.001

G

0.05

2HG

0.01

0.1

0.05

0.1 0.05 0.05

0.1

1 4

1

0.5

0.1

0.5

1

0.5

0.01

2HG5 Compr.

0.005

5 5

0.5

10 10

50

LC 0.4

0.6

0.8

1.0

WC 0.01

0.005

0.5

0.1

0.5

60

80

1

RC 20

Surface

MIU

40

0.05

MMD

0.1

0.2

0.25

0.01

0.005

SMD

0.15

0.5

0.3 0.05

1

5

0.35 0.1

10

T 0.05

0.1

0.5

5

W 2

3

4

5

10

20

30

40

50

Notes: This chart is normalized by the ladies garments fabrics population; suffix 1: warp direction; suffix 2: weft direction; : Japanese linen fabrics including A and B companies fabrics (n = 16); Polish linen fabrics (n = 19), mean value and : ± standard deviation

100

:

ideal men’s suiting (mid-Summer) (Kawabata et al., 2002) in Condition 1 of Figure 3. The KOSHI, SHARI and HARI of Polish linen fabrics and Japanese company B linen fabrics are characteristically high. The mechanical parameters and three basic components of tailorability which are concerned with total appearance value (TAV) as men’s suiting are plotted in Condition 2 of Figure 3. The TAVs of Polish linen fabrics, Japanese company B linen fabrics and Italian linen fabrics were close to the highest value

Figure 1. The mechanical properties of linen fabrics

IJCST 22,2/3

4 Tailored

178

2

Z2

s

0

s

s

–2

2s Drape

2s

Hari

–4

Figure 2. Discrimination graph for the optimum silhouette design of linen fabrics

2s

–4

–2

0 Z1

2

4

Notes: : Polish linen fabrics (n = 19); : Japanese company A linen fabrics (n = 8); ∆: Japanese company B linen fabrics (n = 4); : Japanese linen fabrics excluding A and B companies fabrics (n = 4); : Italian linen fabrics (n = 3)

of 5, so suitable silhouette formability and a high-quality finish suit can be predicted. The Z1Z2 discrimination parameters of the linen fabrics were also calculated using the mechanical properties of the three categories (Kawabata and Niwa, 1991) – cotton-, silk-, and wool-like. The results are shown in Figure 4. The three zones show each of the three categories. Compared with all of the linen fabrics, these are located between “Wool Like” and “Cotton like”. The fabrics other than Polish linen fabrics are located on the cotton-like side, and Polish linen fabrics are located near the wool-like side. Thermal conductivities and air permeability Thermal conductivities and air permeability are shown in Figure 5. The thermal conductivities of Polish linen fabrics are high, so these fabrics are found to be suitable as mid-Summer clothing. The relationship between air resistance and porosity is shown in Figure 6. The relationship between thermal conductivity and porosity is shown in Figure 7. The air resistance decreases as porosity increases, and a tendency for low thermal conductivity was also recognized. However, a tendency related to differences in production regions was not seen. Compositions of trace elements and production districts The trace element levels in Polish linen fabrics and other fabric samples acquired in Japan are shown in Table I. Of the 16 elements for which significant trace levels were

–3σ

Condition 1: Hand –σ σ 0

–2σ

Hand value Stiffness (Koshi) Anti-drape (Hari) Crispness (Shari) Fullness (Fukurami) 1 THV

1

2

1 0

3

4

2

3

4

1

2

3

2

3

1

Trace elements in linen fabrics

5

6

6

7

5 4

5

4

6

5

2

3σ

7

8

8 7

9

9

8

6

3

2σ

9

10

7

8

4

179

10

9

5

Condition 2: Suit appearance Mechanical parameters –3σ –2σ EL2 BS2 SS BP SP (B/W)1/3 (G/W)1/3

2 0.02

3

4

σ

0

5

0.1

0.2

3

4

1.0

5

0.3

2.0

3.0

4.0

0.4 0.5

10

1.5 4.0

40

5.0

1.0 20

2.0

3.0

30

0.3 0.4 0.5

2.0 0.2

3σ

20

0.1

1.0

0.05

2σ

10

0.03 0.04 0.05

0.4 0.5

2

–σ

30

40 50

2.5 5.0

6.0

7.0

Three basic components of tailorability –3σ Formability Elasitic potential Drape TAV

–2σ

0 0 0

–σ

0

2σ

2

4

2

4

2 0

σ

2

4

3σ 6 6 6

4

Notes: This chart is normalized by the men’s suiting population; the perfect property zone is shown by the shaded zone; : Polish linen fabrics (n = 19); . : Japanese . Japanese company B linen fabrics (n = 4); company A linen fabrics (n = 8); ∆: . : Japanese linen fabrics excluding A and B companies fabrics (n = 4); . : Italian linen fabrics (n = 3), mean values are shown

Figure 3. Hand value and suit appearance of linen fabrics plotted on a criteria chart for ideal suiting (mid-Summer)

IJCST 22,2/3

3

2

180

Silk Wool 2s

1

s

Z2

s

2s

0

–1

s

s

–2 s Cotton

–3

2s –3

–2

–1

0

1

2

3

Z1

Figure 4. Wool silk cotton graph for linen fabrics

Notes: : Polish linen fabrics (n = 19); : Japanese company A linen fabrics (n = 8); ∆: Japanese company B linen fabrics (n = 4); : Japanese linen fabrics excluding A and B companies fabrics (n = 4); : Italian linen fabrics (n = 3)

0.11

0.10

l (W/mK)

0.09

0.08

0.07

0.06

0.05

0.04 5 6 7 8 9 0.01

Figure 5. Thermal conductivity l and air resistance AR

2

3

4 5 6 7 89 0.1 AR (kPa·sec/m)

2

3

Notes: : Polish linen fabrics (n = 19); : Japanese company A linen fabrics (n = 8); ∆: Japanese company B linen fabrics (n = 4); : Japanese linen fabrics excluding A and B companies fabrics (n = 4); : Italian linen fabrics (n = 3)

Trace elements in linen fabrics

3 2

0.1 AR (kPa·sec/m)

7 6 5 4 3

181

2

0.01 7 6 5 75

80

85 Porosity* (%)

90

Notes: : Polish linen fabrics (n = 19); : Japanese company A linen fabrics (n = 8); ∆: Japanese company B linen fabrics (n = 4); : Japanese linen fabrics excluding A and B companies fabrics (n = 4); : Italian linen fabrics (n = 3); *: thickness at 0.5 g/cm2

Figure 6. Air resistance AR and porosity

0.11 0.10

l (W/mK)

0.09 0.08 0.07 0.06 0.05 0.04 75

80

85 Porosity* (%)

90

95

Notes: : Polish linen fabrics (n = 19); : Japanese company A linen fabrics (n = 8); ∆: Japanese company B linen fabrics (n = 4); : Japanese linen fabrics excluding A and B companies fabrics (n = 4); : Italian linen fabrics (n = 3); *: thickness at 0.5 g/cm2

seen, three of the elements, Li, Cr, Cu, were present at significantly higher levels in the Italian fabrics than in the fabrics produced in the other two districts. The levels of 11 elements (Al, Ti, Ni, As, Rb, Y, Mo, Ag, Cd, Ba, and La) were significantly higher in the Polish linen fabrics. The levels of only two elements (Si and Cl) were

Figure 7. Thermal conductivity l and porosity

IJCST 22,2/3 Elements

182

Table I. Comparison of element quantities

Li Be B Mg Al Si P S Ci K Ca Sc Ti V Cr Mn Fe Co Ni Cu Zn Ga Ge As Se Br Rb Sr Y Zr Nb Mo Ru Rh Pd Ag Cd In Sn Sb Te I Cs Ba La Ce Pr

Poland (n ¼ 19) Mean(ppb) SD 1.153 0.018 9.525 488.079 128.531 37.929 0 1,437.114 54.223 38.410 180.519 2.287 8.747 0.027 0.582 3.562 13.718 0.281 0.848 1.423 5.311 0.080 0.034 0.023 0.087 0.726 0.095 3.214 0.019 0.089 0.005 0.010 0.003 0.007 0.040 0.038 0.076 0.001 0.294 0.003 Tr 0.001 0.001 0.269 0.002 0.004 0.001

2.198a 0.024 5.513 327.969 61.298b 160.918a 0 6,097.157 155.764a 30.026 448.763 3.975 5.161b 0.019 0.474a 5.152 12.470 0.436 0.627b 1.330a 3.731 0.170 0.055 0.028b 0.142 0.336 0.079b 2.809 0.014b 0.088 0.004 0.008b 0.005 0.012 0.052 0.029b 0.106b 0.002 0.455 0.002 0.000 0.001 0.201b 0.001b 0.003 0.000

Origin Japanese market (n ¼ 16) Mean(ppb) SD 0.306 0.004 9.903 334.243 52.202 283.148 0 8,142.128 232.737 24.392 486.535 4.369 2.377 0.022 0.328 0.710 12.500 0.041 0.240 1.392 3.829 0.007 0.001 0.003 0.067 1.660 0.023 1.414 0.007 0.063 0.003 0.004 0 Tr 0.014 0.011 0.004 0.001 0.723 0.002 0 0.001 0 0.094 0.001 0.002 Tr

0.226a 0.005 13.313 232.221 39.882a 268.307b 0 10,174.247 220.991b 16.398 581.508 5.207 1.864a 0.016 0.384a 0.622 7.115 0.080 0.185a 3.545a 6.487 0.007 0.001 0.002a 0.106 4.688 0.018a 1.642 0.008a 0.076 0.003 0.003a 0 0.017 0.014a 0.044a 0.001 2.102 0.002 0 0.001 0 0.185a 0.001a 0.002

Italy (n ¼ 3) Mean(ppb) SD 5.868 0.006 8.831 326.155 45.865 91.334 0 10,874.434 257.739 39.578 904.343 7.541 3.330 0.019 4.526 0.502 12.503 0.557 0.249 6.945 1.329 0.006 0.002 0.009 0.134 0.366 0.024 1.693 0.006 0.148 0.003 0.004 0 0.001 0.019 0.045 0.005 0 0.275 0.003 0 Tr 0 0.160 0.001 0.003 0

Anova

7.273b 0.004 2.413 324.104 8.348a 83.505ab 0 6,833.870 146.859 ab 4.419 1,203.097 5.333 0.926ab 0.004 4.068b 0.423 2.760 0.682 0.087 ab 8.069b 0.277 0.001 0.000 0.006 ab 0.095 0.106 0.004 ab 1.680 0.001 ab 0.030 0.000 0.002 ab 0 0.001 0.002 0.030 ab 0.003 ab 0 0.0327 0.001 0 0 0.086 0.000 0.000 0

ab ab

**

** **

*

** **

** *

*

** **

*

** *

* *

(continued)

Elements Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Re Os Ir Pt Au Hg Ti Pb Bi Th U

Poland (n ¼ 19) Mean(ppb) SD 0.001 0 Tr 0.003 Tr 0.002 0 Tr 0 0.001 0 Tr 0 0.001 0 0 Tr Tr 0.002 0.001 Tr 1.079 0.287 0.016 0.045

0.001 0 0.004 0.003 0 0 0.001 0 0 0.001 0 0 0.002 0.001 1.296 0.238 0.012 0.037

Origin Japanese market (n ¼ 16) Mean(ppb) SD 0.001 Tr 0 Tr 0 Tr 0 0 0 0 0 Tr 0 Tr 0 0 0 Tr Tr 0.001 0 0.850 0.033 0.009 0.041

0.001 0 0 0 0 0 0 0 0 0 0 0 0.001 0 1.617 0.068 0.006 0.071

Italy (n ¼ 3) Mean(ppb) SD 0.001 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.001 0.001 0 0.309 0.056 0.008 0.084

Trace elements in linen fabrics Anova

0.000 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.000 0.000 0 0.058 0.007 0.002 0.090

Notes: Means and SD values are significantly different *p , 0.05 and * *p , 0.01; Tr, a very small quantity of the element was detected; values without common superscripts are significantly different

significantly higher in the linen fabrics made in Japan. Several elements (Ru, Te, Cs, Eu, Tb, Er, Yb, Ir, and Tl) were detected in Polish linen fabrics but not in the linen fabrics made in Japan, and only one element (Sm) was detected in the linen fabrics made in Japan but not in the Polish linen fabrics. Looking at the eight elements (Mg, Al, Si, S, Cl, K, Ca, and Fe) for which the levels exceeded 10 ppb, the levels of four of these elements (Mg, Al, K, and Fe) were higher in Polish linen fabrics than in the linen fabrics made in Japan, and the levels of the other elements (Si, S, Cl, and Ca) were higher in the linen fabrics made in Japan than in Polish linen fabrics. These results indicate that the Polish linen fabrics contained a larger variety of elements at extremely low concentrations compared to the linen fabrics made in Japan (Figure 8). Conclusions . The mechanical properties of Polish, Italian, and Japanese linen fabrics are different. The silhouettes formed by these fabrics are also different. . The primary hand values calculated from the mechanical properties of linen fabrics, KOSHI, SHARI, and HARI, are high, and the mean value of formability,

183

Table I.

IJCST 22,2/3

0.11 0.10 0.09 l (W/mK)

184

0.08 0.07 0.06 0.05 0.04 0

5

10

15 Sulfur (ppb)

20

25

30 ×103

Notes: : Polish linen fabrics (n = 19); : Japanese company A linen fabrics (n = 8); ∆: Japanese company B linen fabrics (n = 4); : Japanese linen fabrics excluding A and B companies fabrics (n = 4); : Italian linen fabrics (n = 3)

Figure 8. Thermal conductivity l and the quantity of sulfur

.

.

elastic potential and drape of suit appearance show suitable values. Polish, Italian, and Japanese company B linen fabrics can be predicted as yielding suitable silhouette formability and a high-quality finish suit. The thermal conductivities of the Polish linen fabrics are high at approximately 12 percent, so these fabrics are found to be suitable as mid-Summer clothing. The elements found in the fabrics, as well as the amounts of trace elements, differed depending on the production region.

References Inoue, T. and Niwa, M. (2003), “Objective evaluation of the quality of ladies’ garment fabrics ( paper translated from the Japanese ed.)”, J. Text. Eng., Vol. 49 No. 2, pp. 33-45. Japan Linen, Ramie & Jute Spinners’ Association (2010), National Attached Table Classified by Imported Statistical Items, Japan Linen, Ramie & Jute Spinners’ Association, Tokyo, pp. 1-15, available at: www.asabo.com Kawabata, S. (1980), “The standardization and analysis of hand evaluation”, The Hand Evaluation and Standardization Committee, 2nd ed., The Textile Machinery Society, Osaka. Kawabata, S. and Niwa, M. (1991), “Recent progress in the objective measurement of fabric hand”, International Conference of Textile Science ‘91, Technical University of Liberec, Czechoslovakia, Vol. 1, pp. 12-19. Kawabata, S., Niwa, M. and Yamashita, Y. (2002), “Recent developments in the evaluation technology of fiber and textiles: toward the engineered design of textile performance”, J. App. Polym. Sci., Vol. 83, pp. 687-702.

Kawabata, S., Niwa, M., Ito, K. and Nitta, M. (1990), “Application of objective measurement to clothing manufacture”, Int. J. Cloth. Sci. Tech., Vol. 2 Nos 3/4, pp. 18-33. Kawabata, S., Niwa, M., Koztowski, R., Manys, S., Nakano, K. and Inoue, T. (2000), “Fabric hand property of Polish linen fabrics for ladies’ outer wear”, Int. J. Cloth. Sci. Tech., Vol. 12 No. 3, pp. 193-204. Matsudaira, M., Kawabata, S. and Niwa, M. (1984), “Measurements of mechanical properties of thin dress fabrics for hand evaluation”, J. Text. Mach. Soc. Japan (predecessor Journal of J. Text. Eng.), Vol. 37, pp. T49-T57. Mori, S. (1949), Diamond Industrial Complete Book (10) of Linen Production, Diamond, Tokyo. Niwa, M., Nakanishi, M., Ayada, M. and Kawabata, S. (1998), “Optimum silhouette design for ladies’ garments based on the mechanical properties of a fabric”, Textile Res. J., Vol. 68, pp. 578-88. Stanislaw, R., Jadwiga, B. and Krzyztof, H. (1997), “Die ertragsfa¨higkeit der faserflachssorten in verschiedenen umweltbedingungen”, Max-eyth-gesellschaft im VDI (VDI/MEG), Heft 22, pp. 39-40. (The Appendix follows overleaf.) Corresponding author Takako Inoue can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

Trace elements in linen fabrics

185

Thickness and weight

Compression

Shearing

Surface

Bending

Shear stiffness Hysteresis at f ¼ 0.58 Hysteresis at f ¼ 58 Linearity Compressional energy Resilience Thickness at 0.5 gf/cm2 Weight per unit area

2HG 2HG5 LC WC RC T

W

MMD SMD

G

Mean deviation of MIU Geometrical roughness

LT WT RT B 2HB MIU

EM

Tensile

Tensile strain at max. load Linearity Tensile energy Resilience Bending rigidity Hysteresis Coefficient of friction

Symbols Characteristic value

Table AI. Characteristic values of basic mechanical properties and measuring conditions for KESF measurements Weight of specimen per unit area

mg/cm2

Upper limit pressure: 50 gf/cm2 Thickness at 0.5 gf/cm2 pressure

Contactor for geometrical roughness: a steel piano-wire, with 0.5-mm dia. and 5-mm length. Contact force: 10 gf Shear deformation under constant tension of 10 gf/cm Max. shear angle, f ¼ ^ 88

Pure bending Max. curvature, K ¼ ^ 2.5 cm2 1 Contactor for friction measurement: ten parallel steel-piano-wires with 0.5 mm dia. and 5-mm length simulating finger skin geometry,. Contact force: 50-gf

:10 gf/cm

Upper limit tensile force (max. load): 500 gf/cm :50 gf/cm

Strip biaxial deformation

Standard (Kawabata, 1980)

gf/cm degree gf/cm gf/cm – gf cm/cm2 % mm

– mm

– gf cm/cm2 % gf cm2/cm gf cm/cm –

%

Unit

High sensitivity (Matsudaira et al., 1984)

186

Mechanical properties

Measuring conditions

IJCST 22,2/3 Appendix

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

Analysis of tactile perceptions of textile materials using artificial intelligence techniques Part 1: forward engineering B. Karthikeyan School of Engineering and Textiles, Philadelphia University, Philadelphia, Pennsylvania, USA and Department of Electrical and Computer Sciences, College of Engineering, Temple University, Philadelphia, Pennsylvania, USA, and

Forward engineering

187 Received 2 September 2008 Revised 24 May 2009 Accepted 24 May 2009

Les M. Sztandera School of Business Administration, Philadelphia University, Philadelphia, Pennsylvania, USA Abstract Purpose – The first of a two-part series, this paper aims to discuss the design and development of an artificial intelligence-based hybrid model to understand human perception of the tactile properties of textile materials and create an objective system to express those tactile perceptions in terms of measurable mechanical properties. Design/methodology/approach – A forward engineering system using the Model Free Algorithm approach of the Artificial Intelligence Technique to predict the tactile comfort score is presented. Findings – Human perception of tactile sensation is based on the weighted stimulus perceived by the human neural system. Originality/value – Contribution to intelligent textile and garment manufacture. Keywords Artificial intelligence, Modelling, Mechanical properties of materials, Textiles, Programming and algorithm theory Paper type Research paper

Introduction Textile materials form a unique branch in the field of material science. They exhibit an entirely different set of mechanical properties than most other conventional engineering materials. Textile materials find their application not only in the clothing sector but also in several other areas including geo-synthetics, medical transplants, etc. Properties such The research reported was supported in whole by Laboratory for Engineered Human Protection through grant W911QY-04-0001 from Department of Defense/US Army Natick Soldier Systems Center. The authors would like to acknowledge Contracting Officer Technical Representative, Carole Winterhalter, Warfighter Science, Technology and Applied Research Directorate, US Army Natick Soldier Research, Development and Engineering Center, Massachusetts 01760. The authors also thank Dr Howard Schutz, University of California-Davis, California, and Visiting Scientist at Natick, Massachusetts, and Dr Armand Cardello, Senior Research Scientist, Natick, Massachusetts.

International Journal of Clothing Science and Technology Vol. 22 No. 2/3, 2010 pp. 187-201 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011018658

IJCST 22,2/3

188

as inhomogeneity, discontinuity, anisotropicity and easy deformability make them a special case of materials suitable for such distinguished engineering applications (Amirbayat, 1991). They are non-linear and undergo large deformations even under low stress and buckling loads. Hence, the tactile perceptions of a textile material by humans form a different platform of study. A pioneering work of Stylios (1998) laid ground in terms of defining principles for aesthetic measurements of textile materials. It was further advanced when the relationship between the drape attributes and fabric bending, shear and weight was modeled using artificial neural networks (ANNs). It was found that using the natural logarithm of the material property divided first by the weight of the fabric produced the most predictive model (Stylios et al., 2002). In this paper, we analyze and map mechanical properties into perceived tactile comfort. Detailed analyses of both handfeel and mechanical properties and their relationships with tactile comfort, as well as which material properties influence tactile perceptions and why, can be found in Sztandera (2008a, b). Several kinds of testing equipment are available to quantify the mechanical properties of textile materials; the most widely used is the Kawabata Evaluation System (KES-F) for fabrics. The mechanical properties characterized by KES-F systems are listed in Table I (Kawabata et al., 1982). These parameters are analogous to the psycho-physiological evaluation system of human experts who subjectively evaluate the handle property of textile materials (Hu, 2004). In order to evaluate the hand value of a fabric material, the KES-F properties are used to determine the primary hand value, which is used subsequently to derive the total hand value (THV). The THV finds its application in garment manufacturing wherein a two-dimensional fabric material is transformed into a three-dimensional clothing system. In the process, mechanical properties greatly influence the handling and conformation ease of the fabric. Though THV characterizes the fabric in terms of its hand value, it does not depict significantly human perception of tactile comfort properties of the fabric.

Group

Property

Description

Tensile

EMT LT WT RT B 2HB G 2HG 2HG5 LC WC RC MIU MMD SMD W T

Elongation (%) Linearity of load-extension curve (ND) Tensile energy (gf.cm/cm2) Tensile resilience (%) Bending rigidity (gf/cm degree) Hysteresis of bending moment (gf.cm/cm) Shear rigidity (gf/cm. degree) Hysteresis of shear force at 0.5 degrees of shear angle (gf/cm) Hysteresis of shear force at five degrees of shear angle (gf/cm) Linearity of compression-thickness curve (ND) Compressional energy (gf.cm/cm2) Compressional resilience (%) Coefficient of friction (ND) Mean deviation of MIU (ND) Geometrical roughness (mm) Fabric weight per unit area (mg/cm2) Fabric thickness (cm)

Bending Shear Compression Table I. Mechanical property of textile materials evaluated by KES-F module

Surface characteristics Fabric construction

Models based on energy equations (De Jong and Postle, 1977), finite element analysis (Ghosh et al., 1990a, b), stochastic formulations (Behery, 2005), and ANN (Behera and Karthikeyan, 2006) exist to identify the interrelationship between the structure of textile materials and their functional properties. Linear models to predict the tactile comfort of textile materials in terms of both subjective and objective measurements are also found (Wong et al., 2003). Researchers have determined that human tactile perception of textile material is non-linear (Wong et al., 2004) thereby limiting the application of existing models. Moreover, these models are domain specific and their extent of extrapolation is limited. Thus, a system to portray non-linear tactile perception of textile materials in the light of comfort and in terms of their mechanical properties becomes essential. In the current research, a hybrid system based on artificial intelligence techniques was developed to identify the underlying relationship between the fabric mechanical properties and the human perception of tactile comfort. The major tools of AI are ANN and adaptive neuro fuzzy inference system (ANFIS) engines. These tools are capable of identifying/mapping the dynamics of the non-linear relationship that exists in the problem under study.

Forward engineering

189

Materials and methodology A diversified set of fabrics including woven, knitted and non-wovens materials was selected for the evaluation. Laminated fabrics with water, fire, and chemically retardant finishes are included. The range of their mechanical properties as measured using the KES-F module is given in Table II. While these 17 properties form the input of the ANN/ANFIS engine, human perception score of tactile comfort is used as the output. The human perception score is measured using the comfort affective labeled magnitude (CALM) scale shown in Figure 1. The scale developed by the Natick Soldier Systems Center ranges from 2 100 to 100 where a score of 2100 represents the greatest imaginable discomfort and a 100,

Property EMT LT WT RT B 2HB G 2HG 2HG5 LC WC RC MIU MMD SMD T W

Minimum value

Maximum value

0.98 0.48 2.40 22.46 0.01 0.01 0.45 2 0.32 1.70 0.23 0.03 38.35 0.18 0.01 2.18 0.03 4.10

35.87 1.31 61.90 65.30 5.37 6.98 25.54 25.54 39.23 0.58 1.66 94.27 0.38 0.18 27.35 0.23 64.88

Table II. Range of the mechanical properties as measured using KES-F module

IJCST 22,2/3

100

Greatest imaginable comfort

80 Extremely comfortable 60

190 40

Very comfortable

Moderately comfortable

20 0 –20 – 40 – 60 –80

Figure 1. CALM scale

–100

Slightly comfortable Neither comfortable nor uncomfortable Slightly uncomfortable

Moderately uncomfortable Very uncomfortable

Extremely uncomfortable Greatest imaginable discomfort

greatest imaginable comfort. The other labels are distributed in a progressive ratio scale (Cardello et al., 2003). In total, 33 fabric specimens were selected for the evaluation of mechanical properties and the same set of fabrics were evaluated by 50 human subjects for their perceived tactile comfort score. The fabric samples were sequenced in random order for the subjects to evaluate. The mechanical properties were tested with five replicas making a data set of 165 £ 17. The domain specificity is reduced to a greater extent by incorporating fabrics that find application in various domains, namely, battle dress uniforms, ballistic protection, vests, linings, etc. Comfort scale We repeat here the process of the developing the scale, after (Cardello et al., 2003), as it completely defines the way our output variable, tactile fabric comfort, was formed. According to the scale developers (Cardello et al., 2003) in order to develop a sensitive, reliable, and valid labeled magnitude scale of comfort, 35 volunteers, none of whom were members of the descriptive hand panel, were recruited from a random list. Word adjectives that could be used to modify the terms “comfortable” and “uncomfortable” to reflect intensity differences were compiled from previous scaling literature and from standard English language resources. The adjectives “greatest imaginable” and “greatest possible” were included to define scale values commensurate

with a common fixed end-point of positive and negative affective experience, as used in previously developed labeled magnitude scales (Cardello et al., 2003). These adjectives were used to create 41 word phrases, which in combination with two non-polar terms (“neutral” and “neither comfortable nor uncomfortable”), resulted in a total of 43 phrases to be used in scale development. The 43 phrases were printed on separate pages and assembled in random order into testing booklets. Before testing, subjects were provided with written instructions on the procedure to be used in scaling the semantic meaning of the phrases. Oral instructions with an example were also provided. Subjects sequentially rated each of the phrases to index the magnitude of comfort or discomfort connoted by the phrase, using a modulus-free magnitude estimation procedure. In this procedure, subjects assign an arbitrary number to indicate the magnitude of comfort or discomfort reflected by the first phrase (positive numbers used for comfort, negative numbers for discomfort). Subjects then make all subsequent judgments relative to the first, so that if the second phrase denotes twice as much comfort as the first, a number twice as large is assigned; if it denotes one third as much comfort, a number one-third as large as the first is assigned, etc. All ratings were made in spaces provided in the testing booklet. A subset of phrases was chosen to construct a labeled magnitude scale of comfort (Cardello et al., 2003). The criteria for selecting terms were low variability in perceived semantic meaning, parallelism in the terms used to describe comfort and discomfort, and selection of an equal number of comfortable and uncomfortable phrases (a decision based on evidence from the preference scaling literature showing that balanced scales are better for differentiating products). Examination of the standard errors of the geometric means for each of the phrases (Cardello et al., 2003) led to the elimination of several phrases (e.g. “mediocre comfort,” “barely comfortable,” and “a little comfortable”) due to their variable semantic meaning to the subjects. Other phrases were eliminated because of a lack of suitable parallelism in terminology for the purpose of establishing bipolarity (e.g. “superior comfort” and “oppressively uncomfortable”). Applying the remaining criterion to the phrases resulted in the selection of 11 phrases for use in the scale: five associated with comfort, five associated with discomfort, and one neutral term (“neither comfortable nor uncomfortable”) to define the zero point. The geometric mean magnitude estimates of the positive and negative phrases were transformed to range from 0 to þ 100 (positive phrases) and 0 to 2 100 (negative phrases). The phrases were then placed along a 100 mm vertical analogue line scale in accordance with their transformed values. The resulting labeled affective magnitude scale of comfort is shown in Figure 1. The CALM scale shown in Figure 1 has several advantages over other comfort scales commonly used in the literature (Cardello et al., 2003). With this scale, the level of comfort or discomfort experienced by an individual can be readily indexed by simply placing a mark somewhere on the line. This stands in contrast to the difficulty often encountered by subjects using magnitude estimation procedures. However, by having positioned the phrases of comfort/discomfort along the analogue line scale at points representing the magnitude of their semantic meaning as determined by a magnitude estimation procedure, it becomes possible to treat the measured distances along the scale as ratio level data. This stands in contrast to category scales of comfort, which provide only ordinal data. The ratio nature of the CALM scale enables statements to be made about whether a particular sample is 20 percent, 40 percent, three times, etc. as

Forward engineering

191

IJCST 22,2/3

192

Table III. Comparison of a biological and ANN

comfortable (or uncomfortable) as another sample. In addition, it does not require that the data be normalized, as is the case with magnitude estimates. Last, by using the “greatest imaginable” comfort (or discomfort) as end-points on the scale, the scale enables better discrimination between samples/conditions that are either very high or very low in comfort/discomfort and establishes a common ruler by which comfort/discomfort ratings of different subjects can be compared. Artificial neural network An ANN functions as a model based on connectionism. It represents the architecture of a human brain in the sense of interconnected neurons. A neural network essentially consists of several interconnected neurons that can be labeled as “input”, “hidden,” and “output” depending on the layer in which the neuron is present. The analogy between a biological neuron and an artificial neuron is given in Table III. Neural networks are capable of learning the relationship between input and output parameters by adjusting the weights and biases of the interconnected link. A neural network with three layers (one input layer, one hidden layer, and one output layer) can approximate any mathematical function with a satisfactory precision level. There exist neural networks with more hidden layers and the architecture strictly depends upon complexity of the relationship between the inputs and the outputs along with severity of noise in the data (Negnevitsky, 2002). The neural network needs to be trained with data sets whereby the system is provided with the inputs and the outputs; the weights are adjusted in the interconnections to fit with the targeted output. Once the neural network has learned the mapping function, it can predict the output when any unknown instance of the input is fed. This is analogous to how the human brain learns and reacts to unknown situations. Several architectures exist that can be used to model a neural network engine. Feed-forward back-propagation is one of the widely used learning algorithms to train a multi-layer perceptron neural network. It falls under the category of a supervised learning algorithm wherein, at the training phase, the network propagates the input pattern in a sequence through the layers of the network to generate an output. The error in the output, compared with the targeted output, is propagated backwards from the output layer to the input layer. The weights are modified to minimize the error by gradient descent method at each epoch (Baughman and Liu, 1995). Schematic representation of a feed forward neural network with error back propagation is given in Figure 2(a). The 17 mechanical properties form the input vector x(n) and the averaged Tactile perception score forms the output vector Y(n). The total activity at any internal neuron is given by equation (1): Natural neural network

ANN

Soma Dendrite Axon Synapse Electrical potential Threshold potential

Neuron or node Input Output Interconnections Weighted sum Threshold value

Forward engineering

Input signals EMT LT

193

Tactile comfort score

WT

W

Input layer

Hidden layer

Output layer

Error signals (a)

f (x) =

2a (1 – e–bx )

– a;

a = 1, b = 2

1

0

–1

–3

–1

0 (b)

1

3

Notes: (a) Schematic of feed forward back propagation neural network; (b) transfer function to map the inputs to the hidden layer

g kj ¼ f

n X i¼1

Figure 2.

! w kij x ki 2 t kj

ð1Þ

IJCST 22,2/3

194

Where: w kij

is the weight of the ith input to the jth neuron of the kth layer.

x ki

is the ith input.

t kj

is the threshold at the jth node of the kth layer.

f( )

is the activation function.

g kj

is the output of the jth neuron of the kth layer.

The objective of a transfer function is to convert the weighted sum of the inputs to a range that can be [0, 1] or [2 1, 1], and in the present case the latter is followed. The hyperbolic tangent function used as an activation function between the hidden layer and the output layer induces an accelerated learning process (Negnevitsky, 2002). This function, shown in equation (2) and schematically in Figure 2(b), has an additional advantage in the calculation of error gradient. The error gradient is determined just with the partial derivative of the transfer function multiplied by the error in the neuron’s output: f ðxÞ ¼ fðxÞtanh ¼

2a 2a 1 þ e2bx

ð2Þ

where, a and b are constants. Thus, for the neuron j in the output layer, the error gradient is calculated as:

dj ¼ fðxj Þ½1 2 fðxj Þ
Dj where:

dj

is the error gradient of the function f(x).

xj

is the output of jth neuron in the output layer.

Dj

is the difference in the actual output and the calculated output of the jth neuron in the output layer.

Hence, after each epoch, the weights are updated according to equation (3): Dw kij ¼ a:g kj :d kj

ð3Þ

where: Dw kij is the incremental weight of the ith input of the jth neuron of the kth layer.

a

is the learning rate and is usually kept at 0.1.

The details of the architecture and the set parameters for training and testing phases are as shown in Table IV. Thus, the neural network-based model is analogous to the biological system of perception. The mechanical properties are analogous to the stimuli and the conceived perception is reflected in the tactile comfort score. Analogous to the biological neural network, the ANN identifies the relationship between the input and the output. A schematic comparison of the function of a typical biological and ANN system is shown in Figure 3.

Forward engineering

Parameter

Value

Learning algorithm Number of input neurons Number of hidden layers Number of hidden neurons Number of output neurons Number of Epochs Mean square of error (termination criteria) Activation function Total data set Training data set Validation data set Training data set

Feed forward back propagation 17 1 25 1 1,000 0.001 Hyperbolic tangent 165 (33 fabrics £ five samples) 116 (<70 percent of 165) 33 (<20 percent of 165) 16 (<10 percent of 165)

195

Table IV. Details of the neural network architecture

Biological system Body sensors

Signal interpreters

Comfort/ discomfort

Artificial intelligence system Mechanical measurements

ANN/FL engine

Comfort value

Adaptive neuro fuzzy inference system Fuzzy sets possess fuzzy boundaries and the association of any element of the fuzzy set is expressed in terms of the degree of membership to a particular boundary, called membership value. The inferences of the input values are based on rules to predict the output. In ANFIS, the fuzzy sets are derived using a neural network. Adaptive neuro fuzzy system, classified as fused neuro fuzzy system, is a five-layer Tagaki-Sugeno inference system. It uses a neural network to cluster the input data and maps it to the relevant membership function. The rule extraction methodology uses the clustered map to determine the number of rules and antecedent membership functions and then uses linear least squares estimation to determine each rule’s consequent equations. Hence, the input is a fuzzy membership set while the output remains an equation with coefficients for every input function value. A typical Sugeno fuzzy rule can be expressed as: IF x1 is A1 AND x2 is A2 ... AND xn is An THEN g ¼ f( x1, x2, . . . , xn)

Figure 3. Comparison of the function of a typical biological and ANN system

IJCST 22,2/3

196

where, x1, x2, . . . , xn are inputs and A1, A2, . . . , An are fuzzy sets; and g may be a constant or a linear function of the inputs (Lin and Lee, 1996). Here, the mechanical properties evaluated using the KES-F is fed as the input and the coupled neural network system is used to define the membership functions. And, the output layer forms the equation that relates the inputs to the output based on the membership functions and their associated rules. The parameters used to generate the fuzzy inference system are provided in Table V. The neural network identified 29 different membership functions to be included as fuzzy sets and rules. Based on the rules, the relationship surface is formed. For illustrative purpose and to establish the non-linear behavior of the properties, few of the surfaces formed to predict the output are shown in Figure 4. Results and discussion Predictions of ANN and ANFIS The data are split into three parts, namely, training set, validation set, and testing set. The proportions of these data sets are 70, 20 and 10 percent, respectively. The validation set is used at the time of training to avoid the phenomena of over-fitting. Following the training phase, the neural network is tested for its performance using the test data. Details of the testing data are given in Table VI. Comparison of the actual Parameter Influence range Squash factor Accept ratio Reject ratio

Tactile comfort score

Tactile comfort score

50 0 –50 60

Figure 4. Surface of selective input parameters with respect to the output

0.5 1.25 0.65 0.15

–40 – 45 5

50 40 RT 30

10

20 EMT

30

–30 –35

4

3 B 2 1

5

20 25 10 15 G

–20 –30

– 40

–40 0.2

–35

Tactile comfort score

Tactile comfort score

Table V. Parameters for defining the fuzzy inference system

Value

0.15 0.1 T 0.05

0.2

0.25

0.3 MIU

0.35

–50 120 100

80 60 40 20 2HG5

80 60 40

RC

output of the test data and the predicted data is given in Table VI and is shown in Figure 5. The same data set is used to train the ANFIS engine and the prediction results are shown in Figure 6.

Test data number

Actual output

ANN predicted output

ANFIS predicted output

2 6.4 47.04 12.42 2.8 28.66 21.52 2 42.44 48.07 12.96 2 21.03 2 27.54 2 8.38 33.038 81 2 19.019 2 60

0.539746 40.28957 7.150654 4.125191 31.31824 18.21426 238.3522 46.38379 14.3607 221.3752 229.4784 217.8968 31.34551 79.39728 220.741 260

2 4.6709 47.3726 13.5235 2 0.3456 30.6334 21.5246 2 37.2137 52.9692 12.8586 2 18.8693 2 27.9164 2 8.3465 33.038 80.9999 2 19.019 2 60.0002

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Forward engineering

197

Table VI. Details of the test data and the predicted results

Prediction results of tactile comfort score using ANN engine 100 Predicted Actual

80

Tactile comfort score

60 40 20 0 –20 – 40 – 60

0

2

4

6

8 10 Specimen

12

14

16

18

Figure 5. Prediction of comfort score using ANN engine

IJCST 22,2/3

Prediction results of tactile comfort score using ANFIS engine 100 Actual Predicted

80

198

Tactile comfort score

60 40 20 0 –20 – 40 – 60

Figure 6. Prediction of comfort score using ANFIS engine

–80

0

2

4

6

8 10 Specimen

12

14

16

18

Predictions using reduced attributes In order to minimize the “curse of dimensionality” and reduce the computational resources, an attempt is made to reduce the number of attributes and design an engine that can predict the human perception of tactile score using the reduced attributes. In order to reduce the number of attributes, a subset evaluation algorithm was used. The WEKA software is used for the subset evaluation with the “best first” search algorithm. Subset evaluation is based on the worth of a subset of attributes by considering the individual predictive ability of each feature along with the degree of redundancy between them. Best first approach is used as the search method in selecting the attributes. While evaluating the subset contribution, a five cross-validation was set to avoid over-fitting. At the end of the validations, six attributes are identified as the best contributing parameters and are listed below: (1) EMT – elongation percentage. (2) LT – linearity of tensile. (3) G – shear rigidity. (4) HG5 – Shear hysteresis. (5) RC – compressional resilience. (6) MIU – coefficient of friction. The elastic nature of the fabric in the longitudinal and transverse directions that influence the tactile comfort a human subject perceives is reflected by the EMT, LT, and RC properties. Similarly, the viscoelastic nature and the rigidity towards the force in

shear direction are reflected in the G and HG5 properties. The next major influencing stimulation is the frictional property of the fabric, which is reflected in the MIU property. Similarly, the subset evaluation with the search algorithm changed to genetic search was run and it yielded the same set of attributes. The neural network is then trained using the selected attributes (six attributes) as the input and the Tactile Comfort score as output. The coefficient of determination in this case is around 0.96. While using all the 17 attributes, the R 2 value was around 0.99 and with a loss of 0.03, the numbers of attributes are reduced to six. Similarly, the ANFIS engine produced a result with R 2 of around 0.98, but with only five membership functions against the 29 in the full set. The prediction results of ANN engine and ANFIS engine are as shown in Figures 7 and 8, respectively.

Forward engineering

199

Conclusion The process of bio-mimicking the tactile sensory perception of textile materials using AI tools provided excellent results. The system is capable of predicting tactile comfort the way a human subject experiences it. The prediction ability of both the ANN and the ANFIS engines are at par with each other. Using the dimension reduction algorithms, the numbers of attributes that contribute effectively to the output are chosen. The six attributes fall under four groups of human tactile sensations such as elastic nature, rigidity of the material, compressibility and friction, and the subset evaluation schema was able to identify these influencing parameters. Thus, the AI approach provides a successful platform for cognitive perception analysis. In the second part of the paper, a reverse engineering approach will be presented.

Prediction results using ANN engine with reduced data set 100 Predicted Actual

80

Tactile comfort score

60 40 20 0 –20 – 40 – 60

0

2

4

6

8 10 Specimen

12

14

16

18

Figure 7. Prediction of comfort score using ANN engine with reduced data set

IJCST 22,2/3

Prediction results using ANFIS engine with reduced data set 100 Predicted Actual

80

200

Tactile comfort score

60 40 20 0 –20 – 40

Figure 8. Prediction of comfort score using ANFIS engine with reduced data set

– 60 –80

0

2

4

6

8 10 Specimen

12

14

16

18

References Amirbayat, J. (1991), “The buckling of flexible sheets under tension. Part I: theoretical analysis”, J. Text. Inst., Vol. 82 No. 1, pp. 61-70. Baughman, D.R. and Liu, Y.A. (1995), Neural Networks in Bioprocessing and Chemical Engineering, Academic Press, San Diego, CA. Behera, B.K. and Karthikeyan, B. (2006), “Artificial neural network-embedded expert system for the design of canopy fabrics”, Journal of Industrial Textiles, Vol. 36 No. 2, pp. 111-23. Behery, M.H. (2005), Effect of Mechanical and Physical Properties on Fabric Hand, Woodhead, Cambridge. Cardello, V.A., Winterhalter, C. and Schutz, H.G. (2003), “Predicting the handle and comfort of military clothing fabrics from sensory and instrumental data: development and application of new psychophysical methods”, Text. Res. J., Vol. 73 No. 3, pp. 221-37. De Jong, S. and Postle, R. (1977), “An energy analysis of woven-fabric mechanics by means of optical-control theory. Part II: pure-bending properties”, J. Text. Inst., Vol. 68, pp. 62-9. Ghosh, T.K., Batra, S.K. and Barker, R.L. (1990a), “The bending behavior of plain-woven fabrics. Part I: a critical review”, J. Text. Inst., Vol. 81 No. 3, pp. 245-55. Ghosh, T.K., Batra, S.K. and Barker, R.L. (1990b), “The bending behavior of plain-woven fabrics. Part I: the case of bilinear thread-bending behavior and the effect of fabric set”, J. Text. Inst., Vol. 81 No. 3, pp. 255-77. Hu, J. (2004), Structure and Mechanics of Woven Fabrics, Woodhead Publishing in Textiles, CRC Press, Boca Raton, FL. Kawabata, S., Postle, R. and Niwa, M. (1982), “Objective specification of fabric quality, mechanical properties and performance”, Proceedings of the Japan-Australian Joint Symposium on Objective Specification of Fabric Quality, Mechanical Properties and Performance, Kyoto, May 10-21.

Lin, C.T. and Lee, G. (1996), Neural Fuzzy Systems: A Neuro-Fuzzy Synergism to Intelligent Systems, Prentice-Hall, Englewood Cliffs, NJ. Negnevitsky, M. (2002), Artificial Intelligence – A Guide to Intelligent Systems, Pearson Education, Harlow. Stylios, G.K. (1998), “Mechatronic principles for aesthetic measurement of textile materials”, in Adolfsson, J. and Karlsen, J. (Eds), Mechatronics’ 98, Pergamon Press, Oxford, pp. 721-6. Stylios, G.K., Powell, N.J. and Cheng, L. (2002), “An investigation into the engineering of the drapability of fabric”, Transactions of the Institute of Measurement and Control, Vol. 24 No. 1, pp. 33-51. Sztandera, L.M. (2008a), “Identification of the most significant sensory and mechanical properties influencing tactile fabric comfort”, Proceedings of the 8th WSEAS International Conference on Applied Computer Science (ACS’08), pp. 221-5. Sztandera, L.M. (2008b), “Predicting tactile fabric comfort from mechanical and handfeel properties using regression analysis”, Proceedings of the 8th WSEAS International Conference on Applied Computer Science (ACS’08), Venice, pp. 217-20. Wong, A.S.W., Li, Y. and Yeung, P.K.W. (2003), “Neural network predictions of human psychological perceptions of clothing sensory comfort”, Text. Res. J., Vol. 73 No. 1, pp. 31-7. Wong, A.S.W., Li, Y. and Yeung, P.K.W. (2004), “Predicting clothing sensory comfort with artificial intelligence hybrid model”, Text. Res. J., Vol. 74 No. 1, pp. 13-19. Corresponding author B. Karthikeyan can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

Forward engineering

201

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

IJCST 22,2/3

Analysis of tactile perceptions of textile materials using artificial intelligence techniques

202

Part 2: reverse engineering using genetic algorithm coupled neural network

Received 2 September 2008 Revised 24 May 2009 Accepted 24 May 2009

B. Karthikeyan School of Engineering and Textiles, Philadelphia University, Philadelphia, Pennsylvania, USA and Department of Electrical and Computer Sciences, College of Engineering, Temple University, Philadelphia, Pennsylvania, USA, and

Les M. Sztandera School of Business Administration, Philadelphia University, Philadelphia, Pennsylvania, USA Abstract Purpose – The second of a two-part series, this paper aims to explain the design and development of a hybrid system for reverse engineering. Design/methodology/approach – A prediction engine to map the perception of tactile sensations using a neural network engine was developed. Since seventeen mechanical properties form the input and tactile compfort score is used as the output - a direct reversal of the data set becomes impossible, hence, a hybrid approach was employed. The neural net is coupled with a genetic algorithm engine for the reversal process. The trained neural network acts as the objective function to evaluate the property set while the solution set is generated by Genetic Algorithm (GA) engine. Limitation of the GA and a means to overcome it is discussed. Application software based on the current research is also presented. Findings – Human perception of tactile sensations is non-linear in terms of the mechanical properties of textile materials. Originality/value – The paper deals with reverse engineering and discusses application software based on the current research. Keywords Artificial intelligence, Modelling, Mechanical properties of materials, Textiles, Programming and algorithm theory Paper type Research paper

International Journal of Clothing Science and Technology Vol. 22 No. 2/3, 2010 pp. 202-210 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011018667

The research reported was supported in whole by Laboratory for Engineered Human Protection through Grant W911QY-04-0001 from Department of Defense/US Army Natick Soldier Systems Center. The authors would like to acknowledge Contracting Officer Technical Representative Carole Winterhalter, Warfighter Science, Technology and Applied Research Directorate, US Army Natick Soldier Research, Development and Engineering Center, Natick, Massachusetts. The authors also thank Dr Howard Schutz, University of California-Davis, Davis, California, and Visiting Scientist at Natick, Massachusetts, and Dr Armand Cardello, Senior Research Scientist, Natick, Massachusetts.

Introduction Engineered product manufacturing is becoming an indispensable strategy for any manufacturer particularly with products involving direct human interaction. Textile materials contribute significantly to human well-being and require engineered product development. Two principal aspects of textile materials are functionality and comfort. Functionality can be achieved by incorporating specialized materials for the required set of properties. The importance of mechanical properties on the functional behavior of textile materials has been studied extensively. A comprehensive review of surface modification techniques employed for improved functional behavior of textiles can be found in Wei (2009). Unlike functionality, tactile comfort sensation forms a complex platform as it involves psycho-physiological phenomena. In his pioneering work, Stylios (1998) systematically defined principles for aesthetic measurements of textile materials. Stylios et al. (2002) further advanced the work by mapping the relationship between the drape attributes and fabric bending, shear and weight using neural networks. It was reported that using the natural logarithm of the material property divided first by the weight of the fabric produced the most predictive model. Thus, mapping mechanical properties and tactile perception of textile materials pose an interesting arena for research. Since the structure-property relationship of textile materials is non-linear – and the human perception of tactile comfort is relatively complex for empirical identification – a model-free algorithm becomes necessary. Researchers have attempted to identify the functional properties of textile materials based on their mechanical and constructional properties. Energy-based models (De Jong and Postle, 1977), finite element models (Ghosh et al., 1990a, b) and artificial neural network (ANN)-based models (Behera and Karthikeyan, 2006) exist for the aforementioned mapping. Other models include stochastic linear models (Behery, 2005) and subjective models for forward engineering. In the first part of this series, a forward engineering engine was discussed. Literature on reverse engineering of textile materials is found to be scarce. Studies involving reverse engineering based on ANN are reported in the case where the number of inputs and outputs are large in number. Reverse engineering can be approached as a process of reversing the data set/equation or a process of global solution search. In the current research, global solution search based on genetic algorithms is used. We present a hybrid engine that couples the neural network and genetic algorithm to determine the mechanical properties of textile materials. Materials and methodology In the first part of this series, a neural network engine was developed to predict the tactile comfort score of textile materials based on their mechanical properties. The analysis was a forward engineering approach. The neural network has 17 input parameters and one output parameter. Hence, a data set of 165 £ 17 was used to train the net. In the case of reverse engineering, the objective is to identify a suitable set of mechanical properties needed to deliver a required tactile comfort score. Here, training the neural network with one input and 17 outputs is not an advisable approach. Hence, a neural network-genetic algorithm hybrid engine is developed to determine objectively the material properties for a given level of tactile comfort.

Reverse engineering

203

IJCST 22,2/3

204

Genetic algorithm Genetic algorithm falls under evolutionary computation methods and works on the basis of genetic inheritance (Negnevitsky, 2002). It simulates the principles of evolution in several ways. The process of genetic algorithm essentially includes selection, crossover, mutation, and reproduction. Searching the solution space starts with a set of randomly chosen values called parent population. In the present study, 17 parameters are to be selected in a random fashion and one set of 17 parameters form one parent. Similarly, 100 of parents were selected and encoded. The parameters were encoded into 8-bit binary strings. The string is called a chromosome. In the process of crossover, pairs of parent chromosomes are selected for mating. The selection of parents for mating is based on their fitness for a particular objective function (trained neural network). A roulette wheel selection mechanism is used for selection. A probability of 95 percent was set for the crossover, meaning that a chromosome has 95 percent of chance for undergoing crossover. In crossing over, a particular segment of chromosomes is exchanged between the parents. Since the number of parameters is large, each of the parent chromosomes has a 136-bit (8 bit £ 17 parameters) length and hence each 8 bit is involved individually for the crossover. Instead of using a parent as a single entity for crossover, the individual parameter string is crossed over to improve the scope of inheritance. Few of the parent chromosomes are transferred as an elite child to the next generation. About 2 percent of the population is transferred under the elitism tag. Also, the chromosomes undergo the mutation process at the rate of 1 percent per generation. Mutation is the process of reversing the nature of a particular bit from “0” to “1” or vice versa. Thus, the next generation of chromosomes, called child chromosomes, is formed from the parent chromosomes. These chromosomes are again tested for their fitness. When the process is repeated for several generations, those chromosomes that are fit and are capable of satisfying the objective function survive and the remaining chromosomes die off. The entire sequence of generating child chromosomes from the parents is shown in Figure 1, and the parameters of the genetic algorithm are given in Table I. Since the relationship between the tactile comfort score and the mechanical properties is identified using a model-free algorithm, there exists no empirical equation to be used as the objective or fitness function that could be used to gauge the parameters. Hence, the neural network is used as an objective function. The trained neural network is capable of predicting the tactile comfort score for a given set of mechanical properties. The process of selecting a suitable value for 17 mechanical properties is a problem of global optimization search and the solution space is searched using a genetic algorithm approach. Genetic algorithm is used, in the present case, to select the parameters from a global search space and evaluate the selected values through the neural network for its closeness to the objective. Thus, the neural network acts as the objective function for the genetic engine to decide the direction of search. The fitness score can be calculated using: C req C calc FitnessScore ¼ 1 2 C req where: Creq is the required tactile comfort score. Ccalc is the tactile score obtained for the current solution set.

Reverse engineering

Selection Parent-2

Parent-1

205

Cross-over point

Cross-over point Cross-over

Mutation (one of the bits is reversed)

Child-1

Parameter Number of bits per parameter Selection mechanism Crossover percentage Mutation percentage Elitism percentage Stopping criterion

Child-2

Figure 1. Schematic of the process of genetic algorithm

Value 8 Roulette wheel selection 95 1 2 200 generations

The parameters are evaluated using the neural network and the parameter space is navigated accordingly for a better set of parameters based on the predicted value for the current set of selected parameters. Figure 2 shows the schematic of how the neural network-genetic algorithm hybrid system works. Results and discussion The hybrid system was tested by setting the query to generate mechanical properties for a tactile comfort score of 100, which designates the greatest imaginable comfort in the comfort affective labeled magnitude scale. The results after each 50 generations are given in Table II. After about 200 generations, the solution convergence became limited. The tactile comfort score predicted by the neural network for the solution parameters is approximately 100.03. Since genetic algorithm starts with a random set of parameters, the selection space is restricted to the range of values each parameter possesses. The range is set based on the range of the parameters used to train the neural network.

Table I. Operating parameters of the genetic algorithm engine

IJCST 22,2/3

Start

Generate ‘n’ random parents

206 Selection of mates – roulette wheel

Cross-over, mutation, reproduction

Calculate fitness for child population

Feed the child population as input to the neural network

Is the needed comfort score arrived?

Figure 2. Schematic representing the neural network-genetic algorithm coupled hybrid system

Yes Stop

Property

Table II. Solution of genetic algorithm

No

EMT LT WT RT B HB G HG HG5 LC WC RC MIU MMD SMD T W Tactile comfort score

Generation-1

Generation-50

Generation-100

Generation-200

20.56 0.48 47.72 47.78 1.10 0.23 0.69 5.63 6.43 0.45 0.47 33.94 0.37 0.03 13.68 0.09 12.41 64.43

31.88 0.68 29.37 29.56 0.76 0.77 4.58 2 0.02 18.11 0.41 1.59 44.14 0.37 0.05 12.26 0.18 13.60 72.29

28.13 0.55 56.29 27.43 0.45 0.68 3.26 8.29 8.12 0.25 0.85 47.38 0.37 0.08 7.91 0.10 7.77 80.09

24.47 0.94 53.50 26.49 0.55 0.23 6.85 12.85 21.84 0.26 1.47 42.30 0.30 0.12 15.79 0.15 12.45 100.03

Genetic algorithms are sensitive to local solutions; this means if the solution is to search a minimum in the solution space and if there exists several local minima and a global minimum, there is no assurance that genetic engine would reach the global minimum. This problem can be reduced to a greater extent by creating a varied number of parent populations with different lengths of encoding. Thus, the same experiment was repeated with 100 parent chromosomes and with 16- and 32-bit chromosomes. The results of the 200th generation, in each case together with the 8-bit encoded results, are given in Table III. Comparing the converged solutions of different bit length processes depicts an excellent correlation, and hence can be safely concluded that the solution is not locked in a localized space. Averaging out the result of different bit encoded can be considered as a solution obtained through committee of genetic engines. Figure 3 shows the result obtained for a tactile comfort score of 100 using three levels of bit encoding. Validation of the hybrid system can be done by running it to develop a solution for a tactile comfort level whose mechanical properties are known. From the data set used for training the neural network, two samples are drawn whose mechanical properties are listed in Table IV and these are compared with the solution obtained through the hybrid engine after 200 generations. Comparison of the results is shown in Figures 4 and 5. A coefficient of determination around 0.98 for Sample-1 and approximately 0.96 for Sample-2 proves that the genetic engine is capable of converging to the required solution. Combining forward and reverse engineering, instrumental software (COMphi) was developed in Visual Basicw. The user can choose forward or reverse engineering. The system also allows for belief revision and data fusion. In forward engineering, the system is limited for feed-forward back-propagation architecture. In reverse engineering, the bit encoding is the only allowed method with a maximum of 32 bits.

Property EMT LT WT RT B HB G HG HG5 LC WC RC MIU MMD SMD T W Tactile comfort score

8-bit encoding

16-bit encoding

32-bit encoding

24.47 0.94 53.5 26.49 0.55 0.23 6.85 12.85 21.84 0.26 1.47 42.3 0.3 0.12 15.79 0.15 12.45 100.02

24.85 1.03 33.38 32.51 0.17 0.14 2.42 10.67 17.01 0.27 1.53 43.7 0.36 0.09 16.89 0.17 19.48 100.00

26.3 0.51 7.27 49.06 0.08 0.53 4.21 16.55 21.59 0.34 1.51 40.63 0.21 0.08 17.04 0.16 9.12 100.00

Reverse engineering

207

Table III. Comparison of solutions obtained with different bit length of encoding

IJCST 22,2/3

Figure 3. Effect of bit lengths on the solution

Value of mechanical properties (log scale)

W

SMD

T

MMD

MIU RC

WC

HG5

HG

LC

HB G

RT

0.1

8-Bit

16-Bit

32-Bit

0.01 Mechanical properties Note: A means to verify the global solution

Property

Table IV. Validation of genetic solution compared with known samples

WT

1

B

LT

10

EMT

208

Effect of bit length on converged solution 100

EMT LT WT RT B HB G HG HG5 LC WC RC MIU MMD SMD T W

Sample-1 (actual)

Sample-1 (calculated)

Sample-2 (actual)

Sample-2 (calculated)

35.87 0.56 52.70 22.46 0.06 0.01 0.45 1.64 1.73 0.39 0.32 59.73 0.27 0.04 4.52 0.08 16.07

35.75 0.49 58.16 23.22 0.13 0.01 0.81 4.95 3.74 0.36 0.84 81.21 0.24 0.12 9.57 0.05 16.44

1.60 0.90 3.70 48.30 0.78 0.41 2.31 3.41 12.28 0.32 0.22 51.84 0.21 0.04 12.97 0.08 25.96

3.30 0.90 2.76 50.12 0.59 1.78 1.97 4.91 26.13 0.26 0.18 72.77 0.25 0.06 13.34 0.03 23.93

Conclusion The hybrid system based on neural network and genetic engine for the process of reverse engineering is developed and the results are excellent. The system is capable of identifying the set of mechanical properties that gives a required amount of tactile comfort. Evolutionary techniques present a promising tool for global solution search. If the neural network could be trained to map the relationship between the thermal properties and thermal comfort, the hybrid system can be used for reverse engineering

Reverse engineering

Reverse engineering - mechanical properties

MMD

209 W

RC

SMD

T

MIU

WC

LC HG5

HG

G

HB

RT

WT

1

B

LT

10

EMT

Value of mechanical properties (log scale)

100

0.1

0.01 Actual

Figure 4. Validation of genetic solution compared with sample-1

GA solution

0.001 Mechanical properties Note: R2 = 0.97

W

SMD

T

MMD

MIU RC

HG5

HG

G

WC

LC

HB

Actual

RT

WT

B

LT EMT

Value of mechanical properties (log scale)

Reverse engineering - mechanical properties

GA predicted Mechanical properties

Note: R2 = 0.95

the thermal characteristics, too, of textile materials. Thus, the artificial intelligence-based system is capable of identifying the relationship between the mechanical properties and human perception of tactile sensations, and can be successfully used for the process of reverse engineering.

Figure 5. Validation of genetic solution compared with sample-2

IJCST 22,2/3

210

References Behera, B.K. and Karthikeyan, B. (2006), “Artificial neural network-embedded expert system for the design of canopy fabrics”, Journal of Industrial Textiles, Vol. 36 No. 2, pp. 111-23. Behery, M.H. (2005), Effect of Mechanical and Physical Properties on Fabric Hand, Woodhead, Cambridge. De Jong, S. and Postle, R. (1977), “An energy analysis of woven-fabric mechanics by means of optical-control theory part II: pure-bending properties”, Journal of the Textile Institute, Vol. 68, pp. 62-9. Ghosh, T.K., Batra, S.K. and Barker, R.L. (1990a), “The bending behavior of plain-woven fabrics part I: a critical review”, Journal of the Textile Institute, Vol. 81 No. 3, pp. 245-55. Ghosh, T.K., Batra, S.K. and Barker, R.L. (1990b), “The bending behavior of plain-woven fabrics part II: the case of bilinear thread-bending behavior and the effect of fabric set”, Journal of the Textile Institute, Vol. 81 No. 3, pp. 255-77. Negnevitsky, M. (2002), Artificial Intelligence – A Guide to Intelligent Systems, Pearson Education, Harlow. Stylios, G.K. (1998), “Mechatronic principles for aesthetic measurement of textile materials”, in Adolfsson, J. and Karlsen, J. (Eds), Mechatronics’ 98, Pergamon Press, Oxford, pp. 721-6. Stylios, G.K., Powell, N.J. and Cheng, L. (2002), “An investigation into the engineering of the drapability of fabric”, Transactions of the Institute of Measurement and Control, Vol. 24 No. 1, pp. 33-51. Wei, Q. (2009), Surface Modification of Textiles, Woodhead, Cambridge. Further reading Lin, J.J. (2003), “A genetic algorithm for searching weaving parameters for woven fabrics”, Textile Research Journal, Vol. 73 No. 2, pp. 105-12. Corresponding author B. Karthikeyan can contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

The application of codesign in new bra product innovations

Codesign in new bra product innovations

Chi-Shun Liao Department of Marketing and Logistics Management, Hsin Sheng College of Medical Care and Management, Longtan Township, Taiwan, Republic of China, and

Cheng-Wen Lee

211 Received 18 August 2008 Revised 3 June 2009 Accepted 3 June 2009

International Trade Department, Chung Yuan Christian University, Chung Li, Taiwan, Republic of China Abstract Purpose – The purpose of this paper is to discuss how brassiere manufacturers develop new designs for bra products, suitable for individual consumers, through consumer codesign. Design/methodology/approach – New product design that relies on conjoint analysis algorithms can depict multidimensional attribute profiles, such that consumers’ choice behavior reflects their preferences and overall judgment of the profiles. This statistical technique provides a means to codesign and customize bra products and thereby enhance the overall bra design process. Findings – Bra products codesign suggests goals such as attractive appearance, shoulder strap style, vivid/mild color, elegance/sexy lace, comfort/practicality, fabric, lining, comfort/attractive appearance, neckline design, comfort/excellent function cut, sewn cups, and generous quantities. The most preferred combination of attributes for all respondents is a cotton/cotton blend fabric, seamless bra that offers a detachable shoulder strap, lavender color, a two-strap style, lace details, and a low-cut plunge neckline. The paper illustrates consumers’ bra awareness attributes, codesign approach, and individual optimum individualized bra designs. Practical implications – The results provide a useful source of information for product managers, who should consider the use of codesign to design the best products for individual consumers and decrease the risk of design failure, as well as promote consumer loyalty and satisfaction toward the product. Originality/value – The paper provides a unique method to understand the new product codesign structure and make bra product design decisions that integrate optimum individualized design. Keywords Product design, Lingerie, Consumer behaviour Paper type Research paper

1. Introduction Humans’ needs and desires demand the continuous development of new products, which in turn suggests that product designers should consider consumer’s needs and preferences as the main prerequisite of and information pertaining to new product design. Such customer centricity may prompt manufacturers to turn to customer codesign, in which they create products or services that meet the desires and wishes of each individual customer exactly (Berger et al., 2005). Codesign recognizes the different ideals and perspectives that mark consumers and deploys a design process to address this diversity (Wikipedia, 2008). Many codesign activities occur on dedicated interfaces and allow for the joint development of products and solutions between individual customers and manufacturers.

International Journal of Clothing Science and Technology Vol. 22 No. 2/3, 2010 pp. 211-227 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011018676

IJCST 22,2/3

212

An apparel design, though not exactly an exception to more general ideas about product design, offers several peculiarities worth investigating. For example, as one of the most dynamic developers of new product features, the apparel industry creates new products in an atmosphere in which competition is fierce, the product is complex, and the challenge of the constant refinement and innovation is virtually unmatched. Among the various types of apparel, brassiere design requires a particularly lengthy process that demands design creativity, precision pattern making skills, and a detailed knowledge of fabric (Hardaker and Fozzard, 1997). Researchers continue to explore rapid prototyping, body scanning, and other innovative product development technologies to improve this development process. However, many of these efforts continue without the benefit of a thorough understanding of how the manufacturers could obtain and employ various types of consumer information in their current and future development processes (May-Plumlee and Little, 2006). According to the theory of consumer behavior, when consumers buy products, they consider not just a single attribute but base their evaluation on the combination of product attributes in their perceptions and preference decisions, as well as eventually their buying decision. The buying decision represents an essential step in successful new product development, and it may benefit from the effects of codesign. Most previous studies of bras focus solely on attributes such as optimum fit, pattern designs, fabric selection, and pattern development and grading (Hardaker and Fozzard, 1997; Wang and Zhang, 2007). However, by considering the optimization of a single effect, they neglect the overall profile utility produced by the combination of various product attributes. Such a combined profile utility might maximize consumer utility and serve as the reference for codesign optimization. During the new product idea or design development process, conjoint analysis can indicate the optimal product design. Consumers base their evaluations on the combination of many product attributes to choose products that will offer them the maximum benefit. In this structural context, the manufacturer must be able to generate innovative ideas and put them into practice in the shortest possible time. The use of conjoint analysis also can reveal the relative importance and best product combination for consumers with regard to each attribute of brassiere products, which in turn can serve as an important reference for the design and development of new products. As Berger et al. (2005) show, manufacturers possess specific process knowledge (e.g. research and development and production) but often lack interaction capabilities; retailer personnel often lack motivation to learn from their direct interactions with customers or transfer their knowledge. Therefore, the development of new product ideas often relies on a one-way communication process. If manufacturers could consider consumers’ preferences and demands and then integrate them with the innovative ideas of professional designers, they could form a two-way communication path with much greater utility. This benefit also could help solve the problem of inconsistent recognition by designers, manufacturers, and consumers. The unique method developed herein addresses these issues. 2. Literature review 2.1 New product development through codesign New product design gains new potential from emerging and converging technologies, which enable manufacturers to communicate directly with customers and develop

customized products. Apparel firms generally adopt either 3D body scanning to design products that fit better or codesign to create unique product designs. Fiore et al. (2004) argue that codesign represents a customization option, because the product design is based on a customer’s selections from a range of design feature offerings. Codesign provides professional assistance as the customer assesses fashion selections, makes design choices, and facilitates image depictions. This option requires the consumer to interact with a trained design manager, who provides the degree of assistance required for the customer to create customized apparel (Anderson-Connell et al., 2002). Thus, its objective is to deliver goods and services that meet individual customers’ needs, and it implies a new form of cooperation: the manufacturer must interact with the customer to obtain specific information that defines and translates the customers’ needs and desires into a concrete product specification (Berger et al., 2005). Using selectable design features provided by manufacturer, including product attributes such as style, color, and fabric, consumers arrange their own optimum product. In turn, codesign offers greater value for customers because it helps them develop differentiated, unique products and services. As codesigners, customers create unique products from an array of design options and watch the creation take shape on the computer screen. The resulting product offers better fit or meets the customer’s design specifications better. Many products and services are subject to customization, from automobiles and insurance to fashion products (Fiore et al., 2004). Codesigning brassiere products may offer memorable experiences that entice the customer. In this study, the codesign process presents product design options and computer modeling to help customers select the options they prefer, which should foster an engaging experience due to the novelty, creative expression, and interface associated with advanced technology. Specifically, the bra design process, which relies on conjoint analysis, enables the customer to choose the garment, fabric, neckline design, cups style, lace, color, lining, shoulder strap style, and measurements and fit preferences. To ensure the products produced from the codesign process offer maximum efficiency, the analysis reveals those product attributes that may be codesigned and communicated easily for a realistic evaluation. In addition, these attributes must be realistic. 2.2 Brassiere product attributes and evaluative criteria When consumers assess clothes, they consider intrinsic and extrinsic cues. Intrinsic cues are those product attributes that they cannot change or experimentally manipulate without changing the physical characteristics of the product itself (Wheatley et al., 1981). Extrinsic cues are attributes that are not part of the core product (Hill and Garner, 2001). Both types help the consumer distinguish their purchase intention and perceived value (Lee and Liao, 2007). Moreover, these cues should transform into the consideration basis for product designers. In this stage, consumers assess products on various attributes, in relation to what they consider most important, and the product attitudes that influence the codesign get identified. Intrinsic cues includes physical characteristics (e.g. color, lining, and shoulder strap), tactile characteristics (e.g. fabric softness, silkiness, and warmth), and functional properties (e.g. durability, enhance cleavage, and comfort); the extrinsic cues refer to brand name, price, and so forth. Beaudoin et al. (2000) identify 12 attributes that

Codesign in new bra product innovations 213

IJCST 22,2/3

214

correlate with consumer attitudes toward apparel: good fit, durability, ease of care, favorable price, comfort, quality, color, attractiveness, fashionableness, brand name, appropriateness for occasion, and choice of styles. North and De Vos (2002) also show that female consumers’ preferences for apparel items may depend on the joint influence of product attributes, such as quality, style, and price. According to Textile Industry Affairs (2008), apparel purchase decisions actually entail three-tiered events. First, easy care and durability represent the primary purchase requirements. Second, consumers prefer characteristics such as price/value, bleachability, and fabric blend. Third, attributes as recognized brand name, hand wash, cold wash, and dry flat reflect the final decision features. The joint effects of various product attributes on the final decision to purchase a specific item of clothing therefore require consideration in research into consumer purchasing decisions. When consumers buy bras specifically, according to ICT Life Style Research Center (2004), they consider style, color, price, fabric, lining, function, and shoulder strap. These cues manifest themselves in preferences, expectations, and quality assumptions that may become ingrained or generalized as product prejudices (Schutz et al., 2005). Although various experimental studies use physical fabrics or bar samples (Fritz and Gardner, 1988), few employ customer codesign, which prevents them from including more attributes. Furthermore, in general, existing studies imply that consumers value intrinsic characteristic more than extrinsic characteristics, consistent with research that examines this point specifically. Therefore, this study addresses the combined influence of the following intrinsic cues during codesign: fabric, lining, neckline design, cups, lace, colors, and shoulder strap style. 2.3 Conjoint analysis for new product development An analytical technique used extensively in recent years, conjoint analysis relies on the assumption that consumers make purchase decisions by considering multiple attributes of a product simultaneously. To make such decisions, consumers must make trade-offs, because usually one product does not have all of their favored attributes. The results of the analysis provide designers with a hypothetical product, a particular combination of attributes and levels, that is most preferred by consumers, which is especially useful for developing a new product based on the hypothetical “best” product. Conjoint analysis also reveals the importance of involving final customers in the design throughout the steps of generating ideas, defining specifications, revising prototypes, beta testing, and final testing. Specifically, conjoint analysis highlights those characteristics of the product on which the development team should concentrate, which helps minimize the risk of wasted efforts on secondary aspects or features that customer’s view as unimportant (Gustafsson et al., 2003). Most conjoint studies pertain to new product design, because its primary usefulness involves designing physical features for a product formulation (Beane and Ennis, 1987). Green and Srinivasan (1990) point out that approximately 400 commercial applications per year derive from early conjoint analysis, mostly in the form of new product/concept evaluations, repositioning, competitive analyses, and market segmentation. Wittink and Cattin (1989) obtained information from 66 companies that had collectively completed about 1,062 conjoint projects from 1981 to 1985. Gustafsson et al. (2003) point 519 companies in the Germany provided information for their activities in the field of conjoint analysis.

North and De Vos (2002) list several benefits of a conjoint study: . a better understanding of consumers’ selection criteria when purchasing apparel; . more efficient planning of apparel merchandise mixes; . more effective planning of promotional messages and strategies; and . refining training strategies for sales consultants.

Codesign in new bra product innovations 215

3. Method The theoretical framework for conjoint analysis is based on Lancaster’s utility theory, which assumes that a product can be described in terms of a set of multidimensional attribute profiles and that consumers’ choice behavior reflects their preferences for and overall judgment of that set of profiles (Huang and Fu, 1995). According to the prerequisite of selecting the proper preference model, conjoint analysis uses the structural design of a certain factor to carry out different combinations of multiple attribute standards of the products (i.e. every kind of combination is a kind of product idea, and the tester is a stimulus); these combinations provoke the overall assessment value that consumers have for a certain stimulus. A decompositional approach can estimate the part-worth value of every attribute standard and thereby decompose consumers’ preference structure for that product. Green and Wind (1973) suggest that two approaches, namely, noncompensatory and compensatory, can model how consumers process information to arrive at a choice among alternatives. Noncompensatory models assume that people evaluate alternatives on an attribute-by-attribute basis, such that substitutions or trade-offs between attributes do not exist (Huang and Fu, 1995). In contrast to the compositional approach of expectancy value models, conjoint analysis is decompositional and estimates the relative importance of a product’s multidimensional attributes by decomposing a consumer’s overall or global judgment about a set of alternatives into separate and compatible utility scales, after which the original global judgment can be reconstituted (Jaeger et al., 2001). Compensatory models also assume that all multidimensional profiles or alternatives ultimately can be described in terms of single utility numbers, commensurate with one another. In several classifications, researchers use the additive utility model (Gineo, 1990; Halbrendt et al., 1991; Manalo, 1990), which assumes that a respondent adds the values for each attribute to obtain a total value for a combination of attributes. Thus, the total utility of any defined stimulus equals the sum of its parts (Hair et al., 2006). The framework of additive utility also offers several models that can represent the preference function for empirical analysis. Hair et al. (2006) suggest that part-worth functions (i.e. the set of decomposed utility values) provide the most efficient and reliable approach from a statistical estimation perspective. Furthermore, part-worth models provide more flexibility in handing different types of variables. A detailed formulation of the part-worth model can stated as follows: Zj ¼

P X

f i ð yji Þ;

j ¼ 1; 2; . . . ; n;

ð1Þ

i¼1

where Zj is the consumer’s preference rating of the jth stimulus; fi is a function representing the part-worth of each of j different levels of the stimulus object, yji for the

IJCST 22,2/3

216

ith attribute; and yji is the level of the ith attribute for the jth stimulus object. In practice, fi( yji) is estimated for only a few levels of yji depending on the research design and the part-worths for intermediate levels of yji are obtained by linear interpolation. The part-worth function is represented by a piecewise linear curve that approximates any arbitrary shape of the preference function. Furthermore, let X denote a particular alternative that can be described as an ordered p-tuple of p attributes; then: X ¼ ðX 1 ; X 2 ; . . . ; X p Þ; whose component X i ði ¼ 1; 2; . . . ; pÞ represents the level of the alternative X on the ith attribute. The component utility of X i ; U i ðX i Þ is defined as: U i ðX i Þ ¼ U i ðX i1 Þ þ U i ðX i2 Þ þ · · · þ U i ðX ik Þ;

ð2Þ

where U i ðX im Þ; m ¼ 1, 2, . . . , k, is the utility of X with respect to the mth level of the ith attribute. The additive model posits that the total utility for an alternative X, U(X), can be expressed as: U ðXÞ ¼ U 1 ðX 1 Þ þ U 2 ðX 2 Þ þ U 3 ðX 3 Þ þ · · · þ U p ðX p Þ

ð3Þ

equation (1) provides a means to estimate the importance and contribution of each attribute to the total utility of an alternative as expressed in equation (3). Large utilities are assigned to the most preferred levels, and small utilities are assigned to the least preferred levels. The attributes with the largest utilities range are considered the most important in predicting preference, which also represents the optimum product design. 4. Empirical application This empirical process involves: . determining the attributes and levels; . compiling the stimuli (profiles of attributes) to present to respondents; . creating the sampling design; and . determining the reliability and validity. 4.1 Selecting and defining factors and levels With regard to product factors and levels selection, this study mainly considers the following principles: first, the factors and levels must be easily communicated for a realistic evaluation. Second, the factors and levels must be realistic, meaning the attributes must be distinct and represent a concept that can be precisely implemented (Hair et al., 2006). Third, all factors selected must have high levels of consumer attribute recognition and cover the spectrum of levels, which introduces greater potential heterogeneity into the factor choice of interest. The steps associated with selecting factors and standards are as follows: first, focus group interviews with consumers indicate the accessories and evaluation standards they consider when purchasing bras. Second, three interviews with different consumer groups, which included 35 consumers in total, identify 18 primary attributes. Because the resulting analysis would be complex, the attributes should be redefined on the basis of the mean and operational considerations. Three attributes are excluded because they relate closely to others, and seven attributes appear inconsistent with

aspects connected to bra design. The result of these steps indicates the bra attributes that most of consumers’ value: fabric, color, lining, neckline design, cut and sewn cups, lace, shoulder strap-detachable, and style. Thus, the evaluation standards that consumers consider mainly involve comfort, function, appearance, style, sexiness, and practicality. Interviews with several professional designers, each of whom have significant bra design experience and work for underwear brands, indicate the current design direction of bra accessories. These experts indicate that fabric development tends to focus on human-made fiber, cotton/cotton blend, natural fiber, and synthetic-stretch; the cut and sewn cups development process concentrates on two and three pieces sewn and seamless; lace development mainly focuses on the trim, overlaying the cups, and details on the V of the bra; the neckline design entails mostly low-cut plunge and sweetheart; and finally, shoulder strap style development efforts focus on two-strap, halter, and strapless. Therefore, this study identifies eight important product attributes that affect purchase decisions for bras, as Table I details. For each attribute, the specific levels are representative of the characteristics of product differentiations available in the market. In this part, respondents will also consider the importance of each attribute to indicate how important each attribute is in their evaluation criterion. Based on the specific attribute of the bra product, respondents are asked to fill in the degree of importance for each of the item below based on their opinion (1 – unimportant, 5 – extremely important).

Codesign in new bra product innovations 217

4.2 Design stimuli An innovative product design must consider the product’s critical attributes. This study addresses eight important attributes of underwear products; using the selectable design Attributes

Levels

1. Fabric

Cotton/cotton blend Silk Nylon/spandex Two pieces sewn Three pieces sewn Seamless Present Absent Black Lavender Pink Two-strap Halter Strapless Present Absent Lace details on the V of the bra Overlaying the cups Absent Low-cut plunge Sweetheart

2. Cut and sewn cups 3. Shoulder strap-detachable 4. Color 5. Shoulder strap style 6. Lining 7. Lace 8. Neckline design

Table I. Attributes and levels for conjoint analysis

IJCST 22,2/3

218

technology provided by the manufacturer, consumers select levels of these attributes to arrange the optimum product. For the experimental design, this study adopts a full-profile method, which can reduce the number of comparisons through fractional factorial design. A common approach selects orthogonal combinations of attribute levels to reduce the number of product bundles that respondents must evaluate while also enabling measures of the independent contribution of each attribute to the utility function. Orthogonal arrays preclude collinearity between attributes, such that the analysis obtains more efficient and consistent parameter estimates in the empirical estimation of the utility model (Huang and Fu, 1995). The advantage of this design is that it simplifies the ranking task for the respondents and potentially increases the reliability of the results. An orthogonal array identifies a subset of 18 stimuli to structure the interviews for data collection (the partition stimuli appear in Figure 1). In this part, respondents are asked to evaluate 18 bra profiles consisting of the eight attributes used in Table I and each attribute has only one correspondent level. If respondents assign ten points to a profile, respondents will most probably select this bra for their next bra choosing. If respondents assign zero point to a profile, respondents will most probably never select this bra for their next bra choosing. 4.3 Sampling design University students serve as respondents for this study, mainly because 72 percent of consumers between the ages of 20 and 29 years purchase new bras every three months, and among this age cohort, 65.5 percent are college students (ICT Life Style Research Center, 2004). Thus, university students represent the main target market and economic concept. Most university students in Taiwan have experience with buying bras and should be able to differentiate the various product attributes. Furthermore, budgetary constraints prevent representative sampling, so this study employs matched samples of undergraduate student respondents. The sampling method selects and adopts convenience sampling to obtain the samples. Among the various samples, 320 questionnaires were distributed in total, of which 256 were returned. After the

Figure 1. Example evaluation form describing possible bra profiles

Fabric: Nylon/spandex. Cut and sewn cups: Three pieces sewn. Color: Black; Lining: Present. Shoulder strap detachable: Absent. Shoulder strap style: Two-strap

Fabric: Cotton/cotton blend. Cut and sewn cups: Seamless. Color: Lavender; Lining: Present. Shoulder strap detachable: Present. Shoulder strap style: Halter. Lace: Details on the V of the bra

elimination of 42 invalid questionnaires, 214 valid questionnaires remain, for a high-retrieval rate of 80.0 percent and a valid return rate of 66.87 percent. 4.4 Reliability and validity The reliability measure relies on Cronbach’s coefficient a, calculated for all items to assess the internal consistency of the model variables. According to Price and Mueller (1986), a standard coefficient a of 0.60 or higher generally is acceptable for a measure. The value of the Cronbach’s coefficient a of the observed stimuli is 0.887, which indicates that the research achieves good consistency. The validity test considers external and predictive validity (Cattin and Wittink, 1982). Generally, in conjoint studies, internal validity concerns about how good a model is, as a tool of utility prediction within the system. External validity concerns about how good a model is in terms of the extent to which the predicted model can be applied to and across populations outside the system (Hu, 1994). Holdout stimuli compare the actual and predicted preference judgments, and the results show that most of the survey respondents (122) indicate measures of validity, at least sometimes. The predictive validity, computed independently for each respondent, relies on a preference model of the product-moment correlations between the respondent’s 18 calibration profile evaluations and the predictions made, by fitting the calibration model of interest and computing the mean for each preference model across the 214 respondents. The mean correlation shows the conjoint model to be the most effective in terms of predictive ability (r ¼ 0.735). 4.5 Analysis and results Table II displays the means for the bra attributes for all eight items. All respondents attach relatively high levels of importance to the eight product attribute variables. With respect to the fabric, all respondents prefer “comfort,” “practicality,” and “function.” On the color attribute, all respondents prefer “appearance” and “style.” For neckline design, they choose “comfort” and “appearance,” and for cut and sewn cups, they similarly prefer “comfort,” “appearance,” and “function.” “Comfort” is the most selected for the lining, and for the lace attribute, they reveal a preference for “appearance,” “sexy,” and “style.” With regard to the shoulder straps, respondents prefer “comfort” and “appearance,” and for the detachable shoulder strap, they want “practicality” and “function.” 4.5.1 Attributes and evaluation criterion fit analysis. To identify the evaluation criterion on which respondents base their perceptions of attributes, the exploratory research identifies eight relevant attributes of six evaluative criteria. Next, a group of respondents rates, on a seven-point scale, each of the attributes on the six evaluative criteria. The position of each attribute in the perceptual space represents the average factor score for that attribute. The vectors inserted into the reduced perception space that depicts both the attribute group centroids and overlaps among groups enable the projections of group means on each vector to reflect the relative prominence of the predictor variable in that group. This perceptual map then reveals that each attribute gets projected onto each evaluative criterion vector, which shows the degree of similarity and quality direction of a particular evaluative criterion, as shown in Figure 2. Dimension 1 contains three attribute vectors – practicality, function, and comfort – that cluster relatively closer together. Therefore, on this dimension, consumers inherently seek this attribute; this dimension is designated the sense-seeking group.

Codesign in new bra product innovations 219

1.21 1.59 1.65 1.79 1.93 1.46

6.35 5.33 4.85 4.39 4.32 5.95

Comfort Function Appearance Style Sexy Practicality

Table II. Summary statistics of the bra product attributes

Evaluation criterion 4.04 3.26 5.87 5.45 4.89 4.08

2.16 1.87 1.35 1.53 1.80 2.08

Color Mean SD 5.44 4.74 5.27 4.9 4.56 4.63

1.76 1.71 1.39 1.58 1.75 1.76

6.04 5.19 5.22 4.76 4.30 4.99

5.07 4.82 3.82 3.10 3.39 4.59

1.95 1.98 1.94 1.72 1.89 2.01

Lining Mean SD

Attribute

1.39 1.55 1.52 1.63 1.75 1.71

Cut and sewn cups Mean SD

4.0 2.93 4.87 4.43 4.46 3.02

2.14 1.72 1.92 1.90 1.92 1.91

Lace Mean SD

5.10 4.74 5.09 4.65 4.29 4.64

1.97 1.92 1.77 1.80 1.83 2.03

Shoulder strap style Mean SD

220

Fabric Mean SD

Neckline design Mean SD

4.93 5.15 4.99 4.34 3.99 5.26

1.96 1.81 1.79 1.86 1.91 1.89

Shoulder strap-detachable Mean SD

IJCST 22,2/3

3

ND: Neckline design SSD: Shoulder strap detachable SSS: Shoulder strap style

Style Appearance

2

Sexy 1 * Color 0

Codesign in new bra product innovations 221

Practicality Comfort Cut and sewn * ND * Function * Fabric * SSS SSD *

* Lace –1 * Lining

–2

–3

–2

–1

0

1

2

3

Dimension 2 focuses mainly on the attribute vector of appearance, sexy, and style; it represents sensibility preference. The map in Figure 2 clearly shows that consumers regard fabric as the best element if they reside in the sense-seeking group. It has high-attribute popularity, practicality, good functions, and excellent comfort. The next most popular attributes are cut and sewn cups, shoulder strap-detachable, neckline design, and shoulder strap style. The other attributes fall distant from the direction of this evaluative criterion vector, so they are considered of inferior quality and constitute the group that is more similar. That is, consumers believe the other attributes do not strive sufficiently with respect to these three evaluation criteria, which results in negative evaluations. In terms of sensibility seeking, color is rated the best, and neckline design, cut and sewn cups, and shoulder strap style are next. Most of the other attributes are distant from the direction of this evaluative criterion vector, so they are considered of inferior quality and appear in the more similar group. In the diagram, most criteria appear centralized in terms of fabric, color, cut and sewn cups, shoulder strap-detachable, and neckline design, far from the other attributes. However, these attributes have their own advantageous position in the map, which suggests opportunities for new products. Furthermore, the perceptual map identifies that the new product should select specific product characteristics, because the conjoint analysis reveals the linkage of features to preferences and of features to perceptions. Conceptually, this approach can try all feature combinations to determine which is the most preferred. 4.5.2 Conjoint analysis. As noted, conjoint analysis reveals the best design for personalization, and it converts preference ratings into utilities for hypothetical products. Utilities can be divided into part-worth utilities associated with each attribute level of the product, such that the utility value represents a respondent’s overall preference for

Figure 2. Perceptual maps of bra attribute sets

IJCST 22,2/3

222

a particular product profile. A part-worth indicates the preference or utility associated with a specific level of an attribute, estimated for each respondent in a conjoint study. The utilities then are the predicted values from the regression, and the part-worth utilities are the regression coefficients. In this way, conjoint analysis reveals the relative importance of each attribute by considering how much difference each attribute makes to the total utility of a product. In addition to assisting in the development of new products, this result also can contribute to the design of personalized bra products. As Table III shows, the product most suitable for a personalized product for Subject 1 is Product 1 (silk fabric; seamless cut and sewn cups; detachable, halter shoulder strap; black; with lining; lace overlaying the cups; and low-cut plunge neckline), according to her individual attribute importance values: fabric (12.59), color (25.23), neckline design (7.98), cut and sewn cups (14.09), lining (9.76), lace (7.97), shoulder strap-detachable (9.16), and shoulder strap style (13.21). Color is the most important attribute in predicting the preference of Subject 1 but is not very important for the other subjects. Although conjoint analyses refer to individual subjects, the usual goal is to summarize the analysis across subjects, in line with the output in Table IV. The procedure represents the partial utility values for each level of each attribute and a percentage measurement of their importance at both individual and aggregate levels. The most important attribute is shoulder strap style (21.26 percent), followed by bra color (14.66 percent), lace (13.69 percent), and fabric (12.13 percent). Intermediate importance marks lining (10.23 percent), neckline design (10.12 percent), cut and sewn cups (10.11 percent), and shoulder strap-detachable (7.82 percent). As Table IV indicates, the most preferred combination of attributes for all respondents is a cotton/cotton blend fabric bra that uses seamless sewing, features a detachable shoulder strap and two-strap style, has lining, is lavender in color, includes lace details on the V of the bra, and provides a low-cut plunge neckline. This combination gives consumers the highest total worth value. 5. Discussion 5.1 Conclusions Many observers believe that product innovation begins with a new product idea, which causes most innovation studies to focus on new concept generation. However, whereas that approach isolates users’ problems and then seeks solutions to those problems, a codesign approach identifies the product concepts that most likely have value for the consumer. A product with a good core benefit proposition is critical for effective new product development. A perceptual map analysis can reveal the dimensions that consumers use to evaluate a new product and where existing products lie on these dimensions. Product concept creativity thus may be triggered by the mere listing of each feature, because designers instinctively should concentrate more on how to change that feature and understand far more about the product than they otherwise would have. The important attributes to consider when designing bras include fabric, color, lining, neckline design, cut and sewn cups, lace, detachable shoulder straps, and strap style. With regard to the material, it is important that consumers feel comfortable when wearing the bra, it provides easy care practicality, and it ensures a durable bra. The color should be bright and aesthetically pleasing. The neckline design should emphasize comfort, and its appearance should be centralized and firm. On the cut and sewn cups attribute, consumers demands that the bra and their breasts fit perfectly,

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Product

Fabric

Silk Silk Cotton Nylon Cotton Cotton Silk Silk Nylon Nylon Nylon Nylon Silk Cotton Cotton Silk Nylon Cotton

Utility

9.53 7.84 7.78 7.17 6.95 6.92 6.91 5.24 4.79 3.95 3.82 3.77 3.41 2.91 2.48 2.08 1.50 0.97

Black Black Black Black Pink Black Pink Lavender Pink Pink Lavender Black Pink Lavender Lavender Lavender Lavender Lavender

Color Low-cut plunge Sweetheart Sweetheart Sweetheart Low-cut plunge Sweetheart Sweetheart Low-cut plunge Sweetheart Low-cut plunge Sweetheart Low-cut plunge Sweetheart Sweetheart Low-cut plunge Low-cut plunge Sweetheart Sweetheart

Neckline design Seamless Three pieces Two pieces Two pieces Seamless Seamless Two pieces Two pieces Seamless Two pieces Seamless Three pieces Three pieces Seamless Three pieces Seamless Three pieces Two pieces

Cut and sewn cups Present Absent Absent Absent Present Present Present Present Present Present Absent Present Absent Absent Present Absent Present Present

Lining Overlay Absent Absent V bar Absent V bar Overlay V bar V bar Absent Overlay Overlay V bar Absent V bar Absent Absent Overlay

Lace

bar

bar bar

bar

bar

Present Absent Present Present Present Absent Absent Present Absent Present Present Present Present Present Present Absent Absent Absent

Shoulder strap-detachable

Halt Two-strap Halt A halt Two-strap Strapless Two-strap Halt Halt Strapless Two-strap Strapless Strapless Strapless Two-strap Strapless Halt Strapless

Shoulder strap style

Codesign in new bra product innovations 223

Table III. Utilities for Subject 1

IJCST 22,2/3

IMP (%)

Attributes

Levels

12.13

1. Fabric

10.11

2. Cut and sewn cups

7.82

3. Shoulder strap-detachable

14.66

4. Color

21.26

5. Shoulder strap style

10.23

6. Living

13.69

7. Lace

10.12

8. Neckline design

Cotton/cotton blend Silk Nylon/spandex Two pieces sewn Three pieces sewn Seamless Present Absent Black Lavender Pink Two-strap Halter Strapless Present Absent Lace details on the V of the bra Overlaying the cups Absent Low-cut plunge Sweetheart

224

Table IV. Estimated attribute-level utility and relative attribute importance for bras

Utility 0.1347 0.0402 2 0.1750 2 0.0357 2 0.1251 0.1608 0.0215 2 0.0215 0.1320 2 0.1599 0.0279 0.7392 0.0471 2 0.7863 0.0092 2 0.0092 0.1464 2 0.2348 0.0883 0.1476 2 0.1476

such that the bra covers their breasts in full and provides comfort, while also providing a centralized and firm appearance and a body shaping function. The design of the lining focuses on comfort, whereas the lace should be classically elegant in appearance, sexy, and aesthetically pleasing. With regard to the shoulder straps, they should remain in contact with the skin but not leave any marks, with a bright appearance that can attract others’ attention. They also should offer detachable functionality to meet the needs of different occasions, while still offering a body shaping function. Since users regard these attributes of bra design as particularly important, designers should focus on them to improve their bra offerings. Development trends change fast in the underwear industry. Marketers are not always certain how the market will accept the latest trend and may find it useful to use the attribute importance construct as a means to segment markets. Moreover, to master new development trends, new product design should consider both rational viewpoints and perceptual demands. This study considers consumers’ viewpoints and reveals rational and perceptual design features of new bra products. Rationally, the bra design must consider comfort, function, and practicality; perceptually, it should focus on the integration and arrangement of color, appearance, and style. In contrast with considerations of only single viewpoints, these perceptions appear significantly different but align with actual demand. Thus, conjoint analysis enables actual implementation of the codesign concept: consumers play the role of designer and use the assessment criteria to evaluate the product attributes, which leads to a product profile on which new product designs should be based to achieve the best product. Such a method can benefit new product design as well as promote consumer satisfaction and loyalty through codesign. Consequently, through integration with this quantitative method, product design can reach zero risk or zero deviation.

The results provide a useful source of information for product managers, who should consider the use of codesign to design the best products for individual consumers and decrease the risk of design failure, as well as promote consumer loyalty and satisfaction toward the product. The study provides a unique method to understand the new product codesign structure and make bra product design decisions that integrate optimum individualized design.

Codesign in new bra product innovations 225

5.2 Limitations and future research This research focuses solely on Chinese bras and relies on young females, which limits its generalizability. More heterogeneous consumer samples would amplify and verify any cultural impacts on bra codesign. Future research needs to link specific target markets for codesign brassiere and entrance strategies for companies seeking to fill niche. In full-profile conjoint analysis, the relatively large data requirements that are needed to estimate individual utilities, are becoming increasingly burdensome on respondents. A hybrid model approach to conjoint offers a simple but robust approach that uses self-explicated data to obtain a preliminary set of individualized part-worths for each respondent. In addition, each respondent provides full-profile evaluations for a limited number of stimulus profiles. In practical situations, the full-profile method is difficult to execute with the large number of profiles required by larger numbers of attributes and/or levels within attributes. Future research could use hybrid model to provide a practical alternative to the full-profile method. It should be clear that as photographs of the physical prototype stimuli were used as the pictorial images, the present findings pertain to photographic images only and not pictorial representations generated using computer-aided design (CAD) techniques. Future work is required o establish whether CAD images are also acceptable. The codesign process involves a high level of consumer input and stimulation. Some participants expressed concern for their ability to act as designers. They did not feel confident in their ability to put together a bra design. Therefore, this suggests that marketers should focus on the level of consumer involvement and the control over the final product. This study presents an early attempt to understand the emerging paradigm of codesign for consumers, and manufacturers’ perspectives. The study provides an initial blueprint for understanding potential implementation of codesign for the bra industry. References Anderson-Connell, L.J., Ulrich, P.V. and Brannon, E.L. (2002), “A consumer-driven model for mass customization in the apparel market”, Journal of Fashion Marketing & Management, Vol. 6 No. 3, pp. 240-58. Beane, T.P. and Ennis, D.M. (1987), “Market segmentation: a review”, European Journal of Marketing, Vol. 21 No. 5, pp. 20-42. Beaudoin, P., Moore, M.A. and Goldsmith, R.E. (2000), “Fashion leaders’ and followers’ attitudes toward buying domestic and imported apparel”, Clothing and Textiles Research Journal, Vol. 18 No. 1, pp. 56-64.

IJCST 22,2/3

226

Berger, C., Mo¨slein, K., Piller, F. and Reichwald, R. (2005), “Co-designing modes of cooperation at the customer interface: learning from exploratory research”, European Management Review, Vol. 2 No. 1, pp. 70-87. Cattin, P. and Wittink, D. (1982), “Commercial use of conjoint analysis: a survey”, Journal of Marketing, Vol. 46 No. 3, pp. 44-53. Fiore, A.M., Lee, S.-E. and Kunz, G. (2004), “Individual differences, motivations, and willingness to use a mass customization option for fashion products”, European Journal of Marketing, Vol. 38 No. 7, pp. 835-49. Fritz, A.M. and Gardner, A.P.W. (1988), “Lightweight fabrics – a psychophysical study”, Proceedings of the Textile Institute, Textile Institute, Manchester. Gineo, W.M. (1990), “A conjoint/logit analysis of nursery stock purchases”, Northeastern Journal of Agricultural and Resource Economics, Vol. 19 No. 1, pp. 49-58. Green, P.E. and Srinivasan, V. (1990), “Conjoint analysis in marketing: new developments with implications for research and practice”, Journal of Marketing, Vol. 54 No. 4, pp. 3-19. Green, P.E. and Wind, Y. (1973), Multiattribute Decisions in Marketing: A Measurement Approach, Dryden Press, Hinsdale, IL. Gustafsson, A., Herrmann, A. and Huber, F. (2003), Conjoint Measurement: Methods and Applications, 3rd ed., Springer, Berlin. Hair, J.F., Black, B., Babin, B., Anderson, R.E. and Tatham, R.L. (2006), Multivariate Data Analysis, 6th ed., Prentice-Hall, Upper Saddle River, NJ. Halbrendt, C.K., Wirth, F.F. and Vaughn, G.F. (1991), “Conjoint analysis of the Mid-Atlantic food-fish market for farm-raised hybrid striped bass”, Southern Journal of Agricultural Economics, Vol. 23 No. 1, pp. 155-63. Hardaker, C.H.M. and Fozzard, G.J.W. (1997), “Communications: the bra design process – a study of professional practice”, International Journal of Clothing Science & Technology, Vol. 9 No. 4, pp. 311-25. Hill, C.J. and Garner, S.J. (2001), “Consumer perceptions of the efficacy of price as an index of professional service quality”, Services Marketing Quarterly, Vol. 22 No. 2, pp. 29-48. Hu, C. (1994), “Measuring meeting planner’s preferences of hotel site selection: a hybrid conjoint analysis approach”, unpublished Master’s thesis, Purdue University, West Lafayette, IN. Huang, C.L. and Fu, J. (1995), “Conjoint analysis of consumer preferences and evaluations of a processed meat”, Journal of International Food and Agribusiness Marketing, Vol. 7 No. 1, pp. 35-53. ICT Life Style Research Center (2004), Almance of ICT Marketing Data in the Republic of China on Taiwan, Gsharp Corporation, Taipei. Jaeger, S.R., Hedderley, D. and MacFie, H.J.H. (2001), “Methodological issues in conjoint analysis: a case study”, European Journal of Marketing, Vol. 35 Nos 11/12, pp. 1217-37. Lee, C.-W. and Liao, C.-S. (2007), “A cross-cultural perspective of the effect of extrinsic and intrinsic cues on perceived quality”, International Journal of Business and Strategy, Vol. 8 No. 2, pp. 1-24. Manalo, A.B. (1990), “Assessing the importance of apple attributes: an agricultural application of conjoint analysis”, Northeastern Journal of Agricultural and Resource Economics, Vol. 19 No. 2, pp. 118-24. May-Plumlee, T. and Little, T.J. (2006), “Proactive product development integrating consumer requirements”, International Journal of Clothing Science & Technology, Vol. 18 No. 1, pp. 53-66.

North, E.J. and De Vos, R. (2002), “The use of conjoint analysis to determine consumer buying preferences: a literature review”, Journal of Family Ecology and Consumer Sciences, Vol. 30, pp. 32-9. Price, J.L. and Mueller, C.W. (1986), Handbook of Organizational Measurement, Pitman, Marshfield, MA. Schutz, H.G., Cardello, A.V. and Winternalter, C. (2005), “Perceptions of fiber uses and the factors contributing to military clothing comfort and satisfaction”, Textile Research Journal, Vol. 75 No. 3, pp. 223-32. Textile Industry Affairs (2008), “Independent consumer research on apparel characteristics influencing purchase decisions”, available at: www.carelabels.com/apparel1.pdf (accessed April 9, 2008). Wang, J.P. and Zhang, W.Y. (2007), “An approach to predicting bra cup dart quantity in the 3D virtual environment”, International Journal of Clothing Science & Technology, Vol. 19 No. 5, pp. 361-73. Wheatley, J.J., Chiu, J.S.Y. and Goldman, A. (1981), “Physical quality, price and perceptions of product quality: implications for retailers”, Journal of Retailing, Vol. 57 No. 2, pp. 100-16. Wikipedia (2008), “Co-design”, available at: http://209.85.175.104/search?q¼cache:s3EUyQfQmKgJ: en.wikipedia.org/wiki/Co-DesignþCo-Design&hl¼zh-TW&ct¼clnk&cd¼1&gl¼tw (accessed March 19, 2008). Wittink, D.R. and Cattin, P. (1989), “Commercial use of conjoint analysis: an update”, Journal of Marketing, Vol. 53 No. 3, pp. 91-6. Further reading Lilien, G.L. and Kotler, P. (1983), Marketing Decision Making: A Model-Building Approach, Harper & Row, New York, NY. Corresponding author Chi-Shun Liao can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

Codesign in new bra product innovations 227

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

IJCST 22,4

Hand evaluation and formability of Japanese traditional Chirimen fabric

234

Takako Inoue Sugiyama Jogakuen University, Nagoya, Japan, and

Received 9 November 2009 Accepted 14 January 2010

Masako Niwa Nara Women’s University, Nara, Japan Abstract Purpose – Japanese traditional Chirimen fabrics are used for making kimonos, which have a fixed structure and are worn in very particular ways. These fabrics have also been used as dress fabrics in recent years. The purpose of this paper is to investigate the characteristics of the mechanical properties of various types of Chirimen to clarify differences in their hand value (HV) and clothing appearance. Design/methodology/approach – Chirimen fabrics were collected from the largest producing area, the Tango district, plus silk Chirimen and 40 polyester Chirimen samples, resulting in a total of 311 samples. The mechanical properties, HVs, and formability of Chirimen fabrics used for kimono fabrics were compared to those of Western fabrics, and their unique features were clarified. Findings – Values of the weft direction of bending properties of all Chirimen groups, men and women’s suit fabrics, and dress shirt fabrics were at the same level. A significant feature of the mechanical parameters of each Chirimen group (excluding logSP which are compound values of bending properties and shearing properties) was that they were in the range for ideal men’s suiting zone. HV KOSHI of Chirimen is found to be closely related to the bending properties, thickness and weight of the fabric, and HV TEKASA of Chirimen is found to be closely related to the thickness and weight of the fabric. Originality/value – This paper clarifies Chirimen’s mechanical properties which contribute to traditional subjective evaluation by fabric experts. Keywords Japan, Mechanical properties of materials, Fabric testing, Silks, Textile testing, Regression analysis Paper type Research paper

International Journal of Clothing Science and Technology Vol. 22 No. 4, 2010 pp. 234-247 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011048277

Introduction Each region of Japan has unique textile weaves and dyeing methods (Tomiyama and Ohno, 1967) for traditional fabrics. The textile weave and the dyeing method are two sides of the same coin: either the weaving or the dyeing can be done first. Consequently, there are both fiber- or yarn- and piece-dyed fabrics. The Japanese traditional kimono fabrics called Chirimen are produced in Tango, Nagahama, Hokuriku, and Gifu, as well as other districts, and they are typical woven piece-dyed fabrics produced in great amounts throughout Japan. Chirimen is the generic name for silk fabric in which rightand left-laid hard-twist yarn is alternately woven to make weft yarn. There are crimps on the surface, and these crimps create a unique hanging down feeling and tinctorial effect. Chirimen is considered a very high-grade silk fabric, and is used in formal kimono fabrics. Artificial fiber has been used in Chirimen fabrics in recent years.

In this study, the mechanical properties, hand values (HVs), and formability of Chirimen fabrics used for kimono fabrics were compared those of Western fabrics, and their unique features were clarified. The goal of this study is to clarify the mechanical properties which contribute to traditional subjective evaluation by fabric experts. Test sample collection Chirimen fabrics were collected from largest producing area, the Tango district, in 1981 (Komatsu and Niwa, 1981), and silk Chirimen and 40 Polyester Chirimen samples were added to the samples in recent years, resulting in a total of 311 samples. The details are shown in Table I. Chirimen are classified by difference in yarn, fabric construction, and the size of the rugged crimps on the surface (Nakae, 1993). Among the classifications of Chirimen fabrics examined are: . Hitokoshi Chirimen (crimp is minute, and filament yarn that is not twisted is used for the warp yarn; this is the typical Chirimen, with right- and left-laid hard twist yarn alternately woven to make weft yarn). . Kodai Chirimen (filament yarn that is not twisted is used for the warp yarn; thicker hard-twist yarn than that used in Hitokoshi Chirimen is alternately woven right- and left-laid to make weft yarn, namely “two jump” yarn, right/right-laid and left/left-laid hard-twist yarn; there are larger crimps than in Hitokoshi Chirimen; this is also called Futakoshi Chirimen, Futakoshi meaning “two jumps”). . Kawari Muji Chirimen (crimp is more minute than that of Hitokoshi Chirimen; hard twist yarn called Chirimen Yoko (Yoko meaning “weft”) is not used, rather, fancy twist yarn (yarn twisted differently from the usual way, perhaps one yarn made from non-twisted yarn and right-laid yarn, another yarn made from non-twisted yarn and right-laid yarn are twisted to make one yarn) is used; consequently, the degree of shrinkage is small; it has been produced in greater amounts than Hitokoshi Chirimen in recent years). . Mon Chirimen (Chirimen which has warp yarn in plain weave fabric at the surface to highlight the woven pattern; there are different varieties such as Mon Isho Chirimen and Mon Rinzu Chirimen,, etc.). . Fuutsuu Chirimen (woven pattern in the cloth is double-weave, using warp and weft yarn; the pattern is woven in high relief.). In this study, each kind of Chirimen was examined. Fiber

Kind of Chirimen Hitokoshi · Kodai Chirimen Kawari Muji Chirimen

Silk Mon Chirimen

Hitokoshi Chirimen Kodai Chirimen Muji(Mon) Isho Chirimen Rinzu Chirimen Fuutsuu Chirimen Others

Polyester Total

Japanese traditional Chirimen fabric 235

The number of Chirimen 78 26 14 76 63 6 8 40 311

Table I. Chirimen samples

IJCST 22,4

236

Measurement of fabric mechanical properties and the subjective hand-evaluation method Fabric mechanical properties were measured by a KES-FB-AUTO system (Kawabata et al., 1990) under the standard condition (Kawabata, 1980) shown in Appendix, Table AI. The mechanical parameters and three basic components of tailorability were calculated to make a total appearance value prediction (Kawabata and Niwa, 1989). The mechanical parameters are as follows: (Formability) . Weft-directional extensibility: EM2 ð1Þ log10 EL2 ¼ log10 LT2 .

Effective bending stiffness in weft-bending mode: log10 BS2 ¼ log10 ½M 2ð1Þ þ HB2
; M 2ð1Þ ¼ B2K; K ¼ 1

.

ð2Þ

Effective shear stiffness: log10 SS ¼ log10 ½Fsð1Þ þ HG5
; Fsð1Þ ¼ Gf; f ¼ 18

ð3Þ

(Elastic potential) .

.

Bending elastic potential per unit area at K ¼ 2.5 cm2 1:
 ð2:5 2 HB=BÞ2 log10 BP ¼ log10 B 2 Shear elastic potential per unit area at f ¼ 88:
 0 2 2HG 2 2HG5 0 ð8 2 HG=G Þ ; G0 ¼ G þ log10 SP ¼ log10 G 5 2

(Drape) Bending stiffness relating to drape : Shea stiffness relating to drape : 21

pffiffiffi 3 pffiffiffi SS W 3 BS W

ð4Þ

ð5Þ

ð6Þ ð7Þ

where, K, bending curvature (cm ), f, shear angle (degree), M, elastic bending moment (gf cm cm2 1), Fs, elastic shear force (gf cm2 1). Sensory tests were performed for the subjective evaluation of the HV of Chirimen. A standard sample was assumed to represent the standard feel of Chirimen; comparing Chirimen samples with the standard sample, we ranked the Chirimen samples by how strong they felt to the touch, using a scale of 0-10, with the standard sample being 5. This subjective hand-evaluation provided an indication of KOSHI (stiffness) and TEKASA (hand quantity), both of which influence the quality of Chirimen. TEKASA is defined as a sense of great bulk and the rich feeling of cloth with substantial “give” to the touch, a feeling of thickness and warmth as well as elasticity under pressure

(Kawabata, 1980). Judges were four experienced technical engineers who were well acquainted with the silk Chirimen fabrics of the special production site, from the Kyoto Prefectural Institute of Northern Industry. Results and discussion (1) Characteristics of mechanical properties In order to express the features of the mechanical properties, a standardized The Hand Evaluation and Standardization Committee data chart, 201 LDYM, has been created using the mean value and the standard deviation for women’s suiting fabric (Kawabata, 1980). In this study, we added warp and weft direction to tensile properties and bending properties as shown in Figure 1, because Kimono fabrics wrap around a cylindrical body beautifully in the direction of the weft direction and assume a form when worn that hangs down in the direction of the warp. The mean value and the standard deviation for men’s suiting fabrics, dress-shirt fabrics, silk Chirimen, and polyester Chirimen were used to represent the features of Chirimen. Men’s suiting fabrics and dress shirts fabrics have narrower ranges of mechanical properties than women’s suiting fabrics. The distinctive features of silk Chirimen are its low values of bending properties, shearing properties, and thickness & weight, but the values for the weft direction of tensile properties EM2, LT, RT and the bending properties B2 and 2HB have overlapping ranges for men’s and women’s suit-fabric characters, and the ranges are wide. The mean deviation of coefficients of friction MMD and geometrical roughness SMD values are high because there are crepes in the surface, but the coefficients of friction MIU values are low and smooth. The values of extensibility of warp and weft directions EM1, EM2, and the tensile energy WT of polyester Chirimen are low, and its bending properties and shearing properties are close to those for women’s suit fabrics in terms of value level and range. The bending properties of weft direction and the thickness of polyester Chirimen are the similar to those of silk Chirimen. However, values of the surface properties of polyester Chirimen are slightly higher than those for silk Chirimen, and the weight of the fabric is slightly lower than that of silk Chirimen. Thus, it is clear that there are differences as well as similarities between polyester Chirimen and silk Chirimen. The data charts as shown in Figures 2 and 3 have been normalized using the mean value and the standard deviation of each mechanical property of the 271 silk Chirimen fabrics in order to represent the mechanical properties of Chirimen fabrics, and the mean value and the standard deviation of each Chirimen group are plotted. The characteristic ranges of mechanical properties are shown for each group. The values of tensile properties EM1, EM2, WT of Hitokoshi Chirimen and Kodai Chirimen are high, as are those of the surface properties and the thickness and weight; this is most likely due to the fact that there are crepes in the surface. Tensile resilience RT is low, the reason for these Chirimen shrinking easily. The values of the surface properties of Rinzu Chirimen are the lowest in the Chirimen groups, and its surfaces are the smoothest. On the other hand, Fuutsuu Chirimen is made from double-weave cloth, so the thickness values of the fabrics are high, and with the woven patterns brought into high relief, coefficient of friction MIU is also high. Regarding Kawari Muji Chirimen, its thickness and weight values are low and it has the distinctive feature of lower-value tensile properties than those of Hitokoshi Chirimen and Kodai Chirimen. Polyester Chirimen meanwhile, is not easy to stretch, the values of its bending properties of warp direction and shearing properties are high, and the values of its thickness and weight are low; it is understood

Japanese traditional Chirimen fabric 237

IJCST 22,4

HESC mechanical data chart-201 LDYM women’s suit (X-M)/s (n = 220) −4

−3

−2

−1

0

1

2

3

4

EM1

238

Tensile

EM2 1

EM

100

LT WT

0.3

0.4

0.5

0.6

0.7

0.8

0.9 100 70

RT 20

B1

30

Bending

50

60

0.1

2HB1

1 1

0.1

B2 2HB2

40

0.01

0.1

1

0.01

B

0.1

1 1

2HB 0.01

Shear

G 2HG

0.1

0.1

1

10

Surface

Compr.

2HG5 LC

10 0.0

0.2

0.4

0.6

0.8

WC RC MIU

20 0.10

30 0.15

40 0.20

50

60 0.25

70 0.30

80 0.35

0.40

MMD SMD

0.01

0.1 10

T W

Figure 1. The mechanical properties of Chirimen

10 M-s M M+s M-s M M+s Notes: Suffix 1: warp direction, Suffix 2: weft direction; : dress shirt (n = 116); : silk Chirimen (n = 271); mean value and ± standard deviation; men’s winter suit fabrics (n = 214) is indicated by the solid line; polyester Chirimen (n = 40) is indicated by the broken line

that the feature of polyester Chirimen distinguishing it from silk Chirimen is its ability to create a silhouette as clothing when used in clothing. The values of the bending properties of weft direction of all Chirimen groups are at the same level, and they have a range corresponding to those of men’s suit fabrics, women’s suit fabrics and dress shirt fabrics; the values of tensile properties are at the same level as those for men’s suit fabrics and women’s suit fabrics. This can be said to be a distinctive characteristic of Chirimen fabrics.

KESF mechanical data chart for silk chirimen fabrics (X-M ) /s

−4

−3

−2

−1

0

1

2

(n = 271)

3

4

EM1 10

Tensile

EM2 EM

Japanese traditional Chirimen fabric

1

100

0

239

LT 0.4

WT

0.6

0.8

1.0

RT B1

20

30

40

50 0.1

Bending

2HB1

0.001

Compr.

1

0.1

0.01

0.1 0.01

0.1

0.1 0.1

2HG5

Surface

1

0.1 0.01

G 2HG

0.1

0.01

2HB2

2HB Shear

0.01

B2

B

60

1 10

LC WC RC

30

40

50

60

70

80

MIU MMD SMD

0.1

0.01 10

T W 10 M-s M+s M Notes: Suffix 1: warp direction, Suffix 2: weft direction; : Hitokoshi Chirimen(n = 78); : Muji Isho Chirimen (n = 76); : Rinzu Chirimen (n = 63); mean value and ± standard deviation; polyester Chirimen (n = 40) is indicated by the broken line

(2 ) HV and formability Criteria chart for ideal men’s suiting (winter/autumn) (Kawabata et al., 2002) was normalized again using the mean value and the standard deviation of women’s suiting fabrics; the mean value and standard deviation of HV and the mechanical parameters of suit appearance of all Chirimen groups which were calculated using KN201 equation (Kawabata, 1980) are shown in Figure 4. The values of KOSHI, NUMERI (smoothness) and FUKURAMI (fullness and softness) of Chirimen are at the same level as those of dress shirt fabrics, but the value of SOFUTOSA (soft feeling) is high, at the same level

Figure 2. The mechanical properties of Chirimen

IJCST 22,4

KESF mechanical data chart for silk chirimen fabrics (X-M ) /s (n = 271) −4

−3

−2

−1

0

1

2

3

4

EM1

240

Tensile

EM2 EM

1

100

LT 0.4

WT

0.6

0.8

1.0

RT B1

20

30

40

50

60

Bending

2HB1 0.01

B2 0.01

2HB2 B 2HB

0.001

Shear Compr. Surface

1 0.01

0.1

0.01

1

0.1 0.01

G 2HG

0.1

0.1

0.1

1 0.1

2HG5

1 10

LC WC RC MIU

30 0.05

40 0.10

50 0.15

60 0.20

70 0.25

80

0.30

0.35

MMD SMD

0.01

0.1

T W

Figure 3. The mechanical properties of Chirimen

10 M-s M+s M Notes: Suffix 1: warp direction, Suffix 2: weft direction; : Kawari Chirimen(n = 14); : Kodai Chirimen (n = 26); : Fuutsuu Chirimen (n = 6); mean value and ± standard deviation; polyester Chirimen (n = 40) is indicated by the broken line

as the values of men’s suit fabrics and women’s suit fabrics. A distinctive characteristic here of Chirimen is that the values of mechanical parameters for suit appearance are within the range of those of ideal men’s suiting, excluding logSP (equation (5)). The values of the basic components of tailorability are located almost in the middle of the range of men’s suit fabrics and dress shirt fabrics. The mean value and the standard deviation of each Chirimen group are shown in Figures 5 and 6. Values of KOSHI and SOFUTOSA in each group are different, but the values of the mechanical parameters in each group are located within the range of those of ideal men’s suiting (excluding logSP).

Hand values −4s Stiffness (Koshi) Smoothness 2 (Numeri) 0 Fullness (Fukurami) Sofuto feeling 0 (Sofutosa) −2

−3s

−2s

3

4

−s

s

0

5

2

6

6

2

4

4s

8

6

4

0

3s

7

4

2

2s

Japanese traditional Chirimen fabric

9

8

10

8

10

6

241

8

Suit appearance

Mechanical parameter −4s −3s

−2s

−s

s

0

2s

4s

3s

EL2 BS2

1

100

10

SS

0 10

BP SP 3 3

0.1 100

BS/W 1.0

SS/W

1.5

2

2.0

3

4

5

Three basic components of tailorability −4s −3s −2s s Formability Elastic potential Drape −2

0

TAV −4

−2 −2

0 M-s

s

7

2s

4 2

0 M

3.0

6

0

2 0

2.5

2 M+s

3s

4s

6 4

2

8

4 4

8 6

6 6

8 8 8

M-s, M, M+s

Notes: Suffix 1: warp direction, Suffix 2: weft direction; : dress shirt (n = 116); : silk Chirimen (n = 271); mean value and ± standard deviation, men’s winter suit fabrics (n = 214) is indicated by the solid line; polyester Chirimen (n = 40) is indicated by the broken line; Perfect property zone of men’s suit is indicated by the shaded zone

Values of the basic components of tailorability are slightly different from group to group: the elastic-potential values of Hitokoshi Chirimen and Kodai Chirimen are high, indicating that they stretch easily; the drapability values of Muji Isho Chirimen are high, so it is possible that this type can create a beautiful silhouette; the formability and elastic-potential values of Muji Isho Chirimen are higher than for other Chirimen groups, explaining the reason that Muji Isho Chirimen is used to make Tomesode (married women’s formal kimono decorated with five crests and a pattern around the skirt,

Figure 4. HV and the mechanical parameters of Chirimen

IJCST 22,4

242

Hand values −4s Stiffness (Koshi) Smoothness 2 (Numeri) 0 Fullness (Fukurami) Sofuto feeling 0 (Sofutosa) −2

−3s

−2s

3

4

−s

s

0 5

2

6

9

8

6

2

4s

8

6

4

0

3s

7

4

2

2s

10

8

4

10

6

8

Suit appearance

Mechanical parameter −4s −3s

−2s

−s

s

0

2s

3s

4s

EL2 BS2

1

10

100

0.1

SS

10

BP SP

100

3 BS/ W

1.0

3 SS/ W

1.5

2

3

2.0 4

5

Three basic components of tailorability −4s −3s −2s −s Formability Elastic potential Drape−2

0

TAV −4

−2 −2

Figure 5. HV and the mechanical parameters of Chirimen

s

2

M

7

2s

4

0 0 M-s

3.0

6

0

2 0

2.5

2 M+s

4 4

3s

4s

6 4

2

8

8 6

6

8 8

6

8

Notes: Suffix 1: warp direction, Suffix 2: weft direction; : Hitokoshi Chirimen (n = 78); : Muji Isho Chirimen (n = 76); : Rinzu Chirimen (n = 63); mean value and ± standard deviation; polyester Chirimen (n = 40) is indicated by the broken line; perfect property zone of men’s suit is indicated by the shaded zone

the highest-grade kimono and the most formal ceremonial dress) and Houmongi (kimono for formal visiting, quasi-ceremonial dress decorated with a full-body pattern). (3 ) Analysis of mechanical properties affecting hand evaluation How the mechanical properties of Chirimen affect the criteria for subjective evaluation of hand evaluation was examined. First, 139 samples of Chirimen were chosen without disparity among groups from a total of 271 samples of silk Chirimen. A total

Hand values −4s Stiffness (Koshi) Smoothness 2 (Numeri) 0 Fullness (Fukurami) Sofuto feeling 0 (Sofutosa) −2

−3s

−2s

3

4

−s

s

0 5

2

6

9

8

6

2

4s

8

6

4

0

3s

7

4

2

2s

10

8

4

Japanese traditional Chirimen fabric 243

10

6

8

Suit appearance

Mechanical parameter −4s −3s

−2s

−s

s

0

2s

4s

3s

EL2 BS2

1

10

100

0.1

SS

10

BP SP 3 3

100

BS/W 1.0

SS/W

1.5

2

2.0

3

4

2.5 5

3.0

6

7

8

Three basic components of tailorability −4s

−3s

Formability Elastic potential

−2s

−s

0

Drape −2

2 0

TAV −4

−2 −2

2s

4 2

0

s

0

4s

6 4

2

3s

4

8 6

6

0 2 4 6 M-s M M+s Notes: Suffix 1: warp direction, Suffix 2: weft direction; : Kawari Chirimen (n = 14); : Kodai Chirimen (n = 26); : Fuutsuu Chirimen (n = 6); mean value and ± standard deviation; polyester Chirimen (n = 40) is indicated by the broken line; perfect property zone of men’s suit is indicated by the shaded zone

8 8 8

of 47 Hitokoshi Chirimen, seven Kodai Chirimen, seven Kawari Muji Chirimen, 40 Muji Isho Chirimen, and 38 Rinzu Chirimen were chosen. The stepwise-block regression method was applied, using 19 characteristic values of six blocks of the mechanical properties – tensile, bending, surface, shearing, compression, and thickness and weight – and the mean value of each hand evaluation value was used as the subjective value (KOSHI, and TEKASA). The regression equation is shown below:

Figure 6. HV and the mechanical parameters of Chirimen

IJCST 22,4

244

HV ¼ C 0 þ

19  X i¼1

C i1

X 2 2 M i2 X i 2 M i1 þ C i2 i si1 si2

 ð8Þ

where, C0, Ci1, Ci2, constant coefficients of the ith variable terms; Xi, mechanical property of the ith variable term; Mi1, si1, the population mean and standard deviation; Mi2,si2, the square mean and standard deviation. From this analysis, the order of the contribution to HV can be clarified, and we can discover the relationship between the criteria of subjective evaluation and the mechanical properties. The order of contribution of each characteristic block to HV and the degree of correlation between each characteristic block and HV can be apparent from this analysis, and thus it is effective for determining the relationship between the criteria of subjective evaluation and the mechanical properties of fabrics. The contributions of each characteristic block to HV were examined; the contributions to KOSHI are shown in Figure 7 and the contributions to TEKASA are shown in Figure 8. The results were as follows: . The contributions of bending properties to KOSHI are significant, the multiple correlation coefficient being 0.742. . The bending properties of weft direction contribute more to KOSHI than those of warp direction, judging from the figures of contributions. . Thickness and weight contribute to KOSHI, the close positive relationship between weight and KOSHI being particularly clear. . The thickness and weight contribute significantly to TEKASA, while the other mechanical properties hardly contribute at all: TEKASA will be high when the thickness and weight values are high. . The subjective evaluation of KOSHI and TEKASA are especially affected by the thickness and weight of fabrics. Conclusions (1) It is now confirmed that clear differences between polyester Chirimen and silk Chirimen exist. However, bending properties of weft direction and thickness of polyester Chirimen are the similar to those of silk Chirimen. (2) The values of the characteristic mechanical properties of each Chirimen group and their ranges revealed certain distinctive features of Chirimen fabrics: . The values of the bending properties of weft direction of all Chirimen groups are at the same level, and their range corresponds to those for men’s suit fabrics, women’s suit fabrics and dress shirt fabrics. . The values of the tensile properties are at the same level as those for men’s suit fabrics and women’s suit fabrics. (3) The values of the mechanical parameters for suit appearance of all Chrimen groups are within the range of ideal men’s suiting (excluding logSP). (4) In the subjective evaluation of hand evaluation, KOSHI was found to be affected by the bending properties and thickness and weight of fabrics, and TEKASA largely by the thickness and weight of fabrics.

Step 1 bending

2

Contribution to HV

Contribution to HV

Japanese traditional Chirimen fabric logB2 logB1 0 log2HB1 log2HB2

−2 −3

−2

−1

0

1

2

2

logT −2

Step 3 tensile

RT

Contribution to HV

Contribution to HV

(Xi-Mi1)/si1 R = 0.742, (R.M.S = 1.415)

2

logEM1

0 LT −2 −3

logEM2 −2

−1

0

1

2

2

Step 5 Shear

Contribution to HV

Contribution to HV

−2 −3

logG log2HG5

−2

−1

0

1

2

(Xi-Mi1)/si1 R = 0.901, (R.M.S = 0.921)

−1 0 1 2 (Xi-Mi1)/si1 R = 0.837, (R.M.S = 1.158)

3

Step 4 surface

−2

logMMD

−2

−1

logSMD

0

1

2

3

(Xi-Mi1)/si1 R = 0.895, (R.M.S = 0.948)

log2HG 0

−2

MIU

(Xi-Mi1)/si1 R = 0.887, (R.M.S = 0.975)

2

logW

0

−3

3

245

0

−3

3

Step 2 construction

3

2

Step 6 compression logWC

0

LC RC

−2 −3

−2

−1

0

1

2

3

(Xi-Mi1)/si1 R = 0.902, (R.M.S = 0.915)

Notes: Number of samples = 139, Number of subjects = 4; Suffix 1 in the figure of tensile and bending properties: warp direction; Suffix 2 in the figure of tensile and bending properties: weft direction

Figure 7. The contribution of each mechanical property to the HV KOSHI of Chirimen,

logW

−2

logB1

0 log2HB1

−2

2

log2HB2

1 2 3 −1 0 (Xi-Mi1)/si1 R = 0.933, (R.M.S = 0.841)

log2HG5

logG 0 log2HG −2 −3

−2

1 −1 0 (Xi-Mi1)/si1

2

R = 0.939, (R.M.S = 0.803)

3

LT logEM2

logEM1

−2

−2

−1

0

1

2

3

R = 0.925, (R.M.S = 0.896)

−2

Step 5 shear

RT

0

(Xi-Mi1)/si1

Contribution to HV

Contribution to HV

1 2 3 −1 0 (Xi-Mi1)/si1 R = 0.908, (R.M.S = 0.984)

logB2

Step 2 tensile

2

−3

−2

Step 3 bending

2

−3

Contribution to HV

logT

0

−3

Figure 8. The contribution of each mechanical property to the HV TEKASA of Chirimen

Contribution to HV

Step 1 construction

2

Step 4 surface logMMD

2

logSMD

0 MIU

−2 −3

Contribution to HV

246

Contribution to HV

IJCST 22,4

2

−2

1 2 3 −1 0 (Xi-Mi1)/si1 R = 0.938, (R.M.S = 0.817)

Step 6 compression logWC

0

LC RC

−2 −3

−2

−1

0 1 (Xi-Mi1)/si1

2

3

R = 0.941, (R.M.S = 0.793)

Notes: Number of samples = 139, Number of subjects = 4; Suffix 1 in the figure of tensile and bending properties: warp direction; Suffix 2 in the figure of tensile and bending properties: weft direction

References Kawabata, S. (1980), The Standardization and an Analysis of Hand Evaluation Second Edition, The Hand Evaluation and Standardization Committee, The Textile Machinery Society of Japan, Osaka. Kawabata, S. and Niwa, M. (1989), “Fabric performance in clothing and clothing manufacture”, J. Text. Inst., Vol. 80 No. 1, pp. 19-50. Kawabata, S., Niwa, M. and Yamashita, Y. (2002), “Recent developments in the evaluation technology of fiber and textiles: toward the engineered design of textile performance”, Journal of Applied Polymer Science, Vol. 83, pp. 687-702.

Kawabata, S., Niwa, M., Ito, K. and Nitta, M. (1990), “Application of objective measurement to clothing manufacture”, Int. J. Cloth. Sci. Tech., Vol. 2 Nos 3/4, pp. 18-33. Komatsu, K. and Niwa, M. (1981), “Characteristics of physical properties of kimono fabric”, Research Journal of Living Science, Vol. 28 No. 1, pp. 235-43. Nakae, K. (Ed.) (1993), Encyclopedia of Dyeing, Tairyusya, Tokyo. Tomiyama, H. and Ohno, C. (1967), Japanese Traditonal Textile, Tokuma Shoten, Tokyo.

Japanese traditional Chirimen fabric 247

Appendix Mechanical Properties

Symbols Characteristic value

Unit

Measuring conditions Standard (Kawabata, 1980)

Tensile

EM

%

Strip biaxial deformation

LT

Tensile strain at max. load Linearity

–

Upper limit tensile force (max. load): 500 gf/cm

WT RT B 2HB MIU

Tensile energy Resilience Bending rigidity Hysteresis Coefficient of friction

MMD

Mean deviation of MIU Geometrical roughness

gf cm/cm2 % gf cm2/cm Pure bending gf cm/cm Max. curvature, K ¼ ^2.5 cm2 1 – Contactor for friction measurement: ten parallel steel-piano-wires with 0.5-mm dia. and 5-mm length – simulating finger skin geometry. Contact force; 50 gf mm Contactor for geometrical roughness: a steel piano wire, with 0.5-mm dia. and 5-mm length. Contact force; 10 gf gf/cm8 Shear deformation under constant tension of 10 gf/cm gf/cm Max. shear angle, f ¼ ^88

Bending Surface

SMD Shearing

Compression

Thickness and weight

G

Shear stiffness

2HG

Hysteresis at f ¼ 0.58 Hysteresis at f ¼ 58 Linearity Compressional energy Resilience Thickness at 0.5 gf/cm2 Weight per unit area

2HG5 LC WC RC T W

gf/cm – Upper limit pressure: 50 gf/cm2 gf cm/cm2 % mm

Thickness at 0.5 gf/cm2 pressure

mg/cm2

Weight of specimen per unit area

Corresponding author Takako Inoue can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

Table AI. Characteristic values of basic mechanical properties and measuring conditions for KESF measurements

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

IJCST 22,4

Regeneration of 3D body scan data using semi-implicit particle-based method

248

In Hwan Sul

Received 25 July 2009 Accepted 14 December 2009

i-Fashion Technology Center, Konkuk University, Seoul, Republic of Korea and Korean Intellectual Property Office, Daejon, Republic of Korea, and

Tae Jin Kang School of Materials Science and Engineering, Seoul National University, Seoul, Republic of Korea Abstract Purpose – The purpose of this paper is to find automatic post-processing scheme to give textures and motion data to three dimensional (3D) body scan data. Design/methodology/approach – Semi-implicit particle-based method was applied to post-processing of 3D body scan data. The template avatar mesh was draped onto the target scan data and the texture/motion data were transferred to regenerated body. Automatic body feature detection was used to correlate the template body with the target body. Findings – Using semi-implicit particle method, there are advantages in both computational stability and accuracy. The calculation is done in a few minutes and even data with many holes could be used. Originality/value – There are several researches for body feature detection and scan body regeneration but this paper aims for fully automatic method which needs no human intervention. The semi-implicit particle method, which is popularly used for cloth simulation, is applied to body data regeneration. The conventional 3D body scan data, which had no colors and motions can be given textures and motions with this approach. And even the face can be freely interchanged with the use of external face generation software. Keywords Body regions, Image scanners, Motion, Data management, Measurement, Collisions Paper type Research paper

International Journal of Clothing Science and Technology Vol. 22 No. 4, 2010 pp. 248-271 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011048286

1. Introduction Representing the human body shape in three-dimensional (3D) digital format is important for many areas such as reverse engineering in computer-aided design, avatar generation in virtual reality, anthropometric analysis, garment design/simulation and so on. With the development of current 3D scanning technology, various sizes of objects from small pieces of mechanical parts to even a building can be measured within a moderate time. Human body can now be easily scanned, but the data has only coordinate points and cannot be directly used as a 3D avatar because they lack of useful data such as texture data (surface colors), anthropometric feature points and motion data. The main concern of this investigation is to make a useful avatar mesh data from 3D scanning of the individuals. The usefulness of avatar data are defined that the body should have anthropometric feature points, texture coordinates, texture maps, skinning weights, and motion data so that even pose change or dynamic simulation such as cat-walk is possible. To convert a raw 3D can data, which may have noises or holes,

to a useful avatar data, we adopted the idea of iterative closet point to match the mesh points (Allen et al., 2003). That is, if the two mesh data has similar structures, they can be correlated. Using the scheme, a template avatar mesh with texture and motion information was prepared and the avatar mesh vertices were correlated to those of individual’s scan data. To find the same anthropometrical feature points, automatic body measurement algorithm was implemented. The main contribution of this investigation is the use of semi-implicit integration of particle-based simulation to nonrigid mesh registration. The semi-implicit integration method can be used with longer time step compared with those of explicit integration methods and it was successfully used in the simulation of very stiff problems such as cloth. As the body scan data are similar to the clothes in that both are thin shells, the use of semi-implicit integration method showed several advantages over the conventional mesh registration methods. 2. Previous work First, we need an automatic anthropometric measurement algorithm for the body scan data. The specific method of body scanning is beyond our concern because we seek a versatile measurement algorithm which is not sensitive to the raw data noisiness. We tested data from only one of commercial body scanners but the method can be applied to data from other scanners, too. They slightly differ in the scanning method, time and data quality but none of them had the perfect shape of the original body. Some had noisy points and needed post-treatment, while some did not have vertex colors because their inherent scanning methodology captures only the shape, not colors. To reduce the noise and cover the data from all the different scanners, the scan data were converted to slices of point clouds and automatic segmentation was tried. Pargas et al. (1997) first devised a body measurement tool using this slicing method. Nurre (1997) also sliced the raw data and found the bisecting cusp with quick sort algorithms to distinguish the body parts. It was Dekker et al. (1999) who first tried to find marker points automatically. They reduced the noise and filled holes by B-spline interpolation and the cross-sections were analyzed based on rotation angles with respect to the centroid. So they successfully generated a body data with less noise and holes. Wang et al. (2003) also reduced the noise by energy minimization method and recognized the feature points to make a parametric body. But the previous works were interested in the treatment of a single body data and they did not extended their work to matching different bodies so their bodies did not include vertex colors or motion information. Second, matching two different bodies is related to nonrigid registering problem (Allen et al., 2003). We designate the standard body mesh with perfect texture/motion data as a source (or template) mesh and the individual’s 3D scanned body, which may be imperfect and does not include texture or motion data, as a target mesh. Rigid registering algorithm was originally used for merging multiple sets of scan data into one set (Zhang, 1994). It reduces energy from Euclidean distance iteratively until the mesh vertices find the optimal positions. Allen et al. (2003) constructed a parametric space of human bodies from range scans. Their work was template-based nonrigid registration with point-to-point correspondence. This paper uses a very similar approach with Allen’s work but the point-to-point correspondence can lead to local optimum when the incorrect correspondences occur such as when the two bodies have different pose or the scan data has big holes. Such local optima can lead to slow convergence speed. So we

Regeneration of 3D body scan data 249

IJCST 22,4

250

used point-to-element approach instead. The points were not exactly matched to points but they were moved on the surface of the target mesh elements. There were statistical approach to find point correspondence between two bodies without mesh slicing and cusp point searching. Allen et al. (2003) and Anguelov et al. (2005) used Markov network-based feature recognition method so that bodies of different poses can be detected. Their work can be applied to different poses or different bodies. But the inherent nature of Markov network needs an assumption that the target mesh should be subset of the template mesh so arbitrary body scan data cannot be used. But our method is based on automatic feature finding algorithm, so arbitrary meshes data can be used as either source or target mesh. This paper found maker feature points by slicing the point cloud data and grouping them to a certain number of closed loops. The same feature finding procedure was applied both to the template body and target body so that the correspondence can be related. Once specific feature points are found, the rest of mesh vertices which were not on feature points were translated onto the target mesh element surface. The overall vertices were relaxed by minimizing energy from tension, bending and shear forces of the mesh and the linear system for the mesh relaxation was solved iteratively by preconditioned conjugate gradient (pCG) method (Anguelov et al., 2005). And to make points move along the mesh element surface, matrix-based collision detection algorithm was devised. By matching template mesh vertices with the target mesh vertices, the target body could be assigned texture/motion information from those of the template mesh. 3. Automatic body feature recognition 3.1 Data acquisition We used commercially available human body scanner (from Hamamatsu Photonics K.K.) and the exported data contained about 140,000 triangular mesh elements and 281,000 vertices (Figure 1(a)). As the raw data had too high resolution, the raw mesh was reduced using quadric error metrics (Shewchuk, 1994) so that the number of mesh elements became about 10 percent of the original (Figure 1(b)). This reduced mesh data were used for “target mesh” in nonrigid mesh registration process which will be explained in Chapter 5. 3.2 Mesh slicing and iterative closed-loop grouping As our method should find the feature marker points automatically, the first step is to identify each body part correctly. To dismember the whole body mesh into head, bodice, arms and legs, and cross-sections orthogonal to the height direction were acquired as in Pargas et al. (1997). Nurre (1997 did, extrema points were easily found such as hand tips, foot tips, shoulder tips, head tip, and crotch. These tip points were used as starting points for identifying each body part. To find the tip points and dismember the body parts, cross-sections of the reduced mesh were acquired, from foot to head direction. The suitable number of cross-section slices were 100-200. Once each cross-section is found, the point clouds in the cross-section plane were segmented by bisecting planes. Nurre (1997) found this bisecting plane by finding the cusp point, but it was not appropriate around crotch where points can have noise and it was not easy to find cusp point explicitly. Instead nearest neighbor of each point was found and then threshold value was applied. If two points are close under threshold value, they were accepted as neighbors. If no neighbor

Regeneration of 3D body scan data 251

(a)

(b) Notes: (a) Raw data (140,763 vertices, 281,732 triangular elements); (b) after mesh reduction (15,536 vertices, 30,896 elements)

exists around a certain point within threshold, bisecting plane was inserted. Figure 2 shows the effect of threshold value to cross-section segmentation. Now the problem is the determination of suitable threshold value. We found from observation of the cross-sections, that the number of cross-sectional point groups (designated as K) varied from 1 to 4 depending on the body position. So we found heuristic rule for finding

Figure 1. Raw scan data and the simplified mesh

IJCST 22,4

252

(a)

(b)

Figure 2. Segmentation of points cloud with different threshold values (K-resultant number of point groups)

(c) Notes: (a) Successful grouping (K = 3); (b) small threshold (K = 12); (c) large threshold (K = 1)

suitable threshold value. Starting from large threshold value (so that K becomes 1, as shown in Figure 2(c)), the cross-sectional points were grouped and if the output K is different from the expected value, smaller threshold was applied iteratively until desired K value comes out (Figure 2(a)). By repeating this process to all the cross-sections (Figure 3), six body parts (head, bodice, arms, and legs) were acquired (Figure 4). To proceed to the next step, three basic vectors such as arm vector (from left-hand tip to right), up vector (from crotch to head tip), and front vector (same with the scan data’s viewing direction) were defined.

# of curves in layer 1 2

Regeneration of 3D body scan data

3 2~3 3~4

2

3.3 Body feature points detection Body feature points are needed in the garment design or anthropometric study. They are used to specify the locations of anthropometric points such as waist, hips, shoulder, armhole, and so on. The measurements among the feature points can be used to design apparel patterns. Feature points are important not only for garment pattern design, but for our nonrigid mesh registration process because they will be the reference points in matching the source body and target body. We need as many feature points as possible two match different bodies. Each specific feature points of source body, for example waist point, and target body should be at the anthropometrically same position, so that motion and texture data can be transferred exactly. As several body feature points are known by palpating bones through the skin, we used only geometrically detectable ones. But the question is how to detect the same feature point locations among different human bodies. To solve this problem, curvature was used as a criterion for detecting the features. Some features, such as female’s breast, breast bottom, waist, neck, knee, crotch, and armholes, have big curvature changes around them. To find the maximal curvature point, cutting plane was placed (Figure 5(a)) and cross-sectional curve points were found (Figure 5(b)). To remove the effect of point density to the curvature value, cubic B-spline interpolation was done and curvature was calculated from this spline (Figure 5(c)) by converting the spline to third order Bezier curve. And then the feature was assigned to the positive or negative maximal curvature point (Figure 5(d)). Figure 5 is an example for finding breast, bottom breast, neck, and armpit feature points. Feature points found in this step were designated as first feature points. The other feature points (second feature points) were found from the bone information. Table I shows the names of the feature points and bone points. Figure 6 shows the positions of feature points. At times the curvatures do not represent the exact feature points depending on the individual’s body shape. In those cases, relative position to the bone structure was used as an alternative. 3.4 Finding bone positions Figure 7 shows the skeletal structure of the target mesh. These bones are not for motion skinning weights but for getting hints of feature finding. Curvature-based method can find extremal curvature points but the initial position of cutting plane is important. For that purpose, bone positions were found during mesh slicing, supposing that all the

253 Figure 3. Cross-section views of different heights and K values

IJCST 22,4

254

(a)

Figure 4. Results of body part segmentation using slicing

(b) Notes: (a) Cross-sectional point clouds; (b) segmented result

humans have similar bone structure, which means that ratios of intervals between bones are almost constant. For example, knees were supposed to be 50 percent position between hip and foot. Such an assumption does not guarantee an exact knee position, but can be a sufficient condition for roughly positioning cutting plane of Section 3.3. After finding exact features, bones were re-aligned based on the feature point positions.

Regeneration of 3D body scan data 255

(a)

(b)

(c)

(d)

Notes: (a) Cutting planes; (b) cross-section curves and their B-splines; (c) curvature distribution; (d) features found with maximal curvatures

3.5 Finding second feature points Feature which could be found from curvatures were guessed from bone information and they were designated as second feature points. Figure 6 shows the overall feature points. Total 100 feature points were found in this investigation. The more feature points are more helpful for the mesh registration, but our particle-based method does not need such many points. 4. Source mesh data preparation Now that we have target mesh and its anthropometric information, we need source body which has texture and motion information with its feature points. We used graphically modeled body as a source body. We merged the head data from 3D face generation

Figure 5. Curvature-based feature finding

IJCST 22,4

First feature points

Second feature points

Bones

Elbowa, b Wrista, b Thigha, b Kneea, b Calf a, b Anklea, b Abdomenb

Head and Neck Shouldera, bicepa, elbowa, forearma, and handa Hipa, thigha, kneea, calfa, anklea, and foota Chest, breast, waist, abdomen, and pelvis

256

Headtip Handtipa Foottipa Crotch Armpita Breasta Bottom breasta Neckb Hipa Waistb

Table I. List of feature points and bones

Notes: aThe same name exists in two body parts (left arm/right arm, or left leg/right leg); beach has four sub-points (in forward, backward, and left and right direction)

software (Facegenw) to the source body so that we can change the face using the scanned person’s face pictures. Motion data were from database of Carnegie Mellon University Graphic Lab. and the skinning weights were prepared in advance via 3D modeling tool. The source mesh had 6,960 vertices and 13,672 triangular elements (Figure 8). The bones and features are found in the same way with the target mesh. The next step is to match the source body to the target body via mesh registration. 5. Nonrigid mesh registration Originally, rigid mesh registration is a process to merge multiple sets of meshes into one. In this application, the source mesh is merged with the target mesh, but the source mesh is deformed. Therefore, we need the nonrigid mesh registration between source and target mesh. Moreover, the source mesh after registration can have irregular strain distribution and even distorted triangular elements. To recover those poorly registrated mesh elements, we need another process to minimize the internal energy of the source mesh, which is the relaxation process. 5.1 Initial vertex mapping using cylindrical coordinate We want to find the new positions of source body vertex pi on the surface of target body mesh while maintaining the relative position with the nearby feature points. The source and target body have different number of vertices, so the pi’s cannot be matched to vertices of target mesh (qi) one-to-one. The better way is to match with respect to the feature points because they have the same number between two bodies. So our registration algorithm is three-step process. The first step is to move only points nearby feature points. The second step is to move the rest of the points roughly using cylindrical coordinates of the body parts. As the six human body parts of Section 3.2 has cylindrical shape, respectively, the points pi’s can be roughly laid on the target mesh surface (Figure 9(a)). The third step is to relax the deformed mesh using particle-based method. Allen et al. (2003) defined data error, smoothness error and marker error and minimized the total error sum. But the method finds point-to-point correspondence and need many iteration time and assumption that the pose is similar. This paper adopted cloth simulation technique with semi-implicit integration (Garland and Heckbert, 1997) for fast calculation speed and robustness, which can cover data even with holes.

Regeneration of 3D body scan data 257

(a)

(b) Notes: (a) feature points; (b) feature points with measurements

Figure 6. Feature points

IJCST 22,4

258

Figure 7. Bone positions found

Note: These are used for cylindrical mapping, not for motion

The semi-implicit integration method has an advantage that big time step about 10-30 ms can be used in dynamic simulations. 5.2 Mesh relaxation using semi-implicit particle-based method The regenerated body of Figure 9(a) is roughly similar to target body, but it includes misaligned vertices so the surface texture looks warped. This is due to error of cylindrical coordinate mapping. So the mesh elements were relaxed by the particle-based method (Garland and Heckbert, 1997). Allen et al. (2003) used only edge-based error term, but we used tension, bending and shear terms together so that edges and element angles are relaxed in fast time. Among many integration algorithms, implicit integration has an advantage that it can be used with unlimited value of time step. In the case of the cloth simulation, the

Regeneration of 3D body scan data 259

(a)

(b) Notes: (a) Source body with texture coordinates and skinning weights; (b) marker feature points and skeletal structure

Figure 8. Source body and its feature points

IJCST 22,4

260

(a)

(b)

Figure 9. Mesh matching process

Notes: (a) Initial mesh matching using cylindrical coordinates (left: source body, right: target body, middle: regenerated body); (b) mesh relaxation process of the regenerated body

governing equation cannot be fully described in the implicit form but only semi-implicit integration form after Taylor series expansion is possible. So unlimitedly large value of time step cannot be used, but relatively large time step of 10-33 ms can be used. Thus, 1.0 s simulation can be calculated in only 30 loops. Therefore, semi-implicit integration is the most famous method in cloth simulation and our idea is to adopt this to the nonrigid mesh registration and relaxation problem. The governing equation for semi-implicit particle method (Garland and Heckbert, 1997) is as follows:     ›f ›f 2 ›f M2h 2h Dv ¼ h f 0 þ h v 0 ›v ›x ›x

where M is the mass matrix, f, force vector, v, velocity vector, h, time step used. Thus, forces and their derivatives of each mesh vertex should be collected to the linear system. Tension, bending and shear forces among mesh vertices were used and the followings are the definition. Tension:   ðx 2 x Þ j i f tij ¼ kt jxj 2 xi j 2 l 0ij jxj 2 xi j ij, l 0ij , rest

where kt, elastic constant, xi, xj, vertices of edge length of edge (Oh et al. 2006). Tension is the major term which relaxes the mesh. This force limits the edge ij maintains the length at l 0ij . Tension springs are used for among mesh edges and between feature points to match. l 0ij of feature points edge were set to zero and l 0ij of mesh elements were set to half of anticipated length because high strain is more helpful for linear system solving than low strain. Bending force:   ›E b kb e e b b fi ¼ ðu 2 u0 Þ2 ; E ¼ þ ›xi 2 la lb where kb, bending rigidity, u, u0, current and rest angle of bending edge i, e, length of bending edge and la, lb, perpendicular distance of vertices to bending edge (Baraff and Witkin, 1998). Bending force maintains the angle of bending edge. It works best when the target body has holes. Shear force: f se ¼ ks ðg 2 g0 Þ where ks, shear rigidity, g, g0, current and rest angle of triangle e. Shear force maintains the triangular shape. This is to preserve the texture sharpness because the texture is blurred when the triangle deforms too much. These force vectors and their derivatives (Jacobians) were calculated for each vertices and they were collected to one large sparse matrix. This linear system was solved iteratively using pCG method. 10-20 iterations were enough to solve the linear system in each time step. The matrix is symmetric and moreover sparse, so it is advantageous to use iterative method rather than exact inverse-matrix-based method. By simplifying the Jacobian terms (Baraff and Witkin, 1998) and optimizing pCG method (Anguelov et al., 2005), it was possible to simulate 1,000 vertices of mesh in 30 frames per second. The values of material properties were set similar to those of cotton cloth. The time step used was 33 ms and 30 loops of iteration was enough to relax the whole mesh. 5.3 Collision detection between triangles using collision matrix Another important thing in our mesh relaxation is to make vertices of regenerated avatar stay on the surface of the target mesh. That is, we want the vertices of template mesh correlated to the elements of the target mesh (point-to-element correspondence), not to the target mesh vertices. The point-to-point correspondence can lead to local optima when the reference points are misplaced or when the reference points even do not exist (possibly due to scanning noise or holes). Our method maintains the whole mesh shape by tension and bending force and it does not fail even if the target mesh has holes or voids on the surface. To do so, the information among colliding triangle pairs are needed.

Regeneration of 3D body scan data 261

IJCST 22,4

262

The above force terms do not include information about the target mesh’s shape, so we need to constrain the vertex movement using collision detection between regenerated mesh vertices and target mesh triangles. As we have minimized the target mesh and the triangle elements have relatively big size, the vertices can penetrate into the target body depending on the convexity of the target mesh surface. This is analogous to the cloth-to-body collision detection problem in cloth simulation, where the regenerated avatar becomes the cloth and the target mesh becomes body. There are many ways of efficient collision detection but finding the collision time using cubic polynomial solving (Kang and Cho, 2002) is the famous technique for finding collisions between moving triangles. But it has weakness when the two triangles do not have any motion, i.e. the cubic polynomial does not have solutions. In this application, the body (target mesh) does not have motion, so the velocity-based detection can fail in detection. Instead, we used transformation matrix-based translation method. The idea is to move the reference triangles (possibly the triangles of the target mesh) on to the global xy-plane. To implement the transformation easily, local coordinate system was prepared by setting the first vertex of the reference mesh triangle as local origin and its normal vector as local z-axis. Then whole the local system is rotated and translated so that the local origin goes to global origin (0, 0, 0) and the local z-axis becomes (0, 0, 1). And then to make the problem more simple, the target mesh element is distorted so that the other two vertices, which are not on origin, goes to (1, 0, 0) and (0, 1, 0). The whole transformation can be represented by matrix Pi which are assembly of several submatrices. Now Pi is multiplied to the vertex or triangle with which collision detection is needed. Therefore, the original 3D triangle-to-triangle collision detection problem becomes a simple two-dimensional point-in-triangle problem. The operations were actually done by series of matrices and these matrices per each element did not need to be modified once they were prepared in the first time. The transformation matrix Pi of reference triangle i is: P i ¼ M Scale2D · M Shear2D · M Rot2D · M Rot3D · M Trans3D and the Appendix shows the illustration of each matrix operations. Vertices, which were inside the target triangle and were distant within certain range were judged as collided vertices and they were given constraints so that they move only on the surface of the reference triangle. There can be three types of triangle-to-triangle collisions as shown in the Appendix, but using only point-to-triangle collision was enough in this case. Figure 9(b) shows the mesh relaxation process where the vertices stay on the target body surface. Figure 10 shows the final result of mesh relaxation. Although the source (Figure 10(a), left) and target (right) body has slightly different poses, the regenerated body was successfully fitted to the target body, which can be shown in Figure 10(b). Head, hands, and feet mesh vertices were not relaxed to preserve the original shapes. This was inevitable because hands are usually difficult to scan, so we chose to use the source body’s hand mesh. The size and orientation was modified with respect to the thickness of the wrist cross-sections. When the relaxation was finished, motion skinning weight was transferred to regenerated body (Figure 11). Moreover, as the regenerated body completely overlaps with the target body, the motion data also could be transferred to the target body by searching nearest points. Therefore, the source body, regenerated body and target body

Regeneration of 3D body scan data 263

(a)

(b) Notes: (a) Source (left), regenerated (middle), and target (right) bodies; (b) left: target body overlapped with regenerated body, right: cross-sections from shoulder, breast, waist, and hip (solid line: target body, dotted line: regenerated body)

all have the same motion data. Figure 12 shows the motion data applied to three bodies, respectively, with same clothes on. 6. Results and discussion The advantage of particle-based simulation in nonrigid mesh registration is that the simulation speed is relatively fast and the simulation is not stuck to the local optima. The whole procedure took less than 2 min (excluding mesh simplification time of raw scan data) in AMD 3.0 GHz dual-core PC. The most important feature of this investigation is that it needs no intervention of human operator. Moreover, it does not need landmarks to

Figure 10. Result of nonrigid mesh registration

IJCST 22,4

264

(a)

(b)

Figure 11. Motion skinning weight transferred

Notes: (a) Frontal view; (b) side view

Regeneration of 3D body scan data 265

(a) Notes: (a) Source body; (b) regenerated avatar; (c) target body

(b)

(c)

be marked on the scanned person’s body. Bodies of slightly different poses had the same number of landmark feature points and they were used for matching source and target mesh data. These are important benefits for improving throughput of 3D body scanning procedure. The time bottleneck was the mesh simplification process, not the mesh relaxation process. Not only textures but also skinning weight were transferred to the regenerated body and the face, hair, and accessories were replaceable. Figure 13 shows the result for various scan data. The same target body was used for bodies of different gender, weights or heights. Note that the faces have been changed in Figure 13(a)-(c) because we used fixed number of face faces. Via face generation software, the avatar can have 3D face only from the scanned persons’ front picture. The advantage of using particle-based method for relaxation is that it does not need many feature points. Figure 14 shows results when 96, 36 and 18 feature points used. As the small number of feature points used, the texture coordinate deviates from the original position but the visual quality is not bad because the shear force term preserves the overall shape. The feature points are not always on the perfect positions, whether they are found automatically or manually. Ill positioning of feature points can lead to local optima in simulation. Using our method, the ill positioned feature points can be simply eliminated. The bending force term did not make any difference in the above results, but it works when the target body has holes (Figure 15). We put holes manually to the target mesh of Figure 15 to show the effect of the bending force term. Only if the exact feature points and bone structure of the target body are known, the relaxation can proceed by increasing the bending rigidity. High value of bending rigidity makes the material property stiff, so the regenerated body does not penetrate into the target mesh.

Figure 12. Cloth simulation result

IJCST 22,4

266

(a)

(b)

Figure 13. Results for various scan data

(c) Notes: The faces were changed; (a) female no. 2; (b) female no. 3; (c) male no. 1

Additionally, as we measured the feature points and bones, the 3D space around human body can be described with respect to the feature points or bones. Any point coordinate can be described in relative cylindrical coordinates with respect to bodice, arms or legs. For example, Figure 16 shows that initial position of garment fitted to source body can be

Regeneration of 3D body scan data 267

(a)

(b)

(c)

Figure 14. Effect of number of feature points in mesh relaxation

Notes: (a) 96 points; (b) 36 points; (c) 18 points

Note: Assume the feature points were successfully found

translated to the same anthropometric point of regenerated body. This can be useful in digital garment design and manufacturing. But there exists some limitations. As our collision detection assumes the target mesh has mesh elements, scan data with only point cloud cannot be used. In that case, the triangular mesh should be made in advance or the collision detection scheme should be modified. Also the motion pose of regenerated body was slightly different especially around the chest because the two bodies had different length of bones. To overcome this,

Figure 15. Regeneration of scan data with many holes

IJCST 22,4

268 Figure 16. Copying one body’s garment to another body with different pose and (a) (b) (c) (d) size using cylindrical body Notes: (a) Garment patterns fitted to the source body; (b) importing garment to different body; coordinates (c) repositioning using cylindrical transformation; (d) final fitting result

various shapes of source body should be prepared or motion retargeting algorithm should be used. The welding of head mesh to the neck was also an issue because the thickness of the scanned pereson’s neck and the neck from the Facegenw software can be different. This needs further study.

7. Conclusions The raw 3D scan data which lacks of details and vertex colors were converted to useful avatar. The lost meshes of hands and head were replaced by those of the avatar and they could be changeable. Using any external face generation software, the individual’s 3D head can be implanted to the scan data without welding line. As the template avatar contained texture coordinates and skinning weights, the regenerated body was also ready for dynamic simulation and rendering. The feature points for body measurement were found automatically by slicing the raw mesh data and grouping the cross-sections into closed loops. Feature points were used not only for the body mesh matching, but also for the parameterization of Euclidean space to body-relative coordinates. Even if the scan body has various shapes and sizes, the feature points always directed to the same anthropometric point and could be used for positioning initial garment patterns in 3D try-on simulation. The main contribution of the investigation is the use of large time step in mesh registration process by using semi-implicit particle-based method. The avatar can be generated from the raw scan data within a few minutes and the avatar can have not only texture data but also motion data. Even if the raw data has holes or the number of feature points is deficient, the avatar can be generated without harming the visual quality. Therefore, our method has advantages in both speed and stability.

References Allen, B., Curless, B. and Popovic, Z. (2003), “The space of human body shapes: reconstruction and parametrisation from range scans”, ACM Transactions on Graphics ( ACM SIGGRAPH’2003 ), Vol. 22 No. 3, pp. 587-94.

Anguelov, D., Srinivasan, P., Koller, D., Thrun, S., Pang, H. and Davis, J. (2005a), “The correlated correspondence algorithm for unsupervised registration of nonrigid surfaces”, Advances in Neural Information Processing Systems, Vol. 17, pp. 33-40. Anguelov, D., Srinivasan, P., Koller, D., Thrun, S., Rodgers, J. and Davis, J. (2005b), “SCAPE: shape completion and animation of people”, ACM Transactions on Graphics ( ACM SIGGRAPH’2005 ), Vol. 24 No. 3, pp. 408-16. Baraff, D. and Witkin, A. (1998), “Large steps in cloth simulation”, Proceedings of the Computer Graphics, Orlando, FL, pp. 43-54. Dekker, L., Douros, I., Buxton, B. and Treleaven, P. (1999), “Building symbolic information for 3D human body modeling from range data”, Proceedings of the Second International Conference on 3-D Digital Imaging and Modeling ( 3DIM’99 ), Ottawa, pp. 388-97. Garland, M. and Heckbert, P. (1997), “Surface Simplification Using Quadric Error Metrics”, SIGGRAPH 97, Carnegie Mellon University, Pittsburgh, PA. Kang, Y.M. and Cho, H.G. (2002), “Complex deformable objects in virtual reality”, VRST’02, pp. 49-56. Nurre, J. (1997), “Locating landmarks on human body scan data”, Proceedings of the International Conference on Recent Advances in 3-D Digital Imaging and Modeling ( 3DIM’97 ), Ottawa, pp. 285-95. Oh, S., Ahn, J. and Wohn, K. (2006), “Low damped cloth simulation”, The Visual Computer: International Journal of Computer Graphics, Vol. 22, pp. 70-9. Pargas, R., Staples, N.J. and Davis, J.S. (1997), “Automatic measurement extraction for apparel from a three-dimensional body scan”, Optics and Lasers in Engineering, Vol. 28 No. 2, pp. 157-72. Shewchuk, J. (1994), “An introduction to the conjugate gradient method without the agonizing pain”, Technical Report CMUCS-TR-94-125, Carnegie Mellon University, Pittsburgh, PA. Wang, C.C.L., Chang, T.K.K. and Yuen, M.M.F. (2003), “From laser-scanned data to feature human model: a system based on fuzzy logic concept”, Computer Aided Design, Vol. 35, pp. 241-53. Zhang, Z. (1994), “Iterative point matching for registration of free-form curves”, International Journal of Computer Vision, Vol. 13 No. 2, pp. 119-52. Further reading Provot, X. (1997), “Collision and self-collision handling in cloth model dedicated to design garments”, Proceedings of the Graphics Interface, Seattle, WA, pp. 177-89. (The Appendix follows overleaf.)

Regeneration of 3D body scan data 269

IJCST 22,4

Appendix Z

pi0 n(nx,ny,nz) =(r=1,q,f)

270

pi1

Z

MTrans3D

p'i0

Y X

pi2

Y

X

X

p'i1

(a)

p'i2

Z Z p'i0

Y

n'(0,0,1) =(r'=1,0, p2 )

MRot3D

p''i0

n(nx,ny,nz) =(r=1,q,f) X

p'i1

p'i2

Y

p''i1

X (b)

Y

Y qi2

p''i2 MRot2D p''i0

X

ai

qi0

X

qi1

p''il (c) Y

Y

gi

qi0

qi2

q'i2 MShear2D qi1

X (d)

q'i0

Y

q'i1

X

Y

1 q'i2

1 MScale2D X

Figure A1. Submatrices of collision detection matrix Pi

q'i0

1 q'i1

0

1

X

(e) Notes: (a) Matrix MTrans3D; (b) Matrix MRot3D; (c) Matrix MRot2D; (d) Matrix MRot2D; (e) Matrix MScale2D

n (nx,ny,nz) =(r =1,q,f)

Regeneration of 3D body scan data

p'(p'x,p'y,p'z)

Z

p ( px , py , pz)

n'(0,0,1) =(r =1,0,p2 )

Rotation & translation

Y

271

Z

p'(p'x, p'y,0) Y X

X (a) p ( px , py , pz)

p' ( p'x , p'y , p'z) n'(0,0,1)

Rotation & translation

n

Y

Z Y

q' ( q'x , q'y , q'z )

X

q ( qx , qy , qz )

X (b)

r'

p (px,py,pz)

p' r (rx,ry,rz)

n

n'(0,0,1) Rotation Rotation& & translation translation

Y

Z Y X

X

q ( qx , qy , qz )

q

(c) Notes: (a) Point infiltration; (b) edge penetration; (c) face penetration

Corresponding author In Hwan Sul can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

Figure A2. Three cases of triangular collisions

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

IJCST 22,4

272 Received 6 April 2009 Revised 12 September 2009 Accepted 12 September 2009

Warp tension distribution over the warp width and its effect on crimp distribution in woven fabrics Gulcan Ozkan and Recep Eren Department of Textile Engineering, Faculty of Engineering and Architecture, University of Uludag, Bursa, Turkey Abstract Purpose – The purpose of this paper is to investigate warp and weft crimp distribution over the fabric width and how it is influenced by warp tension distribution over the warp width. Design/methodology/approach – An experimental design in this research includes air jet loom, tension sensor, inductive sensor and personal computer. Findings – It is found that warp crimp in the fabric on the loom is higher in the edge zones than the middle of the fabric and warp crimp in the middle is higher than warp crimp in edge zones of the grey fabric. Weft crimp in the edge zones is higher than in the middle of the grey fabric. The reason behind warp tension and warp and weft crimp variations over fabric width is that weft yarn slips towards inside fabric at selvedges and gets relaxed during beat up. Originality/value – It is proved that reducing weft yarn slip and therefore weft yarn relaxation during beat up will reduce warp tension and warp and weft crimp variations and improve the uniformity of fabric properties over the fabric width. Keywords Fabric production processes, Surface properties of materials, Yarn testing Paper type Research paper

International Journal of Clothing Science and Technology Vol. 22 No. 4, 2010 pp. 272-284 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011048295

1. Introduction A weaving warp consists of a multitude of warp ends with a uniform tension. All the warp ends over the fabric width should have the same tension to obtain homogeneous fabric properties over the width. Uneven warp tension over the fabric width may cause unevenness in fabric properties as well as affecting weaving machine performance adversely. Previous researches show that warp tension over the fabric width is not uniform and it is higher in the middle of the warp width than in edge zones (Weinsdorfer et al., 1988, 1991; Ludwig and Gries, 2003). This warp tension distribution may, under certain conditions, increase warp end breaks especially in edge zones of the warp width. The reason for the warp breaks is that warp ends in edge zones which get slack and stick are exposed to higher loading (Ludwig and Gries, 2003). Apart from warp breaks, weft threads which strike warp ends in edge zones increases loom stops in air jet weaving macines (Ludwig and Gries, 2003). The unequal distribution of the warp end tension over the warp width also affects fabric quality. The greater the unevenness in warp tension distribution over the warp width, the greater is the unevenness in the fabric characteristics over the warp width. The unequal distribution of the warp end tension over the warp width causes varying warp and weft crimps over the warp width in both the fabric on the loom and grey fabric (Ozkan, 2005). Also, warp tension distribution over

the warp width, warp crimp distribution in the fabric on the loom and warp and weft crimp distributions in grey fabric over the fabric width vary with different warp and weft densities, warp and weft yarn types and numbers, weave and total warp tension. Previous research investigates warp tension distribution over the fabric width and how it is influenced by weaving machine settings (Weinsdorfer et al., 1988, 1991; Ludwig and Gries, 2003). Warp tension, shed closing angle and type of temple and shed asymmetry are shown to affect warp tension distribution over the warp width. This paper investigates experimentally warp tension distribution over the warp width and its effect on warp crimp distribution in the fabric on the loom over the warp width and warp and weft crimp distribution in grey fabric over the fabric width. Twisted polyester warp yarn and three different cotton-carded ring spun weft yarns are used in weaving fabrics. Fabrics were woven at three different weft densities with two different warp tensions. In this way, the effect of weft density, weft yarn number and warp tension on warp tension distribution over the warp width, warp crimp distribution in the fabric on the loom over the warp width and warp and weft crimp distribution in grey fabric over the fabric width are investigated.

Warp tension distribution

273

2. Experimental A Picanol OMNI air jet loom was used in the production of woven fabrics. Air jet loom and fabric parameters are given in Table I. Warp tension was measured by SCHMIDTH make single end tension sensor which has a measurement interval of 0-200 cN. Loom main shaft angle was measured by an inductive sensor. Tension sensor and inductive sensor were interfaced to a personal computer and warp tension was read with respect to loom main shaft angle by a computer program developed for this purpose using Turbo C programming language. About 200 readings were recorded in each loom revolution. Insertion of tension measuring head onto a warp thread causes a temporary increase in the tension in that thread. Therefore, warp tension readings were recorded after a sufficient number of picks were woven during which the tension in that thread returned to its stable value. Different fabric constructions were woven by changing weft yarn number and weft density and warp tension. Type of warp yarn and weft yarn, warp yarn number and warp density remained the same. Table II shows values of parameters changed during weaving of different fabric constructions. Adjusted total warp tension is a parameter which represents warp tension of all ends measured by loom tension sensor. This parameter is entered from machine computer and used to adjust warp tension by the tension control system of the air jet loom. For each fabric construction, warp tension

Loom parameters

Fabric parameters

Weft insertion system: air jet Shed formation system: cam shedding motion Machine speed: 710 rpm Machine width: 190 cm Number of weft colors: 2 Let off system: electronic Take up system: electronic Shed closing angle: 3208

Weave: plain weave (four shafts) Warp yarn: 150 denier twisted PES yarn Warp density: 33.5 ends/cm Warp width in the reed: 168 cm Warp tension: given in Table II Weft yarn: given in Table II Weft density: given in Table II

Table I. Loom and fabric parameters

IJCST 22,4

274

measurement was carried out over 20 loom revolutions for 13 warp ends over the warp width and average warp tensions were calculated for all warp ends. Then, warp crimp in the fabric on the loom (out off temple zone) was measured for each fabric construction. Warp end tension and warp crimp measurement points on the loom are given in Table III. Warp crimp in the fabric on the loom was measured as follows: the loom was stopped after a length of 50 cm of cloth was woven. Two marks were put on a group of warp yarns between 5 cm interval between the reed and shafts when the heads were level. The loom was then started to run. The loom was stopped again after both marks passed temple area. The distance between the two marks on the fabric on the loom was measured by a ruler. Warp crimp in the fabric on the loom was calculated by using equation (1): ct1 ¼ where:

a length of 5 cm of a group of warp yarns between reed and shafts.

lc

a distance between the two marking group of yarns in the fabric on the loom.

Ct1

warp crimp in the fabric on the loom. Adjusted total warp tension (kN)

Weft density (picks/cm)

1.00 1.75 1.00 1.75 1.00 1.75

14,18,22 14,18,22 18,22,26 18,22,26 18,22,26 18,22,26

29.5 tex (Ne 20/1) (carded cotton) 16.4 tex (Ne 36/1) (carded cotton) 11.8 tex (Ne 50/1) (carded cotton)

Measuring points of warp tension and warp crimp in loomstate fabric Measuring point (cm) Measuring no. (as from left temple)

Table III. Warp tension and crimp measuring points

1 2 3 4 5 6 7 8 9 10 11 12 13

ð1Þ

ls

Weft yarn

Table II. Experimental work program

ls 2 lc lc

6 19 31 43 54 64 72 84 96 111 128 142 160

Measuring points of warp and weft crimp in grey fabric Measuring point (cm) (as from left fabric selvedge) Measuring no. Warp crimp Weft crimp 1 2 3 4 5 6 7 8

6 20 41 63 94 120 139 158

30 65 95 125 155 – – –

Warp and weft crimps in grey fabrics were measured according to ASTM D3883 – 04 standard after fabrics were conditioned under laboratory conditions (ASTM Standards Specifications, 2004). Measuring points over the grey fabric width are given in Table III. 3. Results Warp tension distribution over the warp width Figures 1-3 show warp tension distribution over the warp width for 29.5 tex (Ne 20/1), 16.4 tex (Ne 36/1) and 11.8 tex (Ne 50/1) cotton weft yarns, respectively, with two different tension levels. As seen from all of three figures, warp tension does not have a uniform distribution over the loom width despite the preparation of warp yarn with uniform warp tension over the warp width. Warp tension is lower in the edge zones and 14 picks/cm (1.75 kN)

18 picks/cm (1.75 kN)

22 picks/cm (1.75 kN)

14 picks/cm (1 kN)

18 picks/cm (1 kN)

22 picks/cm (1 kN)

Warp tension distribution

275

35.0

Warp tension (cN)

30.0 25.0 20.0 15.0 10.0 5.0 6

19

31

43

54

64

72

84

96

111

128

142

160

Measuring point (as from left temple-cm)

18 picks/cm (1.75 kN)

22 picks/cm (1.75 kN)

26 picks/cm (1.75 kN)

18 picks/cm (1 kN)

22 picks/cm (1 kN)

26 picks/cm (1 kN)

Figure 1. Warp tension distribution over the warp width for 29.5 tex ( Ne 20/1) weft yarn

40.0

Warp tension (cN)

35.0 30.0 25.0 20.0 15.0 10.0 5.0 6

19

31

43

54

64

72

84

96

111

Measuring point (as from left temple-cm)

128

142

160

Figure 2. Warp tension distribution over the warp width for 16.4 tex ( Ne 36/1) weft yarn

IJCST 22,4

18 picks/cm (1.75 kN) 18 picks/cm (1 kN)

22 picks/cm (1.75 kN) 22 picks/cm (1 kN)

26 picks/cm (1.75 kN) 26 picks/cm (1 kN)

40.0

276

Figure 3. Warp tension distribution over the warp width for 11.8 tex ( Ne 50/1) weft yarn

Warp tension (cN)

35.0 30.0 25.0 20.0 15.0 10.0 5.0 6

19

31

43

54

64

72

84

96

111

128

142

160

Measuring point (as from left temple-cm)

increases towards the middle of the loom. It takes the highest values around the middle of the loom width. It is clearly seen from the figures that an increase in weft density decreases the warp tension variation over the loom width. Warp tension variation over the loom width increases with an increase in total average warp tension level. An increase in weft yarn thickness decreases warp tension variation over the loom width which means that warp tension varies less over the loom width as weft yarn becomes thicker. The effect of warp tension, weft density and weft yarn number on warp tension distribution is given in Table IV. x represents the ratio of the difference between the maximum and minimum warp tension across the warp to minimum warp tension in percentage over the fabric width. As seen from Table IV, the maximum warp tension variation over the fabric width occurs in the fabric woven with 18 picks/cm weft density,

Weft yarn number 29.5 tex (Ne 20/1) (carded) 16.4 tex (Ne 36/1) (carded) Table IV. The change of warp tension and warp crimp in the fabric on the loom over the warp width

11.8 tex (Ne 50/1) (carded)

Change of warp tension x- (%) Adjusted total Adjusted total Weft warp tension warp tension density (1.75 kN ) (1 kN ) (picks/cm)

Change of warp crimp in the fabric on the loom y-(%) Adjusted total Adjusted total warp tension warp tension (1.75 kN ) (1 kN )

14 18 22

65.5 57.6 29.8

150.2 79.9 46.9

34.6 35.1 25.8

100.0 55.5 39.2

18 22 26

102.7 72.3 44.4

199.2 109.2 78.4

84.2 68.1 44.7

100.0 51.5 46.8

18 22 26

148.4 132.7 79.0

217.4 184.5 117.5

79.8 55.3 47.8

100.0 87.9 71.1

11.8 tex (Ne50/1) weft yarn and 1.75 kN total adjusted warp tension (x ¼ 217.4 per cent) and the minimum warp tension variation occurs in the fabric woven with 22 picks/cm weft density, 29.5 tex (Ne20/1) weft yarn and 1.0 kN total adjusted warp tension (x ¼ 29.8 per cent). Between these two extreme cases, warp tension variation over the fabric width decreases with increasing weft density and weft yarn thickness and with decreasing warp tension as represented by x-values. This result shows that a denser fabric woven with a lower warp tension has less warp tension variation over the fabric width. The mechanism of warp tension distribution over the fabric width and the effect of weft density, weft yarn number and warp tension on this distribution can be explained as follows. When a weft yarn is inserted and pushed towards the cloth fell by the reed, it takes on crimp due to interlacing with the warp ends. As fabric is hold from both sides by the temples, the straight weft laid to the shed is beaten up to the fabric at the same length but in a crimped form. Therefore, weft yarn elongates during beat up due to crimp formation and weft yarn tension inside the fabric increases. As the weft yarn tension inside the fabric exceeds the tension of the weft outside the ground warp width, the weft from outside ground warp is drawn inwards. This causes some slip and relaxation of weft yarn between warp ends at edge zones. Towards the middle of the fabric, the slip of the weft yarn between warp yarns decreases progressively due to the increase in interlacing point. In the middle, the weft is hold firmly by the warp ends and no slip occurs. As a result, weft crimp at edge zones of the fabric becomes higher than middle zone of the fabric and in contrast to this, warp crimp at edge zones is expected to be lower than the middle zone. Same amount of warp is fed from the beam for all ends but less is taken up by the fabric at edge zones than middle part of the fabric because of lower warp crimp. This causes lower warp yarn tension at edge zones and higher warp yarn tension at the middle zone of the fabric. The slip of the weft yarn inside the fabric at edge zones was observed on the loom by a simple experiment. A new weft was inserted and the loom was brought to the shed-closing angle. At this position of the loom, a mark was put on the newly inserted pick between the ground selvedge and false selvedge warp yarns. A mark was also put on the reed in line with the mark on the weft. Another mark was put on the newly inserted weft next to the last false selvedge end. After the newly inserted weft was beaten up and fixed to the fabric fell, it was observed that the mark on the weft between the ground and false selvedge ends shifted the left with respect to the mark on the loom. Also, the mark on the weft next to the last false selvedge end shifted towards inside the false selvedge. This was repeated for many weft yarns and the same shift (slip) in the weft inside the fabric was observed. The same experiment was carried out also on a rapier loom and the same shift (slip) in the weft inside the fabric at the fabric edges was observed. Based on the above discussion, it is thought that main reason for warp tension variation over the fabric width is the slip of the weft yarn inwards at edge zones. The stronger the weft yarn is hold by false selvedges outside the ground warp and the more holding force (i.e. friction force) is created between weft yarn and ground warp yarns, the less warp tension variation is expected to occur over the fabric width. Machine settings, fabric construction and false selvedge construction and timing affect the holding force between warp yarn and weft yarn at edge zones and therefore warp tension variation over the fabric width. The effect of machine settings is investigated in the previous

Warp tension distribution

277

IJCST 22,4

278

research works (Weinsdorfer et al., 1988, 1991; Ludwig and Gries, 2003). The effect of weft density and weft yarn number and warp tension on warp tension variation over fabric width is investigated in this paper. As shown from Figures 1-3, increasing general level of warp tension increases warp tension variation over the fabric width. This is because higher warp tension forces weft yarn to take on more crimp during beat up and weft yarn inside fabric elongates and induces more tension. This causes more slip of weft yarn from outside the ground warp towards inside the fabric. As a result of this, weft yarn relaxes more at the edge zones compared to weaving at a lower warp tension. This effect causes higher warp tension variation over the fabric width. Both increasing weft density and increasing weft yarn thickness decrease warp tension variation over the fabric width because both effects increase warp crimp. An increase in warp crimp increases wrapping angle of warp yarn around the weft yarn. This increases friction forces between warp and weft yarns and reduces the slip of the weft yarn towards inside the fabric and the weft relaxes less. Because of this, less warp tension variation occurs over the warp width. Warp crimp mentioned in the above discussion corresponds to warp crimp in the fabric before temples. After passing the temples, the fabric gets contracted widthwise and warp and weft crimp distribution change. Following parts of the paper deal with warp crimp distribution in the fabric on the loom after temples and warp and weft crimps in the fabrics after it is taken off the loom. Warp crimp distribution in the fabric on the loom over the warp width Figures 4-6 show warp crimp distribution in the fabric on the loom over the fabric width for 29.5 tex (Ne 20/1), 16.4 tex (Ne 36/1) and 11.8 tex (Ne 50/1) cotton weft yarns, respectively. In all of three figures, warp crimp in the fabric on the loom in edge zones is higher than warp crimp in the middle of the warp. It was mentioned above that warp crimp in the fabric before the temples was lower at edge zones of the fabric than in the middle and this was the cause of non-uniform warp tension distribution over the fabric width. After the fabric passes the temples, the fabric contracts widthwise and this changes warp crimp distribution as shown in Figures 4-6. This warp crimp distribution

Figure 4. Warp crimp distribution in the fabric on the loom over the warp width for 29.5 tex (Ne 20/1) weft yarn

Warp crimp in loomstate fabric (%)

14 picks/cm (1.75 kN) 14 picks/cm (1 kN)

18 picks/cm (1.75 kN) 18 picks/cm (1 kN)

22 picks/cm (1.75 kN) 22 picks/cm (1 kN)

12.0 10.0 8.0 6.0 4.0 2.0 0.0 6

19

31

43

54

64

72

84

96

111

Measuring point (as from left temple-cm)

128

142

160

Warp crimp in loomstate fabric (%)

18 picks/cm (1.75 kN)

22 picks/cm (1.75 kN)

26 picks/cm (1.75 kN)

18 picks/cm (1 kN)

22 picks/cm (1 kN)

26 picks/cm (1 kN)

12 10

279

8 6 4 2 0 6

19

31

43

54

64

72

84

96

111

128

142

160

Measuring point (as from left temple-cm)

Warp crimp in loomstate fabric (%)

Warp tension distribution

18 picks/cm (1.75 kN)

22 picks/cm (1.75 kN)

26 picks/cm (1.75 kN)

18 picks/cm (1 kN)

22 picks/cm (1 kN)

26 picks/cm (1 kN)

Figure 5. Warp crimp distribution in the fabric on the loom over the warp width for 16.4 tex (Ne 36/1) weft yarn

8 7 6 5 4 3 2 1 0 6

19

31

43 54 64 72 84 96 111 Measuring point (as from left temple-cm)

128

142

160

can be explained as follows: after the fabric leaves the temples, holding forces applied to fabric by temples from both fabric selvedges are removed and the tensioned weft in the fabric causes the fabric to contract widthwise. But, the contraction is not homogenous over the fabric width. It is higher at edge zones of the fabric than in the middle part. It is easy to see this by putting marks on the fabric over the width at equal distances and then by measuring these distances after the fabric passes the temples. More contraction at edge zones causes more relaxation of the weft at edge zones than the middle part. This eases the pressure of the weft yarn on the warp more at edge zones than in the middle and therefore warp crimp becomes higher at edge zones as Figures 4-6 show. This warp crimp distribution is in contrast to warp tension distribution over the warp width. Table IV shows percent warp crimp reduction with respect to maximum warp crimp in the fabric on the loom over the fabric width for each fabric construction. y in Table IV represents the percent ratio of warp crimp reduction (i.e. the difference between the

Figure 6. Warp crimp distribution in the fabric on the loom over the warp width for 11.8 tex (Ne 50/1) weft yarn

IJCST 22,4

280

maximum and minimum warp crimps over the fabric width) to maximum warp crimp in the fabric on the loom. As seen from Table IV, the lowest percent warp crimp reduction over the warp width occurs in the fabric woven with 29.5 tex (Ne20/1) weft yarn and 1.0 kN adjusted total warp tension and the highest warp crimp reduction occurs in the fabric woven with 11.8 tex (Ne50/1) weft yarn and 1.75 kN adjusted warp tension. For each weft yarn number and warp tension, warp crimp reduction over the fabric width decreases in general with increasing weft density. Warp and weft crimp distribution in grey fabric over the fabric width Figures 7-9 show warp crimp distribution in grey fabric over the fabric width for 29.5 tex (Ne 20/1), 16.4 tex (Ne 36/1) and 11.8 tex (Ne 50/1) cotton weft yarns, respectively. In all of three figures, warp crimp in the middle of the fabric width is higher than warp crimp in edge zones of the fabric width. The reason is that when the fabric is taken off the 14 picks/cm (1.75 kN)

18 picks/cm (1.75 kN)

22 picks/cm (1.75 kN)

14 picks/cm (1 kN)

18 picks/cm (1 kN)

22 picks/cm (1 kN)

Figure 7. Warp crimp distribution in grey fabric over the fabric width for 29.5 tex (Ne 20/1) weft yarn

Warp crimp in grey fabric (%)

16 14 12 10 8 6 6

20

41

63

94

120

139

158

Measuring point (as from left fabric selvedge-cm)

18 picks/cm (1.75 kN)

22 picks/cm (1.75 kN)

26 picks/cm (1.75 kN)

18 picks/cm (1 kN)

22 picks/cm (1 kN)

26 picks/cm (1 kN)

Figure 8. Warp crimp distribution in grey fabric over the fabric width for 16.4 tex (Ne 36/1) weft yarn

Warp crimp in grey fabric (%)

10

9

8

7

6 6

20

41

63

94

120

Measuring point (as from left fabric selvedge-cm)

139

158

18 picks/cm (1.75 kN)

22 picks/cm (1.75 kN)

26 picks/cm (1.75 kN)

18 picks/cm (1 kN)

22 picks/cm (1 kN)

26 picks/cm (1 kN)

Warp tension distribution

Warp crimp in grey fabric (%)

7 6.5

281

6 5.5 5 4.5 6

20

41

63

94

120

139

158

Figure 9. Warp crimp distribution in grey fabric over the fabric width for 11.8 tex (Ne 50/1) weft yarn

Measuring point (as from left fabric selvedge-cm)

loom the forces affecting the fabric on the loom are removed and the fabric gets relaxed. During relaxation of the grey fabric, lengthwise contraction occurs more than widthwise contraction as fabric partly contracts widthwise after passing the temples. Lengthwise contraction occurs more in the middle area of the fabric due to higher warp tension than selvedge area of the fabric which has lower warp tension. As a result of this, warp crimp in grey fabric has higher values in the middle part than at selvedge area. Figures 10-12 show weft crimp distribution in grey fabric over the fabric width for 29.5 tex (Ne 20/1), 16.4 tex (Ne 36/1) and 11.8 tex (Ne 50/1) cotton weft yarns, respectively. In all of three figures, weft crimp is higher in edge zones of the fabric than in the middle part. It is thought that higher weft crimp at edge zones is due to the slip and therefore relaxation of the weft during beating up. In grey fabric, weft crimp distribution over fabric width is opposite to warp crimp distribution. Warp crimp is higher and weft crimp is lower in the part of the fabric width where warp tension is higher and vice versa. 14 picks/cm (1.75 kN)

18 picks/cm (1.75 kN)

22 picks/cm (1.75 kN)

14 picks/cm (1 kN)

18 picks/cm (1 kN)

22 picks/cm (1 kN)

Weft crimp in grey fabric (%)

3.8 3.6 3.4 3.2 3 2.8 30

65 95 125 Measuring point (as from left fabric selvedge-cm)

155

Figure 10. Weft crimp distribution in grey fabric over the fabric width for 29.5 tex (Ne 20/1) weft yarn

IJCST 22,4

18 picks/cm (1.75 kN)

22 picks/cm (1.75 kN)

26 picks/cm (1.75 kN)

18 picks/cm (1 kN)

22 picks/cm (1kN)

26 picks/cm (1 kN)

282

Figure 11. Weft crimp distribution in grey fabric over the fabric width for 16.4 tex (Ne 36/1) weft yarn

Weft crimp in grey fabric (%)

6.5 6 5.5 5 4.5 4 3.5 3 30

65

95

125

155

Measuring point (as from left fabric selvedge-cm)

18 picks/cm (1.75 kN)

22 picks/cm (1.75 kN)

26 picks/cm (1.75 kN)

18 picks/cm (1 kN)

22 picks/cm (1 kN)

26 picks/cm (1 kN)

Figure 12. Weft crimp distribution in grey fabric over the fabric width for 11.8 tex (Ne 50/1) weft yarn

Weft crimp in grey fabric (%)

8 7.5 7 6.5 6 5.5 5 4.5 30

65

95

125

155

Measuring point (as from left fabric selvedge-cm)

Table V shows deviation of warp and weft crimps from their maximum values in percentage. x represents warp crimp deviation and y represents weft crimp deviation in percentage. As seen from the table, both weft crimp and warp crimp in grey fabric have deviations over the fabric width between 5 and 15 per cent depending on fabric construction. This deviation may cause variations in fabric properties between middle part and edge zones of the fabric. Deviation in warp crimp increases and in weft crimp decreases with increasing warp tension. Both warp and weft crimp deviations increase as weft yarn gets finer. The weft density does not show a significant effect on warp and weft crimp deviations. 4. Conclusions The distribution of warp end tension over the warp width is not uniform. Warp tension is higher in the middle part than in the edge zones of the warp width. Warp tension change

Weft yarn number 29.5 tex (Ne 20/1) (carded) 16.4 tex (Ne 36/1) (carded) 11.8 tex (Ne 50/1) (carded)

Change of warp crimp in grey fabric x-(%) Adjusted total Adjusted total Weft warp tension warp tension density (1.75 kN) (1 kN) (picks/cm)

Change of weft crimp in grey fabric y-(%) Adjusted total Adjusted total warp tension warp tension (1.75 kN) (1 kN)

14 18 22

7.0 6.5 6.7

9.4 8.1 5.6

7.6 3.9 4.9

6.8 4.9 9.8

18 22 26

7.9 6.8 7.2

9.5 9.0 7.4

13.7 10.6 12.5

23.9 4.1 13.6

18 22 26

6.9 8.6 9.7

12.6 10.3 14.5

14.5 15.2 14.9

13.8 8.2 9.7

over the warp width increases as warp tension increases and weft yarn thickness and weft density decrease. Warp crimp in the fabric on the loom is higher in the edge zones than the middle of the fabric. Change in warp crimp in the fabric on the loom over the fabric width increases with increasing warp tension and decreasing weft yarn thickness and weft density. Warp crimp in the middle is higher than warp crimp in edge zones of the grey fabric. Warp crimp change over the fabric width increases as warp tension increases and weft yarn thickness decreases. No significant effect of weft density on this change has been found. Weft crimp in the edge zones is higher than in the middle of the grey fabric. Weft crimp change over the fabric width increases as weft yarn thickness decreases and warp tension increases. No significant effect of weft density on this change has been observed. The reason behind warp tension and warp and weft crimp variations over fabric width is that weft yarn slips towards inside fabric at selvedges and gets relaxed during beat up. Reducing weft yarn slip and therefore weft yarn relaxation during beat up will reduce warp tension and warp and weft crimp variations and improve the uniformity of fabric properties over the fabric width. False selvedge construction and shed closing angle of false selvedges are thought to be important to adjusted holding forces applied to weft yarn to minimize weft slip and weft relaxation during beat up.

References ASTM Standards Specifications (2004), IS: ASTM D3883-04, ASTM International, West Conshohocken, PA. Ludwig, H.W. and Gries, T. (2003), “Measurements carried out to minimise warp tension variations in weaving machines”, Melliand Textilberichte, 2 June, pp. 55-8. Ozkan, G. (2005), “Investigation of crimp-warp tension relation in woven fabrics”, PhD thesis, Uludag University, Bursa.

Warp tension distribution

283

Table V. The change of warp and weft crimp in grey fabric over the fabric width

IJCST 22,4

Weinsdorfer, H., Wolfrum, J. and Stark, U. (1991), “The distribution of the warp end tension over the warp width and how it is influenced by the weaving machine setting”, Melliand Textilberichte, Vol. 72, pp. 903-5. Weinsdorfer, H., Azarschab, M., Murrweib, H. and Wolfrum, J. (1988), “Effect of the selvedge and the temples on the running performance of weaving machines and on the quality of the fabric”, Melliand Textilberichte, Vol. 35, pp. 364-72.

284 Corresponding author Gulcan Ozkan can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

Fabric-skin friction property measurement system

Fabric-skin friction property

Xu Wang, Ping Liu and Fumei Wang College of Textiles, Donghua University, Shanghai, China Abstract

285 Received 4 August 2009 Accepted 12 December 2009

Purpose – Understanding the friction property between fabric and skin is an important issue in both product design and comfort evaluation. The purpose of this paper is to present a new type of tester to evaluate the friction between skin and clothing. Design/methodology/approach – The proposed fabric-skin friction tester consists of test host, friction equipment, control box, and computer processing system. The suitable testing conditions (tension, friction velocity, and sample width) are recommended by comparing the parameters and actual test circumstance. Findings – As the stress history characteristic of skin, the friction test should be repeated since the second time for each fabric and skin friction pair. The fluctuation of single friction curve and dispersion degree of several friction curves in repeated trials are both smallest under the recommended test conditions (200 cN tension, 500 mm/min friction velocity and 10 cm wide sample) for woven fabric. Originality/value – The proposed fabric-skin friction tester is adapted to measure the dynamical friction properties between human skin and clothing. This is original in comparison to the conventional research method usually found in the literature. Keywords Fabrics, Friction, Skin (body), Textile testing Paper type Research paper

1. Introduction Friction property between fabric and skin is an important topic in clothing comfort and skin healthy. In the early 1980’s, the influence of skin humiture on the perception of fabric texture and pleasantness were investigated. The friction between fabric and human skin was measured by a simple tensometer under various environmental conditions: neutral (comfortable), hot-dry and hot-humid (Gwosdow et al., 1986). Later in the 1990s, the influence of moisture, fiber type and fabric construction parameters on the friction properties between fabric and skin were also investigated by similar method (Kenins, 1994). Subsequently, friction coefficient was proposed to describe the friction properties between fabric material and human skin. A small piece of fabric sample was pasted on the contactor of friction tester before testing. And then the friction coefficient between various fabric materials and individual skin could be measured (Zhang and Mak, 1999; Kondo, 2002; Tang et al., 2007). In other studies, some elastic materials such as silicone were applied as mechanical skin equivalents to rub with fabric. The friction properties between artificial skin and fabric were measured (Ramkumar et al., 2003a, b; Derler et al., 2007) by traditional friction tester. Recently, finite element method (EFE) was introduced to simulate the friction interaction between fabric and skin to analyze the mechanical characteristics (Wu et al., 2003; Wu et al., 2004; Xing et al., 2006). The authors of this paper gratefully acknowledge the support of the National Natural Science Foundation of China (No. 50973015).

International Journal of Clothing Science and Technology Vol. 22 No. 4, 2010 pp. 285-296 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011048303

IJCST 22,4

286

It is found that there is a dearth of suitable instrument simulating the actual dressing situation and its general applicable test method until now. And there has been lack of the dynamic and static friction database between fabric and skin. Under such a circumstance, inventing an appropriate friction instrument and its measurement system become the primary issue. This paper is organized as follows. The fabric-skin friction tester, testing method, materials and experiment conditions is depicted in Section 2. The selected friction characteristics are explained in Section 3, including test parameters and preferred conditions. The conclusions are drawn in Section 4. 2. Experimental 2.1 Friction instrument and method Figure 1 is the schematic diagram of fabric-skin frictional tester remanufactured from a domestic strength-elongation tester (XL-1). It is consists of four major parts: test host, friction equipment, control box, and computer processing system. The friction equipment is assembled by friction trestle, clamp, armrest, roller and attachments, as shown in Figure 2. To ensure a smooth movement, the forearm of the subject under

Upper clamp Sensor

Forearm or simulated forearm Armrest

Fixation jig Computer

Control box

Roller Fabric

Lower clamp

Figure 1. Schematic diagram of fabric-skin frictional tester

F Forearm or simulated forearm Roller

Fabric

Figure 2. Profile schematic of friction assembly

T

testing is laid on the armrest and is fastened with jigs. The top end of the fabric is held by the upper clamp, while the other end of the fabric is held by the lower clamp which exerts uniform tension (T). The fabric is then pulled across the forearm (human forearm or artificial forearm) and the roller. The measurable force (F) of the resistive sensor ranges from 0 to 3,000 cN, and the precision is ^ 1 percent. Sensor and the upper part of the fabric should remain stationary during the test. The friction equipment pulls forearm up and down and thereby provides relatively sliding friction between fabric and skin. The friction displacement ranges from 0 to 500 mm, and can be adjusted freely. The testing velocity is configured from 100 to 1,000 mm/min (stepless speed regulation). The displacement distance and its corresponding friction force are recorded by the connected computer system. Hence, the relation curve can be obtained simultaneously.

Fabric-skin friction property

287

2.2 Materials and conditions 2.2.1 Subject and experiment samples. The subject is asked to rinse any creams or external medication prior to the test. The friction position is selected from the middle of lateral skin of forearm. In order to ensure the comparability and repeatability, each experiment should begin with the label marked on the skin before testing. 2.2.2 Experiment conditions. Various woven fabric are selected as the test samples (Table I) formatted as 10 £ 50 cm and 15 £ 50 cm, respectively. Tests are carried out in a climatic chamber with the temperature at 208C (^ 2 percent) and the relative humidity at 50 percent (^ 5 percent). Subject is asked to rest in this conditioned environment for 30 min before being tested. From our observation, this pre-conditioned procedure is helpful in smoothing the subject, both physically and psychologically. 3. Results and discussion 3.1 One friction curve and its test parameters Figure 3 is the friction characteristic curve which describes the relationship between friction and displacement. Both skin and fabric are stretched under continuous elastic deformation at first stage. Friction monotonically increases from 0 until the maximum. Later on, friction will quickly break through the peak value and declines rapidly simultaneously. Skin and fabric change from static friction state into dynamic friction state. The following three parameters are defined to characterize the frictional properties between fabric and skin for one representative friction curve: (1) Fs (static friction/cN): the crest value of the friction characteristic curve. (2) Fd (average dynamic friction/cN): the mean value of friction force during the selected smooth friction displacement, which is computed by:

Fabric Chiffon Crepe satin Linen fabric Khaki Lining fabric

Fabric density (threads/ Weave Thickness Yarn count (tex) Weight 10 cm) warp £ weft structure (mm) warp £ weft (g/m2) Material 480 £ 300 1,200 £ 580 310 £ 250 640 £ 270 450 £ 330

1/1 plain 5/3 satin 1/1 plain 3/1 twill 1/1 plain

0.246 0.227 0.383 0.534 0.066

13.3 £ 11.1 4£4 27.8 £ 27.8 18.2 £ 38.7 6£6

107.02 73.27 135.79 238.68 46.80

Polyester Silk Linen Cotton Nylon

Table I. Specification of experiment fabric

IJCST 22,4

500 Fs

400

288

F (cN)

Fd

300 200

The selected friction displacement

100

Figure 3. Fabric-skin friction characteristic curve

0

50

100 S (mm)

1 Fd ¼ 100

Z

150

200

150

FdS

ð1Þ

50

It is found that friction behind 50 mm friction displacement changes into relative stable region no matter what kind of friction pair. Therefore, the friction curve during 50-150 mm friction displacement is sufficient to describe the dynamic friction properties between common fabric and skin. (3) FMD (coefficient of mean deviation of Fd/%): the coefficient of mean deviation of friction during the selected friction displacement: R 150 jF 2 F d jdS £ 100ð%Þ ð2Þ FMD ¼ 50 100 £ F d 3.2 Stress history of skin and test parameter in repeated trials When subjected to stress, skin will stretch rapidly at first and then demonstrate very limited extension. If stress has already been exerted and removed afterwards, the reaction time of the first extension phase will delay when the subsequent stress is exerted on skin (Tang et al., 2007). This is called stress history phenomenon. As shown in Figure 4, the stress history phenomenon also occurs in the fabric and skin successive friction experiments. Friction of the first curve (the red curve) is obviously lower than that of the following curves, and its fluctuation is greater. Friction tends to be stable since the second friction curve (the green curve) and the subsequent friction curves (the gray curves) tend to concentrate relatively. Static friction points of the subsequent friction curves delay slightly in comparison. It is believed that skin and its subcutis have not adapted to the extension effect and cannot be stretched fully when rubbing with fabric for the first time. Hence, the reaction time of its first extension phase is shorter, and the friction is comparatively smaller. In our daily life, friction between fabric and skin is happened repeatedly. It is regarded that investigation of friction curve since second test is more practical due to the stress history characteristic. The dispersion degree of friction curves in repeated tests

Fabric-skin friction property

250

200

F (cN)

Second friction curve 150

289 First friction curve

100

50

0

0

50

100 S (mm)

150

Figure 4. Stress history of the fabric and skin friction test

200

is different in various fabric and test conditions. To compare the dispersion degree, the coefficient of variation is defined as follows: CV ¼

s £ 100ð%Þ Fd

ð3Þ

where s and F d denote the standard deviation and mean value of N times of Fd, respectively. The test time N will be discussed in Section 3.7. From our observation, four times of friction test is good enough for woven fabric under the recommended test conditions. However, the dispersion degree of several friction curves is relatively large under some test conditions, each friction pair should be repeated ten times since the second in this paper. 3.3 The relationship between Fs and Fd There is a good accordance between Fs and Fd of same friction curve under each test. The linear approximation is shown in Figure 5 as the solid line. The regression coefficient is 0.9958, indicating that these two parameters have close relationship without independent. In general, Fs is 7 percent ^ 2.5 larger than Fd. Thus, only Fd would be discussed in the following comparative analysis. 3.4 Tension preferences Tension should be applied on fabric to ensure effective contact and friction. Fabric cannot be straightened fully and contact with skin effectively under small tension, which will result in friction instability. In contrast, the greater the tension is, the more severely the skin is stretched and distorted. Besides, the contact area will decrease continuously with the fabric shrinkage under great tension. Thus, the fluctuation of subcutaneous elasticity and the friction signal will be greater. The transient great tension will appear during strenuous exercise, resulting in great friction, skin discomfort or injury. The friction comparisons are carried out on the same subject under 500 mm/min friction velocity. Different tension (100, 150, 200, and 250 cN) is separately imposed

IJCST 22,4

600 y = 1.0535x + 4.2219 R2 = 0.9958

290

Fs (cN)

500

400

300

200

100 100

Figure 5. Relation between Fd and Fs

200

300

400

500

600

Fd (cN)

on a piece of 10 cm wide fabric to maintain straight without elongation (or small elongation without serious necking). As shown in Figure 6, Fd increase with tension for different kinds of friction pair. Common woven fabric can be straightened fully without necking phenomenon under 200 cN tension. In Figures 7 and 8, FMD and CV are both smallest under 200 cN tension. It indicates that the dispersion degree in repeat tests and the fluctuating error on one friction curve are both relatively small. Hence, 200 cN tension is recommended as the conventional test condition. 3.5 Friction velocity preferences Various friction velocities (200, 300, 400, 500 and 600 mm/min) are applied in the test for the same subject under 200 cN tension and 10 cm wide samples. As shown in Figure 9, 600 500

Fd (cN)

400

Chiffon Crepe satin Linen fabric Khaki Lining fabric

300 200 100

Figure 6. Relation between tension and friction

0

100 cn

150 cn

200 cn

Tension (cN)

250 cn

2.0 100 cN 150 cN 200 cN 250 cN

1.8 1.6

Fabric-skin friction property

FMD (%)

1.4

291

1.2 1.0 0.8 0.6 0.4 0.2 0.0 Chiffon

Crepe satin Linen fabric Sample

Khaki

Lining fabric

Figure 7. FMD comparison under different tension

14 100 cN 150 cN 200 cN 250 cN

12

CV (%)

10 8 6 4 2 0

Chiffon

Crepe satin

Linen fabric Sample

Khaki

Lining fabric

Fd does not significantly increase or decrease with the friction velocity. In Figures 10 and 11, FMD and CV are both the smallest under 500 mm/min friction velocity. It demonstrates that the dispersion degree in repeated tests and the fluctuating error of one friction curve are both the smallest. It is believed that muscle deformation caused by stretch cannot recover promptly if the friction velocity is very slow. And stick-slip phenomenon, which brings about signal fluctuation often occurs. In contrast, some of the microcosmic surface friction information can be ignored under high-speed friction. And the tester itself will vibrate which will influence the stability of friction signal.

Figure 8. CV comparison under different tension

IJCST 22,4

500

400

Fd (cN)

292

300 Chiffon Crepe satin Linen fabric Khaki Lining fabric

200

100

Figure 9. Relation between velocity and friction

0

200

300

400 500 Velocity (mm/min)

600

2.0 200 mm/min 300 mm/min 400 mm/min 500 mm/min 600 mm/min

1.8 1.6

FMD (%)

1.4 1.2 1.0 0.8 0.6 0.4 0.2

Figure 10. FMD comparison under different velocity

0.0

Chiffon

Crepe satin Linen fabric Sample

Khaki

Lining fabric

In general, friction velocity has little influence on the friction between fabric and skin, while it has a definite influence on the stability of friction signal. Therefore, the relatively moderate and recommended friction velocity is 500 mm/min for conventional test. 3.6 Sample width preferences Since average length of forearm is about 20 cm, specimen width is selected as 10 and 15 cm to avoid fabric from deflecting or shearing. As shown in Figure 12, friction differs in sample width under 200 cN tension and 500 mm/min friction velocity. Fd of the 15 cm wide sample is bigger than that of the10 cm wide sample, because the contact area between fabric and skin increases.

Fabric-skin friction property

12 200 mm/min 300 mm/min 400 mm/min 500 mm/min 600 mm/min

10

CV (%)

8

293 6 4 2 0

Chiffon

Crepe satin Linen fabric Sample

Khaki

Lining fabric

Figure 11. CV comparison under different velocity

600 10 cm 15 cm

560 520

Fd (cN)

480 440 400 360 320 280 Chiffon

Crepe satin Linen fabric Sample

Khaki

Lining fabric

Results show that friction curve of 10 cm wide sample is more stable. In Figures 13 and 14, FMD and CV of 10 cm wide samples are both smaller than that of the 15 cm width samples. It indicates that the dispersion degree of friction in repeated tests and its fluctuating error in one test are both small. Some specialists believe that the sample size should be big enough to reflect the character of the complete fabric texture. The tester will overload and a higher power sensor will be demanded if the sample is very big. Besides, dynamic impact will easily rise and test error will also increase. While real information of fabric would not be reflected if the sample is very small, thus the experiment meaning will lose (Zhang and Chen, 1998). In theory, the real wearing situation could be actually simulated by large contact area

Figure 12. Relation between sample width and friction

IJCST 22,4

2.4 10 cm 15 cm 2.0

1.6 FMD (%)

294

1.2

0.8

0.4

Figure 13. FMD comparison under different sample width

0.0 Chiffon

Crepe satin Linen fabric Sample

Khaki

Lining fabric

10 10 cm 15 cm 8

CV (%)

6

4

2

Figure 14. CV comparison under different sample width

0 Chiffon

Crepe satin Linen fabric

Khaki

Lining fabric

Sample

between fabric and skin. However, since both ends of human forearm are uneven, the wider the sample is, the more seriously the fabric will deflect. And the error will be greater. Thereby, the recommended appropriate sample width is 10 cm. 3.7 Friction time preferences Data shows that CV of the repeated tests are relatively small (less than 5 percent) under recommended test conditions for woven fabric. To enhance work efficiency, friction time should be massively reduced. This experiment belongs to relatively large discrete

sample because the experiment process is often influenced by many subjective and objective factors. The relative uncertainty E is ^ 5 percent, the probability level is 95 percent (significant level a ¼ 5 percent), and the statistic t (a, N 2 1) is 1.96. According to the following equation, friction time N equals to 3.85. It indicates that four times of test at most would meet the requirement: N¼

t 2 CV2 1:962 CV2 ¼ ¼ 0:154CV2 E2 52

ð4Þ

As a result, each friction pair should be repeated four times of friction test since the second time under the recommended test conditions for woven fabric. While the appropriate friction time for some other special fabric could be calculated by equation (4) in accordance with its CV (Yu et al., 2004). 4. Conclusions The new type of fabric-skin friction tester developed and presented in this work is specifically adapted to measure the friction properties between human skin and clothing in motion state. Four parameters (Fs, Fd, FMD and CV) are proposed to describe the friction behavior. It is found that Fs and Fd of single friction curve have a close relationship without any independence. The first friction curve is obviously lower than the following ones’ because of the stress history characteristic. For woven fabric, the friction test of each friction pair should be repeated four times. The influence of test condition on the friction properties has also been investigated. Friction increases along with tension and sample width, basically has no effect on its velocity. The fluctuation of single friction curve and dispersion degree of several friction curves in repeated trials are both smallest under the recommended test conditions (200 cN tension, 500 mm/min friction velocity and 10 cm wide sample) for woven fabric. References Derler, S., Schrade, U. and Gerhardt, L.C. (2007), “Tribology of human skin and mechanical skin equivalents in contact with textiles”, Wear, Vol. 263, pp. 1112-16. Gwosdow, A.R., Stevens, J.C., Berglund, L.G. and Stolwijk, J.A. (1986), “Skin friction and fabric sensations in neutral and warm environments”, Textile Research Journal, Vol. 56 No. 9, pp. 574-80. Kenins, P. (1994), “Influence of fiber-type and moisture on measured fabric-to-skin friction”, Textile Research Journal, Vol. 64 No. 12, pp. 722-8. Kondo, S. (2002), “The frictional properties between fabrics and the human skin, Part 1: factors of human skin characteristics affecting the frictional properties between fabrics and the human skin”, Journal of the Japan Research Association for Textile End Uses, Vol. 43 No. 4, pp. 36-47. Ramkumar, S.S., Wood, D.J., Fox, J.K. and Harlock, S.C. (2003a), “Developing a polymeric human finger sensor to study the frictional properties of textiles, Part 1: artificial finger development”, Textile Research Journal, Vol. 73 No. 7, pp. 469-73. Ramkumar, S.S., Wood, D.J., Fox, J.K. and Harlock, S.C. (2003b), “Developing a polymeric human finger sensor to study the frictional properties of textiles, Part 2: experimental results”, Textile Research Journal, Vol. 73 No. 7, pp. 606-10. Tang, W., Zhu, H., Meng, Y.J. and Ge, S.R. (2007), “Tribological testing between human skin and fabric”, Lubrication Engineering, Vol. 32 No. 2, pp. 4-6.

Fabric-skin friction property

295

IJCST 22,4

296

Wu, J.Z., Dong, R.G., Schopper, A.W. and Smutz, W.P. (2003), “Analysis of skin deformation profiles during sinusoidal vibration of fingerpad”, Annals of Biomedical Engineering, Vol. 31 No. 7, pp. 867-78. Wu, J.Z., Dong, R.G., Schopper, A.W. and Smutz, W.P. (2004), “Analysis of effects of friction on the deformation behavior of soft tissues in unconfined compression”, Journal of Biomechanics, Vol. 37 No. 1, pp. 147-55. Xing, M.Q., Sun, Z.G. and Pan, N. (2006), “An EFE model on skin-sleeve interactions during arm rotation”, Journal of Biomechanical Eineering, Vol. 128 No. 6, pp. 872-8. Yu, X.F., Bao, Y.P., Wu, Z.P. and Liu, R.H. (2004), Textile Material Testing Technology, China Textile Press, Beijing, pp. 12-13. Zhang, M. and Mak, A.F. (1999), “In vivo friction properties of human skin”, Prosthetics and Orthotics International, Vol. 23 No. 2, pp. 135-41. Zhang, Y. and Chen, H. (1998), “Investigation on testing fabric surface friction properties by the capstan method”, Journal of Tianjin Institute of Textile Science and Technology, Vol. 17 No. 4, pp. 87-91. Further reading Liu, C.H. (1991), Skin Pathology and Physiology, China Medical Science and Technology Press, Beijing, pp. 503-4. Corresponding author Fumei Wang can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

Classification of body shape characteristics of women’s torsos using angles Wookyung Lee School of Natural Science and Ecological Awareness, Graduate School of Humanities and Sciences, Nara Women’s University, Nara, Japan, and

Classification of body shape characteristics 297 Received 17 November 2009 Accepted 14 January 2010

Haruki Imaoka Department of Health Science and Clothing Environment, Faculty of Human Life and Environment, Nara Women’s University, Nara, Japan Abstract Purpose – The purpose of this paper is to classify body shapes using angular defects instead of sizes. Design/methodology/approach – A large amount of dimensional data from a national anthropometry survey was analysed, and a basic pattern and its polyhedron were also used to create a three-dimensional body shape from three body sizes. Using this method, the sizes were converted into nine angular defects. Findings – The authors could define the factors explaining body shape characteristics and classify the body shapes into four groups. The four groups could be characterised by two pattern making difficulties of the upper and lower parts of the body as well as by two proportions, of waist girth to bust girth and bust girth to back length. Furthermore, depending on the age, the authors could understand body shape by the angle made. Originality/value – Using a polyhedron model, the angles could be calculated using an enormous existing data set of sizes. An angular defect serves as an index to indicate the degree of difficulty for developing a flat pattern. If an angular defect of the bust is large, it is difficult to make a paper pattern of a bust dart. On the other hand, if an angular defect of the waist is large, it is easy to make a paper pattern of a waist dart. Thus, each body shape could be simultaneously characterized by two difficulty indices and two proportions of sizes. Keywords Body regions, Classification, Women, Gaussian processes Paper type Research paper

Introduction Measuring body shapes and clarifying their statistical features are fundamental for producing good quality garments at lower production costs. The sizes for representing body shapes have been defined and measured to gather large amounts of data efficiently. Furthermore, it has recently become possible to measure three-dimensional body shapes directly. Thus, the global trend in the production of garments is changing from size to shape oriented. However, we are still discovering how three-dimensional shapes can be created and their essential features extracted. The aim of this paper is to classify body shape characteristics according to angles instead of sizes. Many categorised body shape systems have been used in the

International Journal of Clothing Science and Technology Vol. 22 No. 4, 2010 pp. 297-311 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011048312

IJCST 22,4

298

clothing industry. They are size-oriented systems, whereas our target is a shape-oriented system, one that regards two similar shapes as the same (Cho et al., 2006). Some previous research about young women and men has used concentrated Gaussian curvatures (a.k.a. angular defects) in detail (Masuda and Imaoka, 2004, 2005). Our research is about broad age groups of women. To implement this paper, we analysed a large amount of dimensional data from a national anthropometry survey. A basic pattern and its polyhedron were used to create a three-dimensional body shape from some sizes. This paper limits the subjects to adult female torsos, the phase structures of which are composed of 20 vertices and 31 triangles of polyhedra. Essentially, nine types of angular defect were used to identify features under the consideration of left and right symmetry. The samples were 122 polyhedra, extracted from Japanese body measurement data of adult women (ages 16-79) (HQL, 1997), measured from 1992 to 1994. We determined four categories using principal component analysis and examined the categories with respect to age. We used the basic pattern of the Japanese new Bunka style (Miyoshi, 2004) to create a polyhedral frame. To create the basic pattern, the three sizes of bust girth, waist girth and back length were used. Experimental method Polyhedral model A polyhedral model was prepared as follows: measure sizes; make a basic pattern; close each hole around the neck-base line, armhole line and waistline; prepare a polyhedral model and measure angles. Note that it was not prepared using direct three-dimensional measurements. The polyhedron prepared by the above method is a closed one without holes. The reason for choosing a closed polyhedron is that the actual three-dimensional shape is determined uniquely (though there are many ways to prepare a polyhedron). The solid body model is very useful for directly understanding certain features. Armhole line does not influence the angles and the polyhedra can be classified regardless of such details. We can classify forms regardless of their sizes because angles are invariant under the similarity transformation. A vertex on each hole is a fiducial point to close the hole. Hereafter, this vertex, corresponding to the vertex of the pyramid created after filling the hole, is called the focus. The polyhedron contains vertices (e.g. bust dart points) and foci, which are drawn up and connected to form triangles. Polyhedra were drafted as follows: we used expert pattern system (Corporation FAS, Easy Version 1.01) to make basic clothing patterns graded with their sizes. When bust girth, waist girth and back length are provided as input, the system yields basic clothing patterns of the new Bunka style graded with their sizes as output (Figure 1). Next, we used AutoCAD 2005 (Autodesk Corporation, Educational Version) to draft a 3D polyhedral model from plane basic patterns graded with individual sizes. The polyhedral models (Figure 2) are closed patterns involving foci at the opened neck-base line, armhole line and waistline obtained after drafting basic patterns for the torso according to completion lines. We explain in detail the method for obtaining the focus of the neck-base line (G1) and that of the armhole line (G7) in Figure 3 and the waistline (G13) in Figure 4. The focus is a vertex calculated to close the hole at neck-base line, armhole line or waistline: . Focus of the neck-base line (G1) (Figure 3(a) and (c)). We turn the front and back patterns to centre the side neck point and fit the shoulder lines from the side neck point to the shoulder dart together. G1, the focus of the neck-base line, is a point

.

.

of intersection between an extension line of the front, the centre line of the back torso and an extension line connecting the side neck point and shoulder dart. Focus of the armhole line (G7 ) ( Figure 3(a) and (c)). We obtain A, the point of intersection of an extension of the front and back side lines. Next, we draw a circumscribed circle going through three points, the intersection point A and the front and back axilla points A0 and A00 . A vertical bisector is drawn on the line segment A0 A00 from the centre of the circumscribed circle. The intersecting point G7 on which an extension of the vertical bisector and the circumscribed circle meet is the focus of the armhole line, G7. It contains information about shoulder dart angles, indicated with a dotted line in Figure 3(b), together with the name of each vertex in Figure 3(c). Focus of the waistline (G13 ) (Figure 4(e) and (f )). After closing all waistline darts, the side lines to the front and back of the basic pattern are closed. G13, the focus of the waistline, is the point of intersection of an extension of the front and back centre lines of the basic pattern.

Classification of body shape characteristics 299

Figures 3(c) and 4(f), obtained as shown, are development charts showing half of the body, while that of the whole body is completed assuming left and right symmetry about the front centre line, as shown in Figure 5. Briefly, the polyhedral model comprises triangles connected with 13 types of vertices, including the foci (G1, G7 and G13) and internal dart points (G2-G6, G8-G12).

Figure 1. Graded new Bunka style basic pattern from expert pattern system G1 G7 G6

G1 G7 G6 G5 G2 G4 G3

G1 G7 G10 G12 G8 G11 G9

G1 G10 G7 G9

G8 G6 G5 G4

G5 G4

G10 G11 G1

G7 G6 G5 G2

G13

G13

Front

G13

Back

G13

Side

Top

Figure 2. Polyhedral model and name of each vertex

IJCST 22,4

G3

G2

G4 G5

G1

300

A"

A

G1

G6

G7 G10

G12 A'

Figure 3. First method of making a polyhedral model (G1-G12)

G11

G7 G8

G9

(a)

(b)

(c)

G7

G7 G11

G8

G6 G5

G9

G3

G4

Figure 4. Second method of making a polyhedral model (G13)

(d)

G13 (e)

G13 (f)

Polyhedral model and angular defects Various methods exist for classifying 3D body shapes; however, in the present study, we use the method employing angular defects, such as those of the vertices on polyhedral models (Calladine, 1986; Imaoka and Masuda, 1996; Maltret and Daniel, 2002; Alboul et al., 2005). The angle remains constant under similarity transformation, which is an essential feature of this method. Angular defects are used to evaluate the angles around the vertices by equation (1): Angular defect ¼ 3608 2 sum total of interior angles around the vertex

ð1Þ

This means that larger the value of the angular defect, sharper shape of the top on the polyhedron. The value of a vertex on the plane is zero, the sphere-point is positive and the

G11 G12

G9

G7 G8 G11 G9

G9

G12 G1

G8

Classification of body shape characteristics

G11

G10

G8

G10

G7

G7

G6

G7

G6

G6 G5 G2 G5 G4 G3 G4

G6

301

G8 G11 G9

Figure 5. Plane development chart of a polyhedral mode

G13 Note: Figures 3(c) + 4(f)

saddle-point is negative. The absolute value is an index to indicate the degree of difficulty to flatten a part. We follow the Gauss-Bonnet theorem, essentially, a law of conservation of curvature, indicating that all the angular defects on the vertices of the polyhedral model do not comprise independent characteristics and that their sum is constant: On a polyhedron; sum of angular defects

ð2Þ

þ sum of external angles of a boundary line ¼ 3608 £ Euler number The Euler number of the closed polyhedron used in this paper is 2; therefore, the right side of the equation becomes 7208. Since there is no boundary line, the following formula applies to a closed polyhedron: Sum of angular defects ¼ 7208

ð3Þ

Although the calculation of angles can be achieved by the second cosine law and by measuring the length of the sides of a triangle, a simple method exists that does not use the focus; the angle is calculated from the angle formed at the intersection of the end of each side of the hole. Because the sum of the angular defects is constant, the angles are not independent and it is necessary to consider multiple collinearity characteristics when dealing with statistics. Our polyhedral model is bilateral in form and symmetrical points, lines and surfaces located on both sides of the centre line of the front and back of the basic pattern can be

IJCST 22,4

302

treated the same because they have the same information. Accordingly, one polyhedron is composed of 13-point types, 30 line types and 18 triangle types. The 13 points of angular defects are indicated by symbols from G1 to G13 (Figures 2 and 5). Since G1, G2, G3, G11, G12 and G13 in the polyhedral model are situated on the centre line of the polyhedron, each number corresponds to one angle. However, since the other angular defects are left and right symmetric, each represents two angular defects. Therefore, while considering the numbers, it can be said that the polyhedral model consists of the sum of 20 vertices and 31 triangles. The angular defects from G1 to G13 contain foci G1, G7 and G13. These are the foci of neck-base line, armhole line and waistline, respectively. Therefore, the following relationship applies: Sum of angular defects of one polyhedral model ¼

X ðangular defect situated on the centre line of a polyhedronÞ X þ 2 ðangular defect of symmetryÞ

ð4Þ

¼ ðG1 þ G2 þ G3 þ G11 þ G12 þ G13Þ þ 2ðG4 þ G5 þ G6 þ G7 þ G8 þ G9 þ G10Þ ¼ 7208 Body shape size variation of subjects The Japanese Body Measurement Data of Adult Women (HQL, 1997) of Japanese Industrial Standards (JIS) L 4005-1997 includes data for 11,057 adult females aged 16-79 taken from 1992 to 1994. The number of females in the population is assumed to be 49,907,000. This paper produced a polyhedral model and performed body shape analysis using the subjects of JIS L 4005-1997 (HQL, 1997; JISC, 1997). JIS L 4005-1997 divides body shapes of Japanese adult females into A, Y, AB and B types. The JIS identified body shapes of Japanese adult females as follows: first, the data were divided by height into four types (142, 150, 158 and 166 cm, respectively). Next, the data were divided by bust girth and hip girth into many squares in a cross tabulation. Finally, the JIS defined body shapes with collated zone of the highest emerging proportion as A-type. They divided the remaining body shapes as follows: the hip girth of Y-type is 4 cm smaller than that of A-type, that of AB-type is 4 cm bigger than that of A-type and that of B-type is 8 cm bigger than that of A-type. The standard Japanese adult female has a height of 158 ^ 4 cm, bust of 83 ^ 1.5 cm and hip measurement of 91 ^ 1 cm. We considered height, bust girth, age, number of people and back length from JIS L 4005-1997 “Size data for women”. Also, we considered body type and waist girth from JIS L 4005-1997 “Sizing systems for women’s garments” (Figure 6). As is shown in Table I, combinations of JIS type (four possibilities), age (seven), height (four) and bust (six) make 672 different items. We selected 122 major items as representatives of Japanese females aged 16-79 years, where each item represents more than 20,000 persons and all items represent 8,187,000 persons, which are almost one-sixth of the total female population. Table II is a cross tabulation of 122 selected items among combinations of four JIS types and seven ages. Table III is a cross tabulation of the actual 672 items, which are

JIS L 4005–1997 size data for women (p105) 4.2 Body size of each item according to body type

for women's garments (p3)

4.2.1 A Type

Table 2-1 A type: Height 142 cm

Body type Height Bust girth Age Number of people

Name of body type Basic Bust girth measurement Hip girth Height items Waist Age 10 girth 20 30 40 50 60 70

A 158 cm 83 20 252,000

Bust girth Hip girth Waist girth Cervical-to-Posterior Waist Len. (back length)

83.0 91.0 64.2 37.8

Body type Height Age Bust girth Waist girth Cervical-to-Posterior Waist Len. (back length) Number of people

JIS type (four categories) Age (seven categories) Display range Height (four categories) Display range (cm) Bust (six categories) Display range (cm) Items (number of polyhedrons)

Classification of body shape characteristics

JIS L 4005–1997 sizing systems

A 10 10-19 142 Above 138below 146 77 Above 75.5below 78.5 672

9AR 83 91 158 64 67

70

9AR 158 20 83 64

Figure 6. Method of collecting size data to construct polyhedral models

37.8 252,000

Y 20 20-29 150

AB 30 30-39 158

B 40 40-49 166

146-154 80

154-162 83

162-170 86

303

50 50-59

60 60-69

89

92

70 70-79

78.5-81.5 81.5-84.5 84.5-87.5 87.5-90.5 90.5-94.5

translated so that the total is 122. Each item is weighted by the number of representing people. As a result of testing for goodness-of-fit, we could say 122 subjects in this paper represented the total female population ( p ¼ 0.996). The 122 body shapes were reorganised as 122 polyhedral models by AutoCAD and graded according to the new Bunka style basic pattern. The requisite measurements in the production of polyhedral models are bust girth, waist girth and back length. Because one polyhedral model is prepared for each size type, the paper analysed angular defects G1-G13 in producing 122 polyhedral models.

Table I. Combinations of items

IJCST 22,4

304

Results and discussion Descriptive statistics of angular defects Table IV indicates the weighted mean and standard deviation of each angular defect calculated from the polyhedral models of 122 sizes for 8,187,000 Japanese females aged 16-79. G4, G5, G6, G7, G8, G9 and G10 are doubled because these are symmetrical in each polyhedron. G1, G2, G3, G11, G12 and G13, located on the centre of the basic pattern, occur only once in each polyhedron. However, each value of G2, G3 and G12 is zero; hence, they were omitted. As the top of the polyhedron sharpens, the value of the angular defect increases. The angular defect decreases in the following order: waistline (G13) . armhole line (2 £ G7) . neck-base line (G1) . bust (2 £ G5) [298.58 . 227.88 . 80.08 . 36.98]. Standard deviations as the index of individual differences decreases in the order of waistline (G13) . armhole (2 £ G7) . bust dart (2 £ G5). Principal component analysis We performed principal component analysis on angular defects G1-G13 to clarify the factors expressing individual differences. This paper performed principal component analysis for 8,187,000 persons with 122 types of body shapes using SPSS 13.0 J for Windows, considering the numerical strength belonging to a polyhedral model representing one body shape. In short, we performed principal component analysis on the nine factors remaining after exclusion of constant angular defects (G1 is 808; G2, G3 and G12 are 08) in all body shapes.

Table II. A cross tabulation of 122 items of four JIS type and seven ages

Table III. A cross tabulation of 672 items, which are translated so that the total is 122

Age 10-19 Age 20-29 Age 30-39 Age 40-49 Age 50-59 Age 60-69 Age 70-79 Total

Y

A

AB

B

Total

1.0 5.5 4.2 5.1 3.5 3.2 2.8 25.2

5.1 14.6 8.8 11.6 9.9 7.3 4.3 61.7

2.3 5.4 4.4 8.1 5.1 4.1 1.8 31.4

0.0 0.4 0.0 2.2 0.4 0.8 0.0 3.8

8.4 25.9 17.4 27.0 18.9 15.4 8.9 122.0

Note: Each item is weighted by number of representing people

Age 10-19 Age 20-29 Age 30-39 Age 40-49 Age 50-59 Age 60-69 Age 70-79 Total

Y

A

AB

B

Total

1.8 5.3 4.6 5.1 5.2 3.4 1.7 27.0

4.8 11.1 9.1 11.1 8.6 7.3 4.4 56.5

2.8 5.4 5.1 7.5 5.1 4.0 2.6 32.6

0.4 0.6 1.0 1.5 1.1 0.8 0.5 5.9

9.8 22.4 19.8 25.2 20.0 15.6 9.1 122.0

Note: Each item is weighted by number of representing people

80.0 0.0

Note: 122 polyhedral models

Mean SD

Angular defects (degree)

A focus of neck-base line G1 11.8 1.4

36.9 2.7

8.2 1.0

Vertices of Vertices of busts front armpits 2 £ G4 2 £ G5 2 £ G6 227.8 4.8

Foci of armhole lines 2 £ G7 18.4 2.2

11.4 1.4

Vertices of Vertices back armpits of back 2 £ G8 2 £ G9 23.1 2.1

Vertices of back shoulders 2 £ G10

A focus of waistline G13 298.5 7.3

A vertex of centre back G11 3.8 0.5

Classification of body shape characteristics 305

Table IV. Weighted mean and weighted SD of angular defects

IJCST 22,4

306

Here, we do not execute fixed regularisation because each variable has a common unit of degrees. We rotated (1.18 clockwise) the scatter diagram of principal component marks in Figure 7 for easier interpretation. As a result, a linearised group is observed. After rotation, we present the analytic result of the principal component loading amount (covariance) for each principal component in Table V. Consequently, the significant factors showing the characteristics of body shape are the first and second principal components with coefficients of determination of 65.6 and 34.3 per cent, respectively. These two principal components define the characteristics of body shape. Interpretation of principal component analysis Body types based on angles through principal component analysis are labelled as belonging to quadrants 1-4. The horizontal and vertical axes represent principal components 1 and 2, respectively. The principal component 1 is the characteristics of the lower part which consists of G13, an angular defect of a focus around the waistline and double G8, angular defects of vertices around the back armpits. The principal component 2 is the characteristics of the upper part which consists of double G7, angular defects of foci around the armhole lines and double G5, angular defects of vertices around the busts in Table V. As shown in Table V, the principal component loading of a focus has a positive value and the principal component loading of a vertex has a negative value. We arranged the weighted mean of the angular defects of subjects belonging to the quadrants in Table VI. The value of angular defect is an index to indicate the degree of

Principal component 2

Scatter diagram 10

Figure 7. Scatter diagram of principal components

Table V. Relation between principal components and angular defects

5 0 –25

–20

–15

–10

–5

0

5

10

15

20

25

–5 –10 Principal component 1

Note: 1.1° clockwise rotation

Rotated principal components

Coefficient of determination (%)

Rotated principal component 1 Rotated principal component 2

65.6

Note: 1.18 clockwise rotation

34.3

Principle component loading (covariance) G13:2 £ G8 ¼ 258.9:17.6 ¼ lower part (a focus of waistline: vertices of back armpits) 2 £ G7:2 £ G5:2 £ G10 ¼ 2 27.1:15.8:12.0 ¼ upper part (foci of armhole lines: vertices of busts: vertices of back shoulders)

1 2 3 4

20.2a 16.4b 16.7b 20.1a 18.4

291.8a 304.4b 304.7b 293.2a 298.5

Angular defects (degree)

Mean Mean Mean Mean Mean

Notes: aDifficult to make a flat paper pattern; beasy to make a flat paper pattern

Quadrant Quadrant Quadrant Quadrant Total

Classification

Vertices of back armpits 2 £ G8

Lower part (principal component 1) A focus of waistline G13 222.0a 220.1a 229.7b 230.9b 227.8

Foci of armhole lines 2 £ G7 40.8a 40.9a 35.4b 35.5b 36.9

26.0a 26.1a 21.9b 22.0b 23.1

Upper part (principal component 2) Vertices of Vertices of back busts shoulders 2 £ G5 2 £ G10

Classification of body shape characteristics 307

Table VI. Weighted mean values of angular defects according to each quadrant

IJCST 22,4

308

difficulty to flatten a part. When the value of angular defect is almost 08 on the vertex, it is easy to make flat patterns. On the other hand, when the value of angular defect is almost 3608 on the focus, it is easy to make flat patterns around the focus. Therefore, when the vertices such as G5, G8 and G10 are smaller than the total mean of each, it is easy to make paper patterns around the parts. On the other hand, when the foci such as G7 and G13 are bigger than the total mean of each, it is easy to make paper patterns around the parts. Consequently, the horizontal axis could be interpreted as the difficulty of flattening the lower part, and the vertical axis could be interpreted as the difficulty of flattening the upper part. Relation to the size and classification by angles Figure 8(a) shows the weighted mean value of absolute sizes to clarify the shape characteristics of quadrants 1-4 by angles. Here, we consider the relative over absolute values because classification by angles has similarity invariance. After the bust girth is fixed at a Japanese standard size of 83 cm, the other sizes are converted proportionately in Figure 8(b). Subjects in quadrants 2 and 3 have relatively large waist-to-bust girth ratios. Similarly, we evaluated the ratio of bust to a constant back length in Figure 8(c). Subjects in quadrants 1 and 2 have relatively large bust girth. Accordingly, the pattern of quadrants 1-4 is divided as follows: . Quadrant 1. An hourglass type in which the bust size is large and the waist size is smaller compared with the bust size. . Quadrant 2. A cylindrical type in which the bust size is large and the waist size is larger compared with the bust size. . Quadrant 3. A cylindrical type in which the bust size is small and the waist size is larger compared with the bust size. . Quadrant 4. An hourglass type in which the bust size is small and the waist size is smaller compared with the bust size. Body shape characteristics by age group according to angles To clarify the body shape characteristics by age, we provide a scatter diagram in which weighted mean values of first and second principal components are shown in Figure 9. B 91.8 W 79.3 CtoPW 37.7 quadrant 2

B 91.6 W 74.8 CtoPW 37.7 quadrant 1

quadrant 3 80.7

quadrant 4 B 81.1

B

W CtoPW

67.4 37.9

W CtoPW (a)

Figure 8. Body shape characteristics according to each quadrant

64.1 37.5

B W

83.0 71.7

B W

83.0 67.8

CtoPW 34.1 CtoPW 34.2 quadrant 2 quadrant 1 B W CtoPW

quadrant 3 83.0 69.2 39.0

quadrant 4 B 83.0 W 65.7 CtoPW

(b)

38.4

B

91.8

B

91.6

W 79.3 W 74.8 CtoPW 37.7 CtoPW 37.7 quadrant 2 quadrant 1 B W

quadrant 3 80.4 67.1

CtoPW

quadrant 4 B 81.5 W 64.5

37.7

CtoPW

37.7

(c)

Notes: (a) Weighted mean values of Japanese women’s sizes by the new system of body classification (cm); (b) weighted mean values of Japanese women’s sizes proportionally converted into fixed 83 cm bust size (cm); (c) weighted mean values of Japanese women’s sizes proportionally converted into fixed 37.7 cm back length (cm); *Cervical-to-posterior Waist Len (CtoPW) (back length

Classification of body shape characteristics

Scatter diagram 4

Mean 60 ~ 69

Principal component 2

3 Mean 50 ~ 59 2 Mean 70 ~ 79 Mean 40 ~ 49 1

309

0 –10

–5

0

5

–1

10

Mean 30 ~ 39

–2

Mean 20 ~ 29

–3 –4 Principal component 1

Mean 10 ~ 19

Figure 9. Scatter diagram for weighted mean values of principal components according to age

Figure 10 shows a bar graph indicating the proportion of each body shape in each age group. In the scatter diagram, body shape has a tendency to change to quadrant 2 from quadrant 4 with age (Figure 9). This means that, in comparison with bust, body shape varies from an hourglass type of small bust and slimmer waist to a cylindrical type of large bust and thicker waist. Next, we examined the population ratio of each quadrant by age group (Figure 10). In young persons who are hourglass type with small busts (quadrant 4), the waist tends to be larger compared with the bust. This tendency significantly appears in their 30s; the resulting type is called cylindrical type with small busts (quadrant 3). Persons in their 40s tend to have a larger bust compared with the waist. This type is called hourglass type with large busts (quadrant 1). As women enter their 60s and 70s, their

Age 70-79

48

52 7

Age 60-69 Age 50-59

27

Age 40-49

17

Age 20-29

6

Age 10-19

5

0%

25

29

20

24

43

16

43

49

8

Age 30-39

7

40

45

85

10 95 20% Quadrant 1

40% Quadrant 2

60%

80%

Quadrant 3

Quadrant 4

100%

Figure 10. Bar graph of population by body type according to age (%)

IJCST 22,4

310

waist size tends to increase compared with their bust size. This type is called cylindrical type with large bust and waist (quadrant 2). Conclusion We prepared polyhedral models for 122 types of body shapes of Japanese adult females. By analysing angular defects G1-G13 of 122 polyhedral models, we recognised the following facts: (1) From the results of principal component analysis for a variation covariance matrix, we observed that the factors showing body shape characteristics were the first and second principal components with coefficients of determination of 65.6 and 34.3 per cent, respectively. The factor that best describes the first principal component is the ratio of waist girth and bust girth (an angular defect of the focus of the waistline), and that of the second principal component is the ratio of bust girth and back length (angular defects of the vertices of the bust). (2) The classifications of quadrants 1-4 by angles in this paper are as follows: . Quadrant 1. An hourglass type in which bust size is large and waist size is smaller compared with the bust size. It is a difficult zone to make a flat paper pattern around the waistline and the bust. An index to indicate the degree of difficulty to flatten a part around the waistline and the bust is d and d, respectively, (an index of “d” means “difficult”). . Quadrant 2. A cylindrical type in which bust size is large and waist size is larger compared with the bust size. An index to indicate the degree of difficulty to flatten a part around the waistline and the bust is e and d, respectively, (an index of “e” means “easy”). . Quadrant 3. A cylindrical type in which bust size is small and waist size is larger compared with the bust size. An index to indicate the degree of difficulty to flatten a part around the waistline and the bust is e and e, respectively. . Quadrant 4. An hourglass type in which bust size is small and waist size is smaller compared with the bust size. An index to indicate the degree of difficulty to flatten a part around the waistline and the bust is d and e, respectively. (3) We studied the variation and ratio of mean body shape by angle according to age group. As women grow older, a tendency to shift from quadrant 4 to quadrants 2 or 3 appears in the 40s and is completed in the 70s. In addition, quadrant 1 begins to appear in the 20s, continues to appear in a large proportion until the 50s and fades away in the 70s. References Alboul, L., Echeverria, G. and Rodrigues, M. (2005), “Discrete curvatures and Gauss maps for polyhedral surfaces”, European Workshop on Computational Geometry (EWCG ) 2005 in Eindhoven, The Netherlands, 9-11 March, pp. 69-72. Calladine, C.R. (1986), “Gaussian curvature and shell structures”, The Mathematics of Surfaces in a Conference Organized by the Institute of Mathematics and its Applications at the University of Manchester, Manchester, UK, 17-19 September 1984, pp. 179-96.

Cho, Y.S., Komatsu, T., Takatera, M., Inui, S., Shimizu, Y. and Park, H. (2006), “Posture and depth adjustable 3D body model for individual pattern making”, International Journal of Clothing Science & Technology, Vol. 18 No. 2, pp. 96-107. HQL (1997), Body Measurement Data of the Adult Women, Research Institute of Human Engineering for Quality Life, Japanese Standards Association, Tokyo. Imaoka, H. and Masuda, T. (1996), “An interpretation of cutting and sewing a cloth with respect to curvatures and a proposal of sewing equations”, Journal of the Japan Research Association for Textile End-Uses, Vol. 37 No. 8, pp. 422-9 (in Japanese). JISC (1997), Japanese Industrial Standard: JIS L 4005-1997 Sizing Systems for Women’s Garments, Japanese Industrial Standards Committee, Japanese Standards Association, Tokyo (in Japanese). Maltret, J.-L. and Daniel, M. (2002), Discrete Curvatures and Applications: A Survey, Rapport de recherche LSIS.RR.2002.002, Laboratoire des Sciences de l’Information et des Syste`mes, London. Masuda, T. and Imaoka, H. (2004), “Classification of tight-fitting bodice patterns of young women using concentrated Gaussian curvature”, Sen’i Gakkaishi, Vol. 60 No. 12, pp. 377-85. Masuda, T. and Imaoka, H. (2005), “Classification of curved shapes for tight-fitting bodice patterns of young men using concentrated Gaussian curvature”, Sen’i Gakkaishi, Vol. 61 No. 8, pp. 247-55. Miyoshi, M. (2004), “A theory of a new Bunka basic pattern and the change of basic patterns”, Journal of the Japan Research Association for Textile End-Uses, Vol. 45 No. 4, pp. 272-9 (in Japanese). Corresponding author Haruki Imaoka can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

Classification of body shape characteristics 311

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

Effects of temperature on liquid penetration performance of surgical gown fabrics Wei Cao Family and Consumer Science Department, California State University Northridge, Northridge, California, USA, and

Rinn Cloud

Surgical gown fabrics 319 Received 21 May 2009 Revised 12 January 2010 Accepted 12 January 2010

Baylor University, Waco, Texas, USA Abstract Purpose – Surgical gown fabrics are categorized for liquid penetration resistance by standard tests under specified laboratory conditions, which can be different from the conditions encountered in the surgical environment. The purpose of this paper is to examine the influence of temperature and challenge liquid (CL) type on the effectiveness of liquid penetration resistance of surgical gown fabrics. Design/methodology/approach – One disposable and one reusable surgical gown fabric were tested for liquid penetration using standard methods required in American Society for Testing Materials F2407 for classifying the materials used in Levels 1-3 surgical gowns. Standard test conditions were compared to varied conditions of ambient/fabric temperature (AFT), CL type and challenge liquid temperature (CLT). Analysis of variance was used to determine the effects of variables on liquid penetration. Findings – AFT, CL type and CLT were significant ( p , 0.05) variables for liquid penetration for at least one of the test fabrics. Higher ambient temperature, fabric and liquid temperature conditions resulted in greater liquid penetration. Use of synthetic blood as the CL resulted in higher liquid penetration than observed with distilled water. Research limitations/implications – Results suggest that temperatures within the range of body heat or ambient surgical environments are sufficient to affect liquid penetration of surgical gown fabrics. Also, the use of CLs other than distilled water and the use of CLs warmed to body temperature may be needed to accurately assess the liquid penetration resistance of surgical gown fabrics. Level of protection afforded by surgical gowns may be compromised by variability in these conditions. Originality/value – Conventional wisdom has held that differences between standard testing temperatures and body temperature or ambient temperature in the surgical theatre were insufficient to influence liquid penetration. This study has shown otherwise. No previous studies were found that addressed these variables but our study illustrates their effect on selected materials. Keywords Protective clothing, Liquids, Temperature, Heat Paper type Research paper

1. Introduction Surgical gowns were originated to protect patients from bacterial infections. In recent decades, their protective functions have extended to health care workers due to increased occupational exposure to blood-borne pathogens such as human immunodeficiency virus and hepatitis B and C viruses (Laufman et al., 2000). In response to this concern, the US Occupational Safety and Health Administration (OSHA, 1991), Garner (1986) and the Association of Operating Room Nurses (1992) have

International Journal of Clothing Science and Technology Vol. 22 No. 5, 2010 pp. 319-332 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011071794

IJCST 22,5

320

recommended the use of personal protective equipment (including clothing) to minimize occupational exposure to blood or other potentially infectious body materials. The final rule (1991) of the OSHA specifies that a protective gown should be capable of protecting the wearer “under normal conditions of use and for the duration of time for which it will be used” (p. 64177). A case study by Wong et al. (1998) demonstrated that wearing appropriate protective barriers can significantly improve protection for health care personnel. Because surgical procedures involve varying levels of blood exposure, surgical gown fabrics offering different levels of liquid protection have been made available by manufacturers. A recent standard, developed by the American Society for Testing Materials (ASTM), ASTM F2407-06, standard specification for surgical gowns intended for use in healthcare facilities, classifies a barrier material’s performance into Levels 1-4. Methods specified in this standard and those used in research studies to test and categorize surgical gown fabrics are conducted either under standard laboratory conditions (218C and 65 per cent relative humidity) or unspecified room temperature and humidity conditions, following standard lab procedures with regard to fabric and liquid temperature and liquid type. However, normal conditions of use for surgical gowns are likely to be different from standard laboratory conditions. For example, Brandt (1993) indicated that operating room temperatures range from 15.6 to 25.68C and Mangram et al. (1999) reported relative humidity in operating rooms from 30 to 60 per cent. Normal conditions of use also include interactions of the surgical gown fabric and the body of the wearer. During surgical procedures, body heat passes to the garment from the skin. As the skin temperature increases, the temperature of the garment’s fabric will increase as well (Fort and Hollies, 1970). Hagander et al. (2000) reported a skin temperature range of 27-378C. In still ambient conditions at 208C, the average skin temperature when naked is approximately 338C (Nicholson et al., 2000). Li et al. (2004) measured skin and clothing temperatures of clothed subjects moving from a typical indoor environment (258C, 40 per cent RH, 0.3 m/s) to a hot, humid environment (368C, 80 per cent RH, 0.1 m/s). The temperature of clothing in the trunk area of the body increased from 12 to 408C within two min and then dropped slightly within the next 20 min. When the subject returned to the cooler environment, clothing temperature in the trunk area decreased but remained at 298C after 40 min. It is expected, then, that surgical gown fabric temperature will be affected by body temperature. Standard tests of liquid penetration often use distilled water rather than blood as a screening tool for surgical gown fabrics. Regardless of what liquid is used, the temperature of the liquid is well below the temperature of blood likely to be encountered in the surgical theatre. Blood projecting from a patient may be close to body core temperature when it impacts surgical gown fabrics. Little has been reported regarding the effects of ambient temperature or body heat on the barrier performance of surgical gown fabrics. The current study was designed and conducted to investigate some of these effects. The objectives of the study were: (1) To evaluate the effects of ambient temperature (15.6, 21, 25.68C) on liquid impact penetration properties of fabrics conditioned to the same temperature as the ambient environment with liquid temperature controlled according to the standard requirement. (2) To determine the potential effects of body heat on liquid penetration of surgical gown fabrics by challenging un-warmed (conditioned at 218C) and warmed

(conditioned at 338C) fabrics with two types of liquids at required testing temperature (21 or 278C) and after warming liquids to body core temperature (378C) as measured by: . impact penetration testing (American Association of Textile Colorists and Chemists (AATCC 42); and . hydrostatic pressure testing (AATCC 127). 2. Methods An experimental laboratory study was designed to explore the potential for independent variables to have an effect on liquid penetration of surgical gown fabric. Ambient/fabric temperatures (AFTs), types of liquid and liquid temperature were the variables under consideration. Two fabrics currently in common use in surgical gowns, a disposable and a reusable material, were tested. Fabrics selected for the study were not intended to represent the full range of surgical gown fabrics. Likewise, liquids and temperatures selected for study do not represent the full range of possible liquids or temperatures. 2.1 Surgical gown fabrics One of each category of gown fabric, disposable and reusable, was included in the study. Roll goods of the selected fabrics were provided by two major manufacturers whose products make up a significant market share for Level 2 gown fabrication (Mclntyre, 2005). The thickness of fabrics were characterized following ASTM D1777 (standard method for measuring thickness of textile materials) using a Precisionw material thickness gauge (with a range of 0-25 mm). The weight of fabrics were measured by an Ohausw scale (Maximum 610 g, made in Sweden) calibrated using Nettler Toledow PB 3002-s following ASTM D3776 (standard method for mass per unit area of a woven fabric). Based on the average of 15 randomized specimens, the disposable fabric is heavier (85.6 g/m2) and thicker (0.231 mm) than the reusable fabric (52.8 g/m2 and 0.117 mm, respectively). Our intent was to test a commonly used fabric of each type, not to select fabrics that were similar in these characteristics. Previous research (Leonas and Jinkins, 1997; McCullough, 1993) has identified differences in the performance of nonwoven (disposable) and woven (reusable) surgical gown fabrics. Therefore, testing and analyses were conducted on the two fabrics to determine the separate effects, not to draw comparisons between the fabrics. 2.2 Challenge liquids and liquid characterization Two types of liquid, distilled water and synthetic blood were used as challenge liquids (CL). Distilled water is the standard CL required by the AATCC 42 (impact penetration) and 127 (hydrostatic pressure) tests. Synthetic blood is a useful test liquid in simulating body fluid with its surface tension at the lower end of the range of human fluids (0.0042-0.006 N/m) as ASTM F1670 (resistance of materials used in protective clothing to penetration by synthetic blood) defined. To compare the difference in penetration behaviors of body fluids and standard test liquid, synthetic blood and distilled water were used as CLs in the AATCC 42 test. The synthetic blood was prepared following the procedure specified in ASTM F1670. To document differences in physical properties of the two liquids, the surface tensions of the liquids were measured by a ThermoCahn Dynamic Contact Angle Analyzer.

Surgical gown fabrics 321

IJCST 22,5

322

Kinematic viscosity of the two liquids was measured by a Cannon-Fenske Routine Viscometer. The surface tension of the distilled water used in this study was 65.13 dynes/cm at standard testing temperature (218C); it decreased to 52.29 dynes/cm as the distilled water temperature increased to body core temperature (378C). Viscosity of distilled water was 1.13 mm2/s at 218C and decreased to 1.04 mm2/s when warmed to 378C. The surface tension for synthetic blood measured 42.00 dynes/cm at 218C. This value matches the expected value in the test that provides instructions for mixing it (ASTM F1670). The surface tension of synthetic blood decreased to 35.50 dynes/cm when its temperature increased to 378C. Viscosity of synthetic blood is considerably greater than that of distilled water (4.24 mm2/s at 218C) and also decreased to 3.01 mm2/s as the temperature of synthetic blood was increased to body core temperature (378C). 2.3 Manipulation of ambient/fabric/liquid temperature The ambient temperature adjustment was realized by use of a Heinicke Environmental Chamber. When the temperature of the chamber stabilized, test fabric specimens were placed in the chamber to condition for a minimum of 24 hours. Based on the objectives of the study, temperature settings were adjusted to 15.6, 21, 25.6 (covering the range of temperatures in the surgical environment as reported by Brandt in 1993) and 338C (consistent with average skin temperature as reported by Nicholson et al. (2000)) with ^ 18C control accuracy. The relative humidity was kept at 65 per cent for all temperature settings. A Davis Perception II thermometer was placed in the chamber to monitor the conditions and assure accuracy of the ambient temperature and humidity. Fabric temperature was measured by the YSI Precision 4,000 A thermometer with 903 40B thermistor probe. Liquid temperature settings were based on those required by the standard procedures (21 or 278C) and body core temperature (378C) as warmed in a Precision isothermal water bath located in the environmental chamber. 2.4 Liquid penetration testing ASTM F2407-06 uses two standardized testing procedures for liquid penetration to set criteria for Levels 1-3 surgical gown fabrics. In AATCC 42, liquid penetration is measured by impacting the surface of fabric with a fixed amount (500 ^ 10 ml) of 278C distilled water poured through a standard nozzle from a fixed height (0.6 m) (Figure 1). The equipment used was a Type I Impact Penetration Tester available from AATCC (Order No. 08733). The liquid penetration property of the materials is evaluated by the weight gain of blotting paper placed under the fabric specimen, which is clamped to a 458 metal plate. Weight of the blotter before and after penetration testing was accomplished with an Ohausw scale (Maximum 610 g, made in Sweden) calibrated using Nettler Toledow PB 3002-s. According to the standard, the blotter should not have gained more than 1 g to be considered a Level 2 fabric. Fifteen specimens measuring 178 mm £ 330 mm were tested for each combination of test conditions. AATCC 127 is designed to measure the barrier effectiveness of fabric by subjecting the face of the fabric to 218C distilled water under increasing hydrostatic pressure and observing visible penetration of water droplets on the back surface of the fabric (Figure 2). The equipment used was a TEXTEST FX 3000 Automatic Hydrostatic Head Tester (HYDROTESTER III). A 200 mm £ 200 mm sample of the fabric is clamped horizontally over a cylinder of water. Hydrostatic pressure is steadily increased until penetration of three droplets is observed. To be acceptable for

Surgical gown fabrics 323

Figure 1. Impact penetration tester Note: Type I

a Level 2 fabric, penetration of water droplets should not occur at less than 20 mbar. A total of 15 specimens were tested for each combination of test conditions. 2.5 Data analysis Analysis of variance (ANOVA) was used to determine the effect of each independent variable on the dependent variable (liquid penetration). A confidence level of 95 per cent ( p # 0.05) was set to reject the null hypothesis. The Tukey’s honestly significant difference (HSD) test was used as the post hoc test to determine where significant differences existed when the assumption of equal variances was not violated. The Kruskal-Wallis nonparametric test was used when the assumption of equal variance was violated in ANOVA. Statistical Package for the Social Sciences (SPSS Version 13.0) software was used for data analysis. 3. Results and discussion 3.1 Effects of ambient temperature The means and standard deviations for impact penetration of test fabrics using distilled water and synthetic blood at the three temperatures are shown in Table I.

IJCST 22,5

324

Figure 2. Hydrostatic head tester

Fabrics Disposable Reusable Table I. Liquid impact penetration of fabrics by AFT and CL type

AFT (8C) 15.6 21.0 25.6 15.6 21.0 25.6

DW (278C) M (n ¼ 15) 0.181 0.197 0.225 0.499 0.495 0.493

SD 0.044 0.040 0.123 0.087 0.119 0.120

SB (278C) M (n ¼ 15) 0.669 0.680 0.899 1.651 1.377 1.943

SD 0.185 0.211 0.295 0.511 0.659 1.315

Notes: Results are expressed in grams of liquid that penetrated the fabric; lower numbers indicate less liquid penetration; DW, distilled water; SB, synthetic blood

The three temperatures represent the range of temperatures in surgical environments and the standard environmental temperature for textile testing. The mean distilled water penetration for the two fabrics ranged from 0.181 to 0.499 g while the mean synthetic blood penetration for the fabrics was higher, ranging from 0.669 to 1.943 g.

The ANOVA considered the main effects of AFT and CL type and also the interaction of these two variables. Results of the analysis are given in Table II. The adjusted R 2 values (0.832 and 0.552) indicate that variance in the dependent variable (as measured by weight gain of the blotting paper) was highly explained by the model. Interactions were not significant, therefore allowing for consideration of the main effects. AFT was a significant main effect for disposable fabric but not for the reusable fabric. Based on the Tukey’s HSD test results, liquid penetration of the disposable fabric was significantly higher at 25.68C than at the two lower temperatures (15.6 and 218C). Ambient temperatures corresponding to the high end of the range (25.68C) reported by Brandt (1993) could influence increased penetration of some disposable surgical gown fabrics. CL type was a significant main effect for both disposable and reusable fabric. The overall mean liquid impact penetration of fabrics challenged by synthetic blood (1.657 g) was significantly higher than that of fabrics challenged by distilled water (0.495 g). Significant interaction effects were not found. Distilled water has higher surface tension and lower viscosity than synthetic blood. Higher surface tension is expected to result in reduced wettability and, therefore, less penetration. On the other hand, lower viscosity implies that the liquid should penetrate the fabric pores more easily. The result of higher penetration by synthetic blood than by distilled water, as observed in this study, suggested that the surface tension of the liquid may play a more important role than the viscosity in determining the penetration. In terms of practical significance, a Level 2 gown is expected to allow no more than 1 g of liquid to penetrate the fabric in the impact penetration test. For the reusable fabric at each of the ambient temperatures, the penetration of synthetic blood exceeded 1 g.

Surgical gown fabrics 325

3.2 Effects of body heat Previous literature indicated that the temperature of the garment will change with a skin temperature shift (Fort and Hollies, 1970; Li et al., 2004). Therefore, it is expected that surgical gown fabric temperature will be affected by the skin temperature during wearing. In addition, the temperature of body fluids projecting from patients may be close to the body core temperature, which is higher than the liquid temperatures specified in the test procedures. To explore the potential effects of body heat on liquid penetration performance of gown materials, we varied these conditions. Fabrics conditioned in 338C ambient environment (average skin temperature) and liquids Fabrics

Source

df

F

h

p

Disposable

AFT CL type AFT £ CL interaction Error AFT CL type AFT £ CL Interaction Error

2 1 2 84 2 1 2 84

4.29 435.95 1.09

0.30 0.92 0.17

0.02 * 0.00 * * * 0.34

1.48 109.16 1.23

0.17 0.75 0.17

0.23 0.00 * * * 0.30

Reusable

2

2

Notes: Significance at *p , 0.05, * * *p , 0.001; R ¼ 0.842 (adjusted R ¼ 0.832) for disposable fabric; R 2 ¼ 0.577 (adjusted R 2 ¼ 0.552) for reusable fabric

Table II. ANOVA for liquid impact penetration of fabrics by AFT and CL type

IJCST 22,5

326

warmed to 378C (approximate core body temperature) were compared with fabrics conditioned at 218C and liquids at standard testing temperature (278C for the impact penetration test, 218C for the hydrostatic pressure test). 3.2.1 Impact penetration results. Table III provides the means and standard deviations for impact penetration of test fabrics using warmed (378C) and standard-required (278C) distilled water and synthetic blood at the two AFTs representing the standard testing temperature and average skin temperature. The mean distilled water penetration for the two fabrics ranged from 0.197 to 0.840 g while the mean synthetic blood penetration for the fabrics was higher, ranging from 0.680 to 3.719 g. For both liquids, the lowest penetration occurred in the case of un-warmed liquids challenging unwarmed (218C) fabric. ANOVA was conducted separately on the disposable and reusable material and identified significant main effects for AFT, CL type, challenge liquid temperature (CLT) and the interaction among these three variables (Table IV). The three-way interaction among AFT, CLT and CL type was significant for disposable fabric. However, no two-way interactions were significant, suggesting that further analysis of the interaction is unnecessary. For reusable fabric, the significant interactions were further examined. 3.2.1.1 Main effects for disposable fabric. When disposable fabric temperature increased from 21 to 338C, the overall mean liquid impact penetration (pooled across liquid type and temperature) increased from 0.712 to 0.980 g. CL and CLT were also significant variables for the disposable fabric. When synthetic blood was used instead of distilled water, the overall mean penetration (pooled across fabric temperature and liquid temperature) increased to 1.259 g. When the CLs were warmed to body core temperature (378C), the overall mean penetration of the disposable fabric (pooled across fabric temperature and liquid type) increased from 0.580 to 1.112 g. The results indicate Fabrics Disposable

AFT (8C)

CL

CLT (8C)

M (n ¼ 15)

SD

21

DW

27 37 27 37 27 37 27 37 27 37 27 37 27 37 27 37

0.197 0.439 0.680 1.531 0.257 0.840 1.185 1.637 0.495 0.612 1.377 3.016 0.527 0.526 2.067 3.719

0.040 0.258 0.211 0.461 0.111 0.566 0.588 0.513 0.119 0.246 0.659 1.120 0.119 0.116 1.063 1.162

SB 33

DW SB

Reusable

21

DW SB

33

DW SB

Table III. Liquid impact penetration of fabrics by AFT, CL type and CLT

Notes: Results are expressed in grams of liquid that penetrated the fabric; lower numbers indicate less liquid penetration; DW, distilled water; SB, synthetic blood; AFT, ambient/fabric temperature; CL, challenge liquid type; CLT, challenge liquid temperature

Fabrics

Source

df

F

h

p

Disposable

AFT CL CLT AFT £ CL Interaction AFT £ CLT Interaction CL £ CLT Interaction AFT £ CL £ CLT Interaction Error AFT CL CLT AFT £ CL Interaction AFT £ CLT Interaction CL £ CLT Interaction AFT £ CL £ CLT Interaction Error

1 1 1 1 1 1 1 112 1 1 1 1 1 1 1 112

17.86 237.39 85.65 0.71 0.02 2.47 6.23

0.37 0.82 0.66 0.10 0.00 0.14 0.22

0.00 * * * 0.00 * * * 0.00 * * * 0.40 0.89 0.12 0.01 * *

4.88 309.14 29.44 6.09 2.22 17.74 0.11

0.20 0.85 0.46 0.22 0.14 0.37 0.00

0.03 * 0.00 * * * 0.00 * * * 0.02 * 0.14 0.00 * * * 0.74

Reusable

Notes: Significance at *p , 0.05, * *p , 0.01, * * *p , 0.001; R 2 ¼ 0.758 (adjusted R 2 ¼ 0.743) for disposable fabric; R 2 ¼ 0.648 (adjusted R 2 ¼ 0.626) for reusable fabric; DW, distilled water; SB, synthetic blood; AFT, ambient/fabric temperature; CL, challenge liquid type; CLT, challenge liquid temperature

Surgical gown fabrics 327

Table IV. ANOVA for liquid impact penetration of fabrics by AFT, CL type and CLT

that warmed disposable fabrics challenged with warmed liquids will allow greater liquid penetration than their unwarmed (standard temperature) counterparts. 3.2.1.2 Interaction effects of fabric temperature and CL type for reusable fabric. Figure 3 provides a graphed depiction of the interaction of AFT and CL type for the reusable fabric. The effect of AFT differed by type of liquid with the effect being evident only for synthetic blood. When the reusable fabric temperature increased from 21 to 338C, the mean synthetic blood penetration increased but the mean penetration of distilled water remained unchanged. These results suggest that the effects of AFT may be more detrimental when surgical gowns are impacted by blood than by water. Such differences are not accounted for in current testing procedures.

Mean impact penetration (g)

3.5 3

Synthetic blood Distilled water

2.5 2 1.5 1 0.5 0

21

33 AFT (°C)

Figure 3. Mean impact penetration of reusable fabric by the interaction of AFT and CL type

IJCST 22,5

328

3.2.1.3 Interaction effects of CL type and CLT for reusable fabric. Figure 4 provides a graph of the interaction of CL type and CLT for reusable fabric. The figure indicates that the effect of CLT on liquid impact penetration of reusable fabric also differed by type of liquid. As liquid temperature increased from 27 to 378C, mean liquid impact penetration of synthetic blood increased dramatically. However, mean penetration of distilled water increased only by a small amount. Viscosity tests for the two CLs indicated that synthetic blood experienced more viscosity reduction than distilled water when the temperature of the liquid increased from standard testing temperatures (21 or 278C) to the body core temperature (378C). The data also demonstrate that, regardless of liquid temperature, more penetration occurred for synthetic blood than for distilled water. AFT had a significant effect on the mean impact penetration for both categories of fabrics. Mean liquid impact penetration increased when the fabric was warmed to average skin temperature. When testing of unwarmed and warmed fabric was done with distilled water at the liquid temperature required in the standard, the differences were not enough to impact protection classification. However, when combined with changes in liquid type and liquid temperature, the penetration results exceeded the criterion of maximum penetration for a Level 2 rating for both fabrics. The normal conditions of use for surgical gown fabrics comprise a complex set of variables that should be considered both independently and collectively to obtain a more complete understanding of the factors affecting protective garment performance. 3.2.2 Hydrostatic pressure results. The second test of liquid penetration used in the classification of surgical gown fabrics measures the resistance of fabrics to penetration by liquid under rising hydrostatic pressure. Higher pressure measurements indicate better resistance to penetration (lower penetration). Table V shows the means and standard deviations for hydrostatic pressure required to induce liquid penetration of surgical gown fabrics by challenging unwarmed (conditioned at 218C) and warmed (conditioned at 338C) fabrics with distilled water at the required testing temperature (218C) and after warming liquids to body core temperature (378C). The mean hydrostatic pressure required to induce liquid penetration of surgical gown fabrics by distilled water at two temperatures of fabric and liquid ranged from

Figure 4. Mean impact penetration of reusable fabric by the interaction of CL type and CLT

Mean impact penetration (g)

4 3.5 3

Synthetic bood Distilled water

2.5 2 1.5 1 0.5 0 27

37 CLT (°C)

Fabrics Disposable

Reusable

AFT (8C)

CL

CLT (8C)

M (n ¼ 15)

SD

21

DW

33

DW

21

DW

33

DW

21 37 21 37 21 37 21 37

33.467 29.867 31.633 29.233 32.367 27.300 32.333 27.133

2.812 1.620 1.420 1.678 1.316 1.953 1.839 1.356

Notes: Results are expressed in mbar of pressure required to induce liquid penetration of fabric; a higher number indicates greater resistance to liquid penetration (lower penetration); AFT, ambient/ fabric temperature; CL, challenge liquid type; CLT, challenge liquid temperature; DW, distilled water

Surgical gown fabrics 329 Table V. Hydrostatic pressure required to induce liquid penetration of fabrics by AFT and CLT

27.133 to 33.467 mbar. Owing to equipment constraints, synthetic blood could not be used as a CL in this test. The ANOVA results are given in Table VI. Ambient fabric temperature was a significant main effect for the disposable fabric but not for the reusable fabric. CLT was a significant main effect for both disposable and reusable fabrics. The interactions were not significant for either fabric. 3.2.2.1 Main effect of AFT. For the disposable fabric, there was a significant increase in liquid penetration (decrease in hydrostatic pressure) of fabrics related to increasing AFT (Figure 5). The overall mean hydrostatic pressure required to induce penetration of water drops decreased from 31.667 to 30.433 mbar when fabric was warmed to 338C, reflecting lower resistance to liquid penetration. These results provided further demonstration of the potential effects of body heat to warm fabrics and subsequently increase liquid penetration. 3.2.2.2 Main effect of CLT. Increasing the temperature of the CL (distilled water) to body core temperature resulted in a significant difference in overall mean liquid penetration of both fabrics (Figure 6). For the disposable fabric, the overall mean hydrostatic pressure required to induce penetration decreased from 32.550 to 29.550 mbar when the fabric was challenged by 378C distilled water. For the reusable fabric, the overall mean hydrostatic pressure was reduced even more, from 32.350 to 27.217 mbar. Fabrics

Source

df

F

h

p

Disposable

AFT CLT AFT £ CLT Interaction Error AFT CLT AFT £ CLT Interaction Error

1 1 1 56 1 1 1 56

5.492 35.220 1.005

0.30 0.61 0.14

0.02 * 0.00 * * * 0.32

0.054 144.638 0.012

0.00 0.85 0.00

0.82 0.00 * * * 0.91

Reusable

Note: Significance at *p , 0.05, * * *p , 0.001; R 2 ¼ 0.721 (adjusted R 2 ¼ 0.706) for disposable fabric; R 2 ¼ 0.427 (adjusted R 2 ¼ 0.396) for reusable fabric; AFT, ambient/fabric temperature; CLT, challenge liquid temperature

Table VI. ANOVA for hydrostatic pressure required to induce liquid penetration of fabrics by AFT and CLT

IJCST 22,5

Figure 5. Hydrostatic pressure required for liquid penetration by AFT for disposable fabric

31.6 Hydrostatic pressure (mbar)

330

31.8

31.4 31.2 31 30.8 30.6 30.4 30.2 30 21

33 AFT (°C)

33 21°C

Figure 6. Hydrostatic pressure required for liquid penetration by CLT

Hydrostatic pressure (mbar)

32

37°C

31 30 29 28 27 26 25 24 Disposable

Reusable

Less pressure was required to induce liquid penetration of surgical gown fabrics when warmed distilled water was used to challenge the fabric than when the standard testing procedure was followed. These results suggest that challenges by fluids at body temperature are more likely to result in penetration than distilled water at standard testing temperature. 4. Conclusions, limitations and implications Our intent was to investigate whether temperature changes in the range of the ambient surgical environment and body temperature would result in differences in penetration resistance of fabrics. Results of the current study indicated that changes in ambient temperature over the reported operating room temperature range are sufficient to result in a significant increase in liquid penetration of at least one type of disposable fabric currently used in surgical gowns. Use of distilled water in laboratory penetration testing of surgical gown materials results in significantly lower penetration than

results obtained with synthetic blood for the two fabrics tested. Fabric that is warmed to the body skin temperature is penetrated by liquids more easily than fabric that is conditioned and tested at standard laboratory conditions. Warm liquids will penetrate some surgical gown fabrics in higher quantity than is realized by standard laboratory tests using distilled water or even synthetic blood at cooler temperatures. Owing to the difficulties of warming fabrics separately from the ambient temperature, these variables were co-varied. The effects may be somewhat different if a method can be established to warm fabrics separately from the ambient environment. The findings cannot be generalized to all surgical gown fabrics due to the limited selection of fabrics for this study. Findings from this study may assist the development of new standards for evaluating performance of surgical gown materials and the research/development of innovative gown fabrics capable of providing protection in conditions likely to be encountered in the operating room. The findings from this research should be considered in evaluating the protective performance of surgical gown fabrics as well as of other types of protective clothing. This study followed procedures described in ASTM F2407-06, standard specification for surgical gowns intended for use in healthcare facilities, to test liquid penetration of surgical gown fabrics. This study did not take into account the role of seams or edges of fabric, nor did we consider the effects of body movement or fabric curvature. These additional factors should be considered in future work.

References Association of Operating Room Nurses (1992), “Recommended practices: protective barrier materials for surgical gowns and drapes”, AORN Journal, Vol. 55, pp. 832-5. Brandt, B. (1993), “Surgical gowns: a survey of wearer and purchaser satisfaction with protection”, Nonwovens Industry, Vol. 9, pp. 114-9. Fort, L.E. and Hollies, N.R.S. (1970), Clothing: Comfort and Function, Marcel Dekker, New York, NY. Garner, J.S. (1986), “CDC guidelines for prevention of surgical wound infections”, American Journal of Infection Control, Vol. 14, pp. 71-80. Hagander, L., Midani, H., Kuskowski, M. and Parry, G. (2000), “Quantitative sensory testing: effecting of site and pressure on vibration thresholds”, Journal of the Peripheral Nervous System, Vol. 5, pp. 251-2. Laufman, H., Belkin, N.L. and Meyer, K.K. (2000), “A critical review of a century’s progress in surgical apparel – how far have we come?”, Journal of American College of Surgeons, Vol. 191, pp. 554-68. Leonas, K.K. and Jinkins, R.S. (1997), “The relationship of selected fabric characteristics and the barrier effectiveness of surgical gown fabrics”, American Journal of Infection Control, Vol. 25, pp. 16-23. Li, Y., Li, F., Liu, Y. and Luo, Z. (2004), “An integrated model for simulating interactive thermal processes in human-clothing system”, Journal of Thermal Biology, Vol. 9, pp. 567-75. McCullough, E.A. (1993), “Methods for determining the barrier efficacy of surgical gowns”, American Journal of Infection Control, Vol. 21, pp. 368-74. Mclntyre, K.B. (2005), “Spunlace: a market on the move: manufacturers scramble to take advantage of wipes market growth, but can it last?”, Nonwovens Industry, Vol. 36, pp. 60-7.

Surgical gown fabrics 331

IJCST 22,5

332

Mangram, A.J., Horan, T.C., Pearson, M.L., Silver, L.C. and Jarvis, W.R. (1999), “Guideline for prevention of surgical site infection”, American Journal of Infection Control, Vol. 27, pp. 97-132. Nicholson, G.P., Scales, J.T., Clark, R.P. and de Calcina-Goff, M.L. (2000), “A method for determining the heat transfer and water vapor permeability of patient support systems”, Medical Engineering & Physics, Vol. 22, pp. 155-6. OSHA (1991), “Occupational exposure to bloodborne pathogens: final rule”, Federal Register, Vol. 56, pp. 64004-182. Wong, D.A., Alexander, J.A. and Kay, L. (1998), “Risk of blood contamination of health care workers in spine surgery: a study of 324 cases”, Spine, Vol. 23, pp. 1261-6. Further reading AATCC (2004), AATCC Technical Manual, American Association of Textile Chemists and Colorists, Research Triangle Park, NC. ASTM (2006), Annual Book of ASTM Standards, 11.03, American Society for Testing and Materials, West Conshohocken, PA. US Department of Health and Human Services, Centers for Disease Control and Prevention (2004), HIV/AIDS Statistics and Surveillance, Division of HIV/AIDS, National Center for Infectious Diseases, Atlanta, GA. Corresponding author Wei Cao can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

Thermo-physiological comfort of a PES fabric with incorporated activated carbon Part I: preliminary physical analysis R. Splendore Associazione Tessile e Salute, Biella, Italy

Thermophysiological comfort 333 Received 20 November 2009 Accepted 28 April 2010

F. Dotti Dipartimento di Scienze dei Materiali e Ingegneria Chimica, Politecnico di Torino, Torino, Italy

B. Cravello Associazione Tessile e Salute, Biella, Italy, and

A. Ferri Dipartimento di Scienze dei Materiali e Ingegneria Chimica, Politecnico di Torino, Torino, Italy Abstract Purpose – The purpose of this paper is to evaluate the thermo-physiological comfort of a knitted polyester (PES) fabric which contains activated carbon particles in the back-side. Design/methodology/approach – According to the manufacturer’s intention, the activated carbon particles, added in the PES extrusion process, give permanent attributes to the garment, such as odour resistance, UV protection and evaporative cooling. These features should make the modified PES ideal for sportswear. Standard fabric characteristics (morphology, mass per unit area, thickness) have been evaluated for two similar fabrics, the one containing the modified PES yarn and the other one made of conventional PES yarn. The investigated thermo-physiological properties were air permeability (AP), water vapour resistance (Ret ), thermal resistance (Rct ), thermal conductivity and diffusion, drying rate, vertical wicking, horizontal liquid diffusion area and buffering capacity. They have been measured in controlled thermal and humidity conditions in a climatic chamber. Findings – The modified fabric is more hydrophilic than the conventional one, thanks to the carbon particles sorption ability. Thus, the liquid management of the modified PES fabric was found to be better. On the other hand, liquid desorption was slow and the drying time was longer. Moreover, the dry heat and the vapour transfer were found slightly worse for the modified PES, probably due to the lower AP. Originality/value – The paper shows a comprehensive fabric characterization of a functionalized fabric, highlighting the positive and negative effects of activated carbon particles on the liquid, vapour and heat management. Keywords Thermal testing, Fabric testing, Carbon, Clothing Paper type Research paper

The authors gratefully acknowledge the Piemonte Regional Government that financed this work within the HITEX project (D.G.R n. 227-4715 del 27 November 2006). Special thanks to the National Research Council of Biella (CNR-ISMAC) for the technical support in performing the fabric characterization.

International Journal of Clothing Science and Technology Vol. 22 No. 5, 2010 pp. 333-342 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011071802

IJCST 22,5

334

1. Introduction Human comfort is defined as freedom from pain, a state of physiological, psychological and physical harmony between a human being and the environment (Slater, 1985). Comfort is not a textile property: it is a human feeling, a condition of ease or well-being which is influenced by many factors including textile properties (Collier and Epps, 1999). Comfort of clothing can be classified into three different categories: thermo-physiological comfort, tactile sensation and pressure sensation (Li, 2001). Thermo-physiological comfort is related to the way clothing buffers and dissipates metabolic heat and moisture (Verdu et al., 2009), which depends on many textile properties, such as thermal resistance, water vapour permeability, thickness, fabric structure, finishing, etc. During normal wear, insensible perspiration is continuously generated by the body. Heat and moisture vapour must be dissipated to maintain thermoregulation. In transient wear condition, characterized by intermittent pulse of activity, or in hard climatic conditions, sensible perspiration and liquid sweat must be rapidly managed by the clothing in order to maintain comfortable conditions of wear (Umbach, 1993). In the design of a item of clothing, especially sportswear, the ability of the material to transport moisture and to dry quickly is crucial to give the wearer the perception of thermal comfort (Barker, 2008). In literature, the evaluation of thermal comfort can be performed via three different methods: (1) by means of fabric physical analysis (Verdu et al., 2009); (2) by using a heated and sweating manikin (Celcar et al., 2008); and (3) by testing the item of clothing on a panel of people (Martı´nez et al., 2009). The first one is the most used and well-established method: textiles properties such as water vapour resistance, thermal resistance, wicking, buffering capacity give an idea of the fabric liquid, vapour and heat management. Testing laboratories carry out these analysis according to the specific standards (ISO11092, 1993); it is a cheap and fast method but it only gives information on the textile material features. The use of a manikin adds on information about the effect of clothing design. For example, the pumping effect, that is the effect of the air flow between skin and item of clothing can be taken into account by using a moving manikin. However, this is a more expensive technique compared to the physical analysis because the manikins are relatively rare. Moreover, the results are not of general validity because the manikin design varies considerably from lab to lab (Holme´r, 2004). The wear trial method is the most time consuming because a proper number of tests must be carried out for obtaining reliable results due to large physiological variability. On the other hand, this method considers the psychological component of comfort: thermal and humidity perceptions are usually assessed via a subjective scale (from warm to cold, from dry to wet) and a global comfort judgment is given by the wearer (Martı´nez et al., 2009). Preliminary tests on a knitted polyester (PES) fabric which contains activated carbon particles are shown in this paper. A comparative evaluation with a conventional PES fabric has been carried out. Both fabrics are intended for sports use and they are made of two interlaced layers with a different structure. The back-side of one fabric contains activated carbon particles trapped within the PES fibres. According to the manufacturer’s statement, the main benefit of activated carbon is the enormous surface area created by

the activated carbon pores and this feature should accelerate evaporation. Thus, evaporative cooling should be enhanced in the modified PES fabric. Moreover, when a hydrophobic textile such as PES is worn with high sweat rate, part of the sweat drips down the sweating area and is no longer available to contribute to wet heat loss (Weder et al., 2008). Thus, the increase of PES sorption ability is another positive effect obtained through activated carbon incorporation. Thermal and water vapour resistance, thermal conductivity and diffusion has been measured as well as liquid management properties related to the material sorption ability and hydrophilicity, such as drying rate, vertical wicking and horizontal liquid diffusion area (HLDA).

Thermophysiological comfort 335

2. Fabric characterization PES fabrics were characterized according to the following tests: . optical microscopy; . optical-based fibre diameter analyser (OFDA); . determination of mass per unit area (EN12127: 1998); . determination of thickness (ISO5084:1996); and . air permeability (AP) (ISO9237:1995) Fabric A is the conventional PES fabric and fabric B is the modified one. Figure 1 shows the optical microscopy image of a fabric B back-side fibre. Activated carbon appears as black particles in relief on the polymer surface. The OFDA analysis showed no difference in yarn diameter between fabric A and B front-side: the mean fibre diameter is 11.07 m (Sd ¼ 2.03 m) for fabric A and 11.52 m(Sd ¼ 2.46 m) for fabric B. Fabric B back-side yarns have a larger mean diameter of 15.10 m (Sd ¼ 2.21 m). Mass per unit area, fabric thickness and AP are reported in Table I. In the determination of mass per unit area, a 10 £ 10 cm sample was weighted after a 24-hour

Figure 1. Optical microscopy image: carbon particles in fabric B back-side

IJCST 22,5

336

conditioning at 208C and 65 per cent relative humidity (RH). Fabric thickness was measured using the fast system after one night conditioning under a 20 gf cm2 2 load, according to ISO5084:1996. Thickness results were confirmed by Alambeta analysis (data not shown). AP was measured, after one night conditioning, by FX 3300 (TEXTEST Instruments, Switzerland) according to the UNI EN ISO 9237. Fabric B was found thicker than fabric A. Moreover, AP of fabric B was smaller than that of fabric A. Mass per unit area of fabric B is about 7 per cent larger than that of fabric A while its thickness is 15 per cent larger than fabric A thickness. This suggests that fabric B is more porous and more air is trapped into the fabric structure. 3. Thermo-physiological properties 3.1 Water vapour resistance and thermal resistance Measurement of fabric response to water vapour loads and heat transfer were measured via the skin model prototype Permetest (Sensora Instruments, Czech Republic). This instrument measures dry thermal resistance (Rct), water vapour resistance (Ret) and the percentage water vapour permeability (%WVP) with respect to the empty plate. The measurements were performed in a climatic chamber set at 358C and 65 per cent RH according to ISO11092. 3.2 Thermal conductivity and diffusion Thermal conductivity and diffusion were measured by the Alambeta system (Sensora Instruments, Czech Republic). This instrument was used for a fast measurement of the transient and steady state thermo-physical properties of textile fabrics. By means of the Alambeta device, besides the classical stationary fabrics’ thermal properties such as thermal resistance (r) and thermal conductivity (l), we can also assess transient thermal characteristics such as thermal diffusion (a) (Hes, 2002). Thermal conductivity and diffusion were chosen as index of heat flow trough a textile fabric. 3.3 Vertical wicking Vertical wicking was measured by the tensiometer K100 (Kruss, Germany). To perform the test a fabric sample (1.5 £ 7 cm) is hung to the tensiometer balance. Then, the sample cross-section is put in contact with the water surface. The sample weight increase is due to capillary rise of liquid. This is an index of fabric wettability. 3.4 Horizontal liquid diffusion area The following technique was used for the evaluation of fabric sorption ability: 1 ml of liquid water was poured on the fabric sample and the diameter of the wet area was measured after a fixed time. A large area implies a large mass-transfer area for the drying process: the larger the area the higher the drying rate and the more intense the cooling effect. On the other hand, a small area implies a large sorption capacity per unit area. Both vertical wicking and HLDA are indices of fabric hydrophilicity.

Table I. Mass per unit area, fabric thickness and AP

Fabric A Fabric B

Mass per unit area (g m2 2)

Thickness (mm)

AP (mm s2 1)

148.54 159.50

0.646 0.742

1,325 973

3.5 Drying rate The drying rate was measured in a climatic chamber set at 358C and 50 per cent RH. A prototype of temperature and humidity wearable sensor developed within the Hitex project (Corbellini et al., 2008), measured the temperature and humidity of a fabric sample after being wetted with 1 ml of water. The temperature and humidity profiles had been recorded for 2 h. 3.6 Buffering capacity A dried sample was hung vertically in a sealed climatic chamber at 238C and 30 per cent RH. After a residence time of 90 min, the sample was weighted. The same process was repeated changing the climatic RH (238C and 50 per cent RH and 70 per cent RH). The water gain was reported as a percentage over the dried weight of the sample. Water gain is plotted against RH and the slope of the line is the buffering capacity (Verdu et al., 2009).

Thermophysiological comfort 337

4. Results and discussion The water vapour resistance, relative water vapour permeability and the thermal resistance of the two fabrics are shown in Table II. Each measurement was repeated five times. The conventional PES fabric has a lower water vapour resistance and a lower thermal resistance. Thermal resistance results were also confirmed by the Alambeta analysis (data not shown). Thus, the PES fabric has a better heat and vapour management than the modified PES one. This result is probably due to fabric A thinner structure and larger AP. Other thermal properties like thermal conductivity and thermal diffusion are shown in Table III. Each measurement was repeated three times. Thermal conductivity shows similar results for fabric A and B indicating a similar conductive heat flow through the fabrics. Thermal diffusion was higher for fabric B. In heat transfer analysis, thermal diffusion (a) is the ratio of thermal conductivity (l) to volumetric heat capacity (r cp ): a ¼ l=r cp For fabrics with similar thermal conductivity, thermal diffusion depend on the product of density (r) and specific heat capacity (cp ), that is the ability of a given volume of a substance to store internal energy while undergoing a given temperature gradient.

Fabric A Fabric B

Ret (Pa m2 W2 1)

%WVP (%)

Rct £ 102 3 (K m2 W2 1)

1.8 ^ 0.04 2.2 ^ 0.04

73.9 70.8

11.63 ^ 0.5 15.07 ^ 2.6

Notes: Ret, absolute water vapor resistance; %WVP, relative water vapour permeability; Rct, thermal resistance

Fabric A Fabric B

Thermal conductivity (l) (W/m *K)

Thermal diffusion (a) (m2/s)

0.039 ^ 0.002 0.040 ^ 0.002

0.129 ^ 0.007 0.166 ^ 0.011

Table II. Permetest results

Table III. Alambeta results

IJCST 22,5

338

Thus, fabric B rapidly adjusts its temperature to that of the surroundings due to the lower amount of energy stored. The result of the vertical wicking test is shown in Figure 2: the mass increase is plotted versus time. The fabric B has a higher wettability due to the carbon particle inclusion and more liquid is retained by the fabric. In the horizontal liquid diffusion test the wet area diameter measures 8.4 cm for fabric A, while it measures 5.7 cm for fabric B. This behaviour is probably ascribed to the carbon particles and the larger thickness of fabric B. This means that the same amount of liquid is concentrated in a smaller area for fabric B. This result can have a double effect as far as thermal comfort is concerned: on the one hand, the carbon modified PES fabric has the ability to remove more sweat per unit area, thus giving a dry sensation during the initial phase of physical activity; on the other hand, under the same amount of sweat to remove, a smaller wet area makes the evaporation rate slower. The mass-transfer area for the evaporative process is reduced rather than increased in contrast with the manufacturer’s statement. Thus, fabric drying should be prolonged and, as a consequence, evaporative cooling should be less intense for fabric B. To support this hypothesis the drying rate was analysed. The humidity and the surface temperature of two fabric samples wetted with the same amount of water (1 ml) were measured. The test was carried out in a climatic chamber at controlled conditions (358C, 50 per cent RH). The temperature profiles are shown in Figure 3. The fabric surface temperature decreases due to water evaporation. The slope of the descending curve changes with time: water evaporates at high rate removing great amount of latent heat at the beginning. As the amount of water in the fabric decreases, the evaporation rate slows down because water is supplied to the fabric surface at a smaller rate. When liquid water is almost finished the fabric temperature starts increasing again and it returns to the initial temperature when no more liquid water is present in the fabric. As shown in Figure 3 the evaporation process is slower for carbon modified PES fabric since the curve reaches its minimum after 60 min, about ten min later than the conventional PES one. This is the combined effect of the smaller mass-transfer area and the strong intermolecular interactions between activated carbon and water molecules that oppose water desorption. The fabric surface humidity was also measured during the drying process as shown in Figure 4.

0.20

Mass (g)

0.15

0.10

0.05

Figure 2. Vertical wicking of PES (fabric A) and modified PES (fabric B)

Fabric A Fabric B 0.00 0

10

20

30

40

50 Time (s)

60

70

80

90

100

Thermophysiological comfort

34.00

Temperature (°C)

33.00 32.00 31.00

339

30.00 29.00 Fabric A

28.00

Fabric B 27.00 0

15

30

45

60

75

90

105

120

Time (min)

Figure 3. Drying rate: temperature profile

Relative humidity (%)

95.00 Fabric A Fabric B

85.00 75.00 65.00 55.00 45.00 0

15

30

45

60

75

90

105

120

Time (min)

The humidity profile shows three different phases: at the beginning a steep humidity increase marks the moment when water is poured on the fabric. Then, humidity keeps constant for a period to the saturation value for each fabric. In this phase liquid water is abundant on the fabric surface and the controlling process is evaporation. As the amount of water decreases, the humidity on the surface decreases below the saturation value and the controlling process becomes the water transport from the bulk to the fabric surface. It can be observed that the saturation humidity value is larger for fabric B than fabric A. This is coherent with the result of the horizontal liquid diffusion test: as carbon particles retain water and the same amount of water is concentrated in a smaller area, the surface humidity of fabric B is larger than that of fabric A. The buffering capacity is shown in Figure 5. The fabric weight increase with respect to the dried sample is less than 0.6 per cent for both fabric A and B. Thus, activated carbon has a strong tendency to form intermolecular bonds with liquid water but these bonds are not strong enough to capture water molecules in the vapour phase. The heat, vapour and liquid management properties of the two fabrics are summarised in a radar plot as shown in Figure 6. The properties are included in this plot

Figure 4. Drying rate: humidity profile

IJCST 22,5

0.60 Fabric A Fabric B

340

Weight increase (%)

0.50 0.40 0.30 0.20 0.10

Figure 5. Buffering capacity

0.00 20

30

40

50

60

70

80

Relative humidity (%) Air permeability 10 8 Drying time

6

Rct

4 2 0 HLDA

Figure 6. Thermo-physiological properties of fabric A and B compared in a radar plot

Vertical wicking

Ret

Fabric A Fabric B

Buffering capacity

with scales from 0 to 10. The ratings are calculated by dividing the corresponding values by the greatest value for that specific property. For Rct, Ret and drying time, for which the lower the value of the property the better the performance, the rating is calculated with the reciprocal ratio. The radar plot evidences a different behaviour of the two fabrics: the modified PES fabric has better performances in the absorbency properties, such as vertical wicking and buffering capacity, while the conventional PES fabric is better as far as heat and moisture management is concerned. The modified PES fabric should be comfortable as long as sweating is moderate removing humidity from skin giving the user a comfortable dryness sensation. When physical activity is intense and a large evaporative cooling is needed to obtain thermoregulation, the modified fabric should be less comfortable due to the slow liquid desorption, which slows down the evaporative process. To confirm our hypothesis, wear trials are being carrying out to express an overall comparative comfort evaluation of T-shirt made of the two fabrics. The methodology and the results of the wear trial campaign will be presented in the second part of this work.

5. Conclusion The thermo-physiological characterization of two knitted PES fabrics was carried out. Both fabrics are intended for sports use and they are made of two interlaced layers. The back-side of one fabric contains activated carbon particles trapped within the PES fibres. According to the manufacturer’s statement carbon inclusion is aimed at improving PES hydrophilicity and cooling effect. Heat, vapour and liquid management was investigated. Vertical wicking and HLDA confirmed that the modified fabric had a better hydrophilicity than the conventional one. The HLDA of the modified fabric was found to be smaller since it retain more water per unit area thanks to the carbon particles sorption ability. Thus, on the one hand, the liquid management of the modified PES fabric is better but, on the other hand, desorption is slow and drying time is longer for the modified fabric than the conventional one. As a consequence, wet heat loss is smaller for the modified PES. As far as dry heat loss and vapour management is concerned, dry thermal resistance and vapour resistance of the modified fabric is slightly worse. In this thermal condition (T 258C) also thermal diffusivity play an important role in modified PES thermal flow management, allowing a rapid adjustment of its temperature to that of the surroundings. According to these results, the desired effect of improving the evaporative cooling by increasing the mass-transfer area has not been achieved. On the basis of preliminary wear trials, the modified PES fabric is more comfortable than the conventional PES one as long as sweating is moderate because it easily removes humidity from skin, giving the user a comfortable dryness sensation. When physical activity is intense and a large evaporative cooling is needed for thermoregulation, the modified fabric seems to be less comfortable due to the slow liquid desorption, which slows down the evaporation process. Moisture management together with thermal properties are important factors also during post activity phase (recovery), allowing human temperature to return to physiological ranges. Wear trials are currently in progress with the aim of expressing an overall comparative comfort evaluation of a T-shirt made of the two fabrics. In the second part of this work, the results of the wear trial campaign will be treated extensively and discussed on the basis of the fabric characterization presented in this paper. References Barker, R.L. (2008), “Multilevel approach to evaluating the comfort of functional clothing”, JFBI, Vol. 1 No. 3, pp. 1-4. Celcar, D., Meinander, H. and Gersˇak, J. (2008), “Heat and moisture transmission properties of clothing systems evaluated by using a sweating thermal manikin under different environmental conditions”, Int. J. Clothing Sci. Tech., Vol. 20 No. 4, pp. 240-52. Collier, B.J. and Epps, H.H. (1999), Textile Testing and Analysis, Prentice-Hall, Englewood Cliffs, NJ. Corbellini, S., Ferraris, F. and Parvis, M. (2008), “A system for monitoring workers’ safety in an unhealthy environment by means of wearable sensors”, Instrumentation and Measurement Technology Conference Proceedings, Warsaw, 1-3 May, Vol. 1, pp. 951-5. Hes, L. (2002), “Recent developments in the field of users friendly testing of mechanical and comfort properties of textile fabrics and garments”, paper presented at the World Congress of the Textile Institute, Cairo, 23-27 March.

Thermophysiological comfort 341

IJCST 22,5

342

Holme´r, I. (2004), “Thermal manikin history and applications”, Eur. J. Appl. Physiol., Vol. 92, pp. 614-8. ISO11092 (1993), Textiles – Physiological Effects – Measurement of Thermal and Water-vapour Resistance under Steady-state Conditions, International Organization for Standardization, Geneva. Li, Y. (2001), “The Science of clothing comfort”, Textile Prog. Ser., Vol. 31 Nos 1/2, pp. 22-32. Martı´nez, N., Gonza´lez, J.C., Rosa, D. and Alca´ntara, E. (2009), “A methodology of selecting a suitable garment for sports use”, Int. J. Clothing Sci. Tech., Vol. 21 Nos 2/3, pp. 146-54. Slater, K. (1985) in Charles, C. (Ed.), Human Comfort, Thomas, Springfield, IL. Umbach, K. (1993), “Moisture transport and wear comfort in microfiber fabrics”, Melliand Engl., Vol. 74 No. 2, pp. 78-80. Verdu, P., Rego, J.M., Nieto, J. and Blanes, M. (2009), “Comfort analysis of Woven/Polyester fabrics modified with a new elastic fiber, part 1 preliminary analysis of comfort and mechanical properties”, Textile Res. J., Vol. 79, pp. 14-23. Weder, M., Rossi, R.M., Chaigneau, C. and Tillmann, B. (2008), “Evaporative cooling and heat transfer in functional underwear”, Int. J. Clothing Sci. Tech., Vol. 20 No. 2, pp. 68-78. About the authors R. Splendore obtained the degree in Biology (1999) at University of Turin. She has specialized in Clinical Pathology at the University of East Piedmont-Novara (2004). Her research and clinical laboratory activities (2004-2007) focused on the study of human physiology parameters. Since 2008, her research activity focused on skin physiology and textile-skin interaction and on thermo-physiological comfort of apparel in the High Textile Technology Laboratory in Biella. R. Splendore is the corresponding author and can be contacted at: [email protected] F. Dotti obtained the degree in Chemical Engineering (2003). Her research activity (2005-2006) concerned electrospinning and filtration with textile media at CNR-ISMAC (Consiglio Nazionale delle Ricerche – Istituto per lo Studio delle Macromolecole). She obtained Master in Health and Textile (2006). Since 2007, her research activity concerns thermo-physiological comfort of textiles in the High Textile Technology Laboratory in Biella. B. Cravello obtained the degree in Biology (1996). Her research activity (1996-1999) focussed on textiles in the ISMAC Institute of Biella and from 2000 to 2002 in the Farmacogenomics Laboratory of Biella. Since 2003, she is a Scientific Referee at Associazione Tessile e Salute, a non-profit body that works for the safety and the quality of the textile product. Since 2004, her research activity concerns thermo-physiological comfort of textiles in the High Textile Technology Laboratory of Biella. A. Ferri obtained the degree in Chemical Engineering at Politecnico di Torino (1999) and PhD at Politecnico di Torino (2002). Since 2006, her research activity concerns thermo-physiological comfort of apparel. She served as a Referee of the Textile Research Journal and Research Assistant at the Chemical Engineering Department of Politecnico di Torino.

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

Algorithms of the Automatic Landmark Identification for various torso shapes Hyunsook Han and Yunja Nam Clothing and Textile Department, Seoul National University, Seoul, South Korea, and

Automatic Landmark Identification 343 Received 28 December 2009 Accepted 4 April 2010

Su-Jeong Hwang Shin Apparel Design and Manufacturing Program, Texas Tech University, College of Human Science, Lubbock, Texas, USA Abstract Purpose – The purpose of this paper is to provide algorithms of the automatic landmark extraction software program that are applicable for any torso shape. Design/methodology/approach – In this study, Automatic Landmark Identification (AULID), an automatic landmark extraction software program, was developed to extract consistent landmark locations from any torso shape. A methodology of geometrical characteristics of the body surfaces around each landmark was used for the algorithms and implemented with Cþþ . The accuracy of the AULID was tested on various torso shapes. The verification methodology consisted of mean difference (MD), mean absolute differences (MAD), and one-way analysis of variance. Duncan test for multiple comparisons was used to evaluate the significant differences of MAD values among different torso groups. The MAD values were compared to the anthropometric survey allowable errors. Findings – The algorithms of AULID provided both accuracy and consistency of identifying landmarks on any body torso types. Originality/value – Most 3D body scanning systems often show landmark location errors when dealing with nonstandard body shapes. None of automatic landmark extraction software program provides consistency of identifying landmarks in various body shapes. However, algorithms of AULID, an automatic landmark extraction software program, in this study are only consistent definitions for identifying landmarks in any torso shape. Keywords Computational geometry, Measurement, Garment industry, South Korea Paper type Research paper

1. Introduction An automatic body measurement method using 3D body scan technology has been developed and utilized in the apparel industry. The accuracy of body measurements is very important for apparel manufacturers to develop patterns and sizing systems. However, most automatic body scan measurement methods often show landmark location errors when dealing with nonstandard body figures (Ashdown and Dunne, 2006). The error nullifies the advantage of saving time of automatic body measurement system and makes body scan measurements inaccurate. The inaccuracy causes ineffective sizing systems for apparel mass production. The accuracy and consistency of measurements are related to algorithms of the automatic landmark extraction that are usually predefined by 3D body scan developers.

International Journal of Clothing Science and Technology Vol. 22 No. 5, 2010 pp. 343-357 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011071811

IJCST 22,5

344

There are three approaches to automatic landmark extraction. The first approach is to identify a landmark by using geometric characteristics of body surfaces around each landmark. The second approach is to use statistical relationships among landmarks. The third approach is to match the individual body shape to a template with model landmarks marked on it. In the first approach, Dekker et al. (1999) detected landmarks based on the relation between the surface shape of each body part and other landmarks (Dekker et al. 1999). A limitation of the study (Dekker et al. 1999) was that only the accuracy of sizes was verified rather than the landmark positions. Wang et al. (2003) extracted feature points using fuzzy logic, and the feature points of the human body were extracted from the relation between the surface shape of each body part and other landmarks (Wang et al. 2003). However, Wang et al. (2003) did not verify the accuracy of the feature points because it focused on the generation of a body feature model. Leong et al. (2007) automatically detected landmarks through image processing and computer geometry by logically and mathematically analyzing feature point definitions (Leong et al. 2007). In the second approach, Azouz et al. (2006) used the learning of spatial relations among the characteristics of landmarks and body scan data with landmark sets (Azouz et al., 2006). The learned information was formalized into a pair-wise Markov network. Each node on the network corresponding to the landmark position was a random variable. An edge on the network indicated the positional relation between a pair of landmarks. In addition, Azouz et al. (2006) performed statistical inference on the Markov network for positioning landmarks. Azouz et al. (2006) verified the accuracy of the extracted landmarks. In the third approach, Au and Yuen (1999) recognized features by creating an original feature model. Landmarks were placed on a torso mannequin, and the original feature model is scanned. Then, each individual landmark in the original feature model was compared and matched to the point clouds in the scanned feature (Au and Yuen, 1999). These approaches had both advantages and disadvantages. However, none of them were verified for consistency of identifying landmarks in various body torso shapes. Thus, the purpose of this study is to provide algorithms of the automatic landmark extraction that are applicable for any torso shape. In this study, algorithms of an automatic landmark extraction software program, called as Automatic Landmark Identification (AULID) were developed and tested results to provide consistent landmark locations in any torso shape. 2. Research method 2.1 The 3D body scan subjects A data set of the 5th Size Korea was used in this study. The 5th Size Korea National Sizing Survey was conducted between 2003 and 2004 in South Korea. Subjects, 1,704 males and 1,718 females, were scanned with the WB4 body scanner (Cyberware Co. Ltd, USA) during the Size Korea National Sizing Survey. Table I shows rudimentary statistics on age, height, and BMI of the subjects. Subjects were age over 20 years old. Average of

Table I. Basic descriptive statistics of the subjects

Gender

Age

n

Mean height

Mean BMI

Male Female Total

20 , 69 20 , 69

1,704 1,718 3,422

1,693 mm 1,570 mm

24.2 22.9

men’s BMI was 24.2 and women’s BMI was 22.9. Average of men’s height was 1,693 mm and women’s height was 1,693 mm. A data set of 20 females and 20 males for each body figure group was selected from the 5th Size Korea and tested with algorithms of identifying landmarks in this study. Algorithms were developed for identifying five landmarks on the torso: nipples, underbust, waist, abdomen and hip point. Table II shows the landmarks were related to body figure factors such as body weight, waist shapes and abdomen shapes. Each body figure was defined with the body figure factors. Therefore, the factors were considered for identifying landmarks in each body figure group. For example, body figures were grouped by body weights to identify the landmarks of nipple and under bust because the breast shapes were related to obesity (Park and Sohn, 1996; Sohn and Ko, 2000; Cho and Sohn, 2001). The waist was determined with shapes of the waist (¼ body figure factor). The waist landmark was identified with a concave point on the torso from the front view, and the existence of concave point was different by the waist types (e.g. X or H type). The abdomen landmark was identified with the most forward protruded point on the abdomen shape from the side view. The existence of the protruded point was different by the abdomen types (e.g. protrusion type or flat type). The hip landmark was identified with the most backward protruded point from the side view. The body figures could not be grouped by the hip shapes since all body figures had the same protruded point.

Automatic Landmark Identification 345

2.2 Methodology of AULID In this study, AULID, an automatic landmark extraction software program, was developed to extract consistent landmark locations from any torso shape. A methodology of identifying landmarks was based on the maximum value, the minimum value, the radical slope changes in front view, the silhouettes, and the cross sections. In addition, statistical position of each landmark was used when any distinct geometrical feature and criterion was not identified. Algorithms of the AULID were implemented with Cþþ, and the following coordinate system was used: the leftward direction is to be the x-axis; the upward direction, the y-axis; and forward direction, the z-axis. The x, y, z value of a landmark referred to as width location, height location and depth location, respectively. 2.3 Verification of the accuracy and consistency The accuracy and consistency of the AULID were verified by evaluating both mean differences (MD) and mean absolute differences (MAD). The MD was calculated by subtracting the measure of each manually marked (MM) point from the measure of each corresponding automatically identified (AI) point. A positive (þ ) MD means that the measure of the AI point is larger. In addition, one-way analysis of variance and

Landmark

Body figure factors

Body figures

Nipple Underbust Waist Abdomen Hip

Body weight Body weight Waist shape Abdomen shape No

Underweight/normal weight/overweight Underweight/normal weight/overweight X type/H type Protrusion type/flat type No

Table II. Landmarks on the torso and Body figures

IJCST 22,5

346

Duncan test for multiple comparisons were used to verify if there were any MAD value differences among body figure groups in the significance level set at p # 0.05 level (a . b . c). The MAD of each measurement was compared with the allowable technical errors in the anthropometric survey (ANSUR) Natick/TR-89-044 technical reference (Gordon et al. 1989). The allowable errors are calculated by the following formula where d1,2 are the difference between the first and second measurement (Gordon et al. 1989): sffiffiffiffiffiffiffiffiffiffiffiffi P 2 d1;2 2N 3. Automatic Landmark Identification 3.1 Bust point landmark identification (only for women) (1) Height location determination. The height of bust points were determined as a “first point where slop degree change from minus to plus value (PslopC) on the side silhouette (Sils) from up (armpit height) to down” (PslopC(Sils)). This approach was suitable for finding any slop degree changes on the bust points even on the flat bust shape and the protruded abdomen shape (Figure 1). The following algorithm for identifying height location was programmed in Cþ þ : Algorithm. The height of nipple Target zone, T ¼ {P[ fSils(set of side silhouette points, front): P.y<armpit.y} Step 1: Calculate slop from up to down with two points, Pup and Pdown. in order. Step 2: If (sign of the slop>0) then do ‘step 1’ again with next two points Else set the Pup.y as the height of nipple. (2) Width location determination. The width location of the nipple point on the side silhouette was significantly different from that of actual nipple point. Table III shows the differences in width position between the actual bust point and the side silhouette bust point. The actual nipple point were more outward than the side silhouette nipple point by 18-35 mm. The bust points of obese body type were more outward than that of

Bust point

Bust point

Max protruding point

Slop –

Max protruding point Slop +

Figure 1. Identifying bust points

Woman Man Maximum protruding point method (a)

Woman Slop change method (b)

normal type as shown in Figure 2. Table IV shows the mean ratio “distance between nipple points” to “bust width”. Therefore, this study adjusted the width position of the side silhouette bust point to be mean ratio position of “distance between nipple points” to “bust width” as shown in Table IV. The following algorithm for identifying width location was programmed in Cþ þ :

Automatic Landmark Identification

Algorithm. The width location of nipple Target zone, T ¼ {P[ fSecCnipple(set of front cross section points): P.y ¼ nipple.y} Step 1: Calculate the statistical width location of nipple, the nipple.x: nipple.x (right nipple) ¼ C.x – ((bust width £ mean ratio of ‘distance between nipple points’ to “bust width”)/2) where C.x is the x-coordinate of center point of fSecC. Step 2: Calculate the x-distance with the nipple.x for all points in T x-distance ¼ ABS(P.x-nipple.x) Set the point of minimum x-distance as the nipple point.

347

3.2 Underbust point identification (only for women) This study identified the underbust point on a sagittal section. Among many sagittal sections, we used a sagittal section (Ssbust) which passing through the nipple point because the underbust position had been clearly revealed at the section. The following algorithm for identifying the underbust point was programmed in Cþ þ , and the result shows in Figure 3. Body type (n) Low weight (10) Normal weight (10) Over weight (10)

Max protruding point Real BP at side view

Mean

SD

18.2 24.2 36.5

9.3 16.8 6.7

Table III. Difference in width position between actual nipple point and side silhouette nipple point (unit: mm)

Max protruding point at side view Real BP

Normal weight (a)

Over weight (b)

Figure 2. Difference in width position between actual nipple point and side silhouette nipple point

IJCST 22,5

348

Algorithm. The underbust point Target zone, T ¼ {P[ fSecS nipple (set of sagittal section points, front): Pu.y
Table IV. The ratio of “distance between nipple points” to “bust width”

Women Men

Mean

m 2 3s , m þ 3s

0.520 0.614

0.403 , 0.637 0.491 , 0.736

Source: Size Korea 3D scan measurements of adults

BP

Under bust pt || Max distance pt

Figure 3. Underbust point identification method

Y –

Sagittal section at nipple

Pu

Step 1: Find the Pup, Pdown Pup ¼ point of underbust height Pdown¼ point of middlehip height Step 2: Find the Pconcave Calculate perpendicular distance from the line LPup,Pdown Pconcave¼ point of Max(perpendicular_distance(P, LPup,Pdown)) Step 3: Calculate the upper angle (Angleup) with the Pup and Pconcave Angleup¼ angle between two vectors, v(1,0) and v(Pdown.x -Pup.x, Pdown.y -Pup.y) Step 4: If (Angleup<908) then waist type ¼ X Else waist type ¼ H

Automatic Landmark Identification 349

Figure 4 shows waist types that were classified by using three points; point of underbust height (Pup), point of middle hip height (Pdown) and point of the largest distance from the line connecting those two points (Pconcave). When torso type classification was based on front silhouette within a range between bust height and hip height, many women did not have the concave points within the range, or the concave points were found under bust. Therefore, the search range for women’s waist concave points was set between Pup at underbust height and Pdown at middle hip height. Pup was determined at armpit height (instead under bust height) for men, and Pdown was determined at middle hip height. The middle hip position was determined at least not to be above the waist height using statistical waist height range. Table V shows the statistical waist height range based on proportion of the waist height location. The ratio of “vertical distance between waist and crotch” to “vertical distance between back neck and crotch” were applied to define the middle hip position (Pdown).

UpPt Up angle

UpPt Up angle

ConcavePt

ConcavePt

Side angle

Side angle DownPt

DownPt

Woman (a)

Man (b)

Figure 4. The points, angles on the torso front silhouette used for waist type classification

IJCST 22,5

350

(2) Waist point identification method of waist type X. After waist type classification, if it is classified into X type, the concaveP is determined as the waist point. The following algorithm of identifying waist points for waist type X was programmed in Cþ þ : Algorithm. The waist point of waist type X Target zone, T ¼ {P[ Silf (set of torso front silhouette middlehip.y
points):

(3) Waist point identification method of waist type H. The bodies of waist type H have no geometrical body surface features around the waist. Therefore, the following methods were tested for defining the waist position: Method I using small of the back point and Method II using statistical mean position of waist. Method I: Using the small of the back point. Method I was based on “the small of the back point” to find waist landmark location. The small of the back was defined as the point where the spine had the largest indent when viewed from the side. It was necessary to investigate the small of the back point could match with the waist landmark on 20 women’s torsos. The following algorithm was programmed in C þ þ for testing Method I (using the small of the back point): Algorithm. The waist point of waist type H, small of back Target zone, T ¼ { P[ bSils(set of torso side silhouette points, back): middlehip.y
“Waist-crotch distance” to “back neck-crotch distance” Women Men Table V. Statistical waist height range based on proportion of the waist height location

Mean (SD) m 2 3s , m þ 3s m 2 2s , m þ 2s m2 s,mþs

0.429 (0.026) 0.350 , 0.507 0.376 , 0.481 0.402 , 0.455

Source: Size Korea 3D scan measurements of adults

0.420 (0.030) 0.331 , 0.510 0.361 , 0.480 0.391 , 0.450

Automatic Landmark Identification

Waist height

Waist height

Waist height Back concave pt

Back concave pt

(a) Small difference

Back concave pt

(b) Moderate difference

351 Figure 5. Height difference between small of back point and actual waist point

(c) Large difference

Method II was based on geometric mean of waist height’s ratio to find the waist landmark location. The geometric mean ratios, “mean ratio of waist height to stature” and “mean ratio of waist-crotch distance to back neck-crotch distance”, were compared to investigate which one was close to actual waist height location. The following algorithm was programmed in C þ þ for testing Method II (using the statistical mean of waist height’s ratio): Algorithm. The waist point of waist type H, mean position Target zone, T ¼ {P[ fSils(set of torso side silhouette points, front): middlehip.y
Sex (n) Women (20) Men (20)

Height difference between actual waist and “mean ratio position of waist height to stature” Mean MAD SD 5.3 17.8

14.7 20.0

21.8 16.5

Height difference between actual waist and “mean ratio position of waist-crotch distance to back neck-crotch distance” Mean MAD SD 2 0.1 10.2

8.6 12.6

15.5 12.3

Table VI. Comparison of height difference between actual waist and two mean waist position (unit: mm)

IJCST 22,5

352

3.4 Abdomen point identification method (1) Abdomen types. In physical measurement, the abdomen point was set at the most forward protruded point between underbust height and hip height. However, the definition could not be applied to all people because many of them did not have the most forward protruded abdomen. In addition, it was difficult to make a clear definition of abdomen point because there was no skeletal feature at the abdomen and the shape of abdomen was various. According to the existing definition for physical measurement (ISO 8559, 1989), the abdomen point was found at the body of prominent belly (see Figure 6 (a) prominent type). However, it was difficult to set abdomen point at the subject whose body did not have prominent point at the belly because actual abdomen protruded on the lower or higher part (see Figure 6 (b) (c) (d)). In physical measurement of those obscure abdomen shape subjects, measurers had tendency to determine the abdomen point subjectively near the navel. However, the manual method could not be used for an automatic scanning procedure. (2) Abdomen point identification method. In this study, abdomen point was defined as the most forward prominent point at the front side silhouette similar to the definition of physical measurement. However, we limited the search range to the statistically possible range using the ratio of “vertical distance from abdomen to crotch” to “vertical distance from back neck to crotch”. Table VII shows the geometric search range that was based on the ratios, and the following algorithm was programmed for searching an abdomen point: Algorithm. The abdomen point Target zone, T ¼ {P[ fSils(set of torso side silhouette points, front)}

Max protruding pt Max protruding pt Abdomen Pt

Abdomen Pt

Max protruding pt

Figure 6. Abdomen types – side view

Abdomen Pt

Abdomen Pt Max protruding pt

(a) Prominent type ‘)’

(b) Obscure type ‘]’

(c) Obscure type ‘\’

(d) Obscure type ‘/’

Abdomen-crotch distance to back neck-crotch distance Women Men Table VII. Ratio of “vertical distance from abdomen to crotch” to “vertical distance from back neck to crotch”

Mean (SD) m 2 3s , m þ 3s m 2 2s , m þ 2s m2 s,mþs

0.266 (0.044) 0.134 , 0.398 0.178 , 0.354 0.222 , 0.310

Source: Size Korea 3D scan measurements of adults

0.310 (0.044) 0.177 , 0.443 0.221 , 0.399 0.266 , 0.355

Step 1: Calculate the abdomen search range Down limit ¼ crotch.y þ ((vertical distance from back neck to crotch distance £ m) 2 3s) Up limit ¼ crotch.y þ ((vertical distance from back neck to crotch distance £ m) þ 3s) where m ¼ Mean ratio of ‘abdomen-crotch vertical distance’ to ‘back neck-crotch vertical distance’ Setp 2: abdomen point ¼ point of Max(P.z)

Automatic Landmark Identification 353

3.5 Hip point identification method Hip point was easily found on all bodies because buttock was protruded clearly regardless of body types. The hip point was determined at the most backward prominent point when viewed from the side. We limited the search range to m ^ 3s of ratio of “vertical distance from hip to crotch” to “vertical distance from back neck to crotch”. Table VIII shows the geometric search range, and the following algorithm was programmed in Cþ þ to find hip point: Algorithm. The hip point Target zone, T ¼ {P[ bSils(set of torso side silhouette points, back): crotch.y
Mean (SD) m 2 3s , m þ 3s

Women

Men

0.088 (0.023) 0.018 , 0.158

0.106 (0.026) 0.028 , 0.184

Source: Size Korea 3D scan measurements of adults

Body types (n)

MD

Underweight (20) Normal weight (20) Overweight (20)

2 2.2 2 1.3 2 1.2

Height MAD SD 2.9 3.0 1.2

2.7 3.9 5.1

MC

MD

a a b

1.9 0.2 0.3

Width MAD SD 3.2 2.5 1.2

4.1 3.5 4.3

Table VIII. Ratio of “vertical distance from hip to crotch” to “vertical distance from back neck to crotch”

MC A A A

Note: The different alphabet represents that there is significant difference in MAD among groups in multiple comparison (Duncan) at level of p # 0.05 (a . b . c)

Table IX. Average and standard deviation of the height and width location errors for bust point (unit: mm)

IJCST 22,5

354

the width MD was not significantly different among body types. The height MAD of overweight was significantly smaller than other body types. This is interpreted that the bust point position was clear because the overweight body type had more prominent breast than other body types. 4.2 Underbust point Table IX shows the results of the difference in height and width position between “automatically identified (AI) underbust point” and “manually marked (MM) underbust point”. The MAD of height was small and was less than the allowable error 10 mm according to ANSUR. There was no significant difference among body types. This result indicates that AULID in this study had high accuracy of identifying underbust point location on any body type (Table X). 4.3 Waist point The accuracy of AULID was tested with two waist types, X and H since different methods were applied for each waist type in this study; “waist concave point method” for waist type X and the “mean waist height method” for waist type H. Table XI shows the difference between “automatically identified (AI) points” and “manually marked (MM) points” by the waist types. As shown in Table XI, women’s waist type X had MAD of 4.5 mm and women’s waist type H, 7.1 mm. The both women’s waist types were less MAD than the allowable error 11 mm according to ANSUR. Men’s waist type X had MAD of 15.5 mm and men’s waist type H, 11.4 mm. The both men’s waist types were greater MAD than the allowable error of ANSUR because it was not appropriate for men using the concave point on the front view torso in physical method. However, the AULID in this study had high accuracy of identifying waist landmark location on women’s waist types, showing less MAD than the allowable error.

Table X. Average and standard deviation of the height location errors for underbust point (unit: mm)

Table XI. Average and standard deviation of the height location errors for waist point (unit: mm)

Body types (n)

MD

MAD

SD

MC

Underweight (20) Normal weight (20) Overweight (20)

20.1 20.4 20.4

0.4 0.5 0.4

1.1 1.3 0.8

a a a

Note: The different alphabet represents that there is significant difference in MAD among groups in multiple comparison (Duncan) at level of p # 0.05 (a . b . c)

Sex

Body types (n)

Waist concave point method MD MAD SD

Women

Waist type Waist type Waist type Waist type

0.4 – 15.3 –

Men

X H X H

(20) (20) (20) (20)

4.5 – 15.5 –

6.7 – 10.1 –

Mean waist height method MD MAD SD – 2 1.2 – 9.2

– 7.1 – 11.4

– 11.6 – 12.3

4.4 Abdomen point The accuracy of using AULID was tested with two abdomen types; prominent abdomen and obscure abdomen. In addition, two different search ranges were tested using mean (m) and standard deviation (s) of ratio of “vertical distance between abdomen and crotch” to “vertical distance between back neck and crotch”. One was from “m 2 1s” to “m þ 1s” and the other was from “m 2 2s” to “m þ 2s”. Table XII shows the difference between “automatically identified (AI) points” and “manually marked (MM) points” by the abdomen types. The prominent abdomen type showed MD and MAD of the height between 0 and 5 mm. However, the obscure abdomen type showed MD and MAD of the height between 0 and 47 mm. The difference of m ^ 2s range was greater than that of m ^ 1s range. Thus, this study decided the m ^ 1s as the search range of abdomen. Since the allowable error of abdomen was not available in ANSUR, the allowable error of waist (11 mm) was used for the allowable error of abdomen in this study. According to the allowable error of waist, the MAD of prominent abdomen type was less than the allowable error. However, the MAD of obscure type was more than the allowable error. It was due to unclear definition for the obscure abdomen type in physical measurement method.

Automatic Landmark Identification 355

4.5 Hip point Table XIII shows the difference between “automatically identified (AI) hip points” and “manually marked (MM) hip points”. Compared to the allowable error (7 mm) in ANSUR, the MAD of height was much less than ANSUR. This result indicates that AULID provided both accuracy and consistency on identifying hip landmark location. 5. Conclusion This paper provided algorithms of the AULID that reflected various body torso types. Landmarks were related to body figure factors: the body weight, the waist shape and the abdomen shape. Each body figure was grouped by body weights (¼ body figure factor) to identify the landmarks of bust points and under bust. The waist was

Sex

Body types (n)

Search range: m ^ 1s MD MAD SD

Search range: m ^ 2s MD MAD SD

Women

Obscure abdomen type (20) Prominent abdomen type (20) Obscure abdomen type (20) Prominent abdomen type (20)

2 3.0 0.4 16.3 4.2

28.0 20.8 24.9 3.5

Men

Sex (n) Women (20) Men (20)

17.0 3.2 23.2 4.4

19.9 6.0 26.7 8.2

40.2 2.2 46.6 3.7

45 3.4 47.2 5.9

Mean

MAD

SD

3.1 1.8

3.4 2.8

4.5 3.2

Table XII. Average and standard deviation of the height location errors for abdomen point (unit: mm)

Table XIII. Average and standard deviation of the location errors for hip point

IJCST 22,5

356

determined with the waist shape (¼ body figure factor). The waist landmark was identified with a concave point on the upper body from the front view, and the existence of concave point was different by the waist shapes (e.g. hourglass shape or rectangular shape). The abdomen landmark was identified with the most forward protruded point on the abdomen shape from the side view. The existence of the protruded point was different by the abdomen types (e.g. protrusion type or flat type). The hip landmark was identified with the most backward protruded point from the side view. The body figures could not be grouped by the hip shapes since all body figures had the same protruded point. The results of MAD tests shows that our algorithms of AULID provided both accuracy and consistency of identifying landmarks on any body torso types, except unclear landmark definitions in physical methods. Whenever we found inaccuracy of identifying landmark locations, the problems were due to unclear physical landmark definitions. Certain definitions of the physical landmarks could not be applied for all different body torso types. For example, abdomen landmark location was not easy to be identified in obscure abdomen shape. The waist landmark location with a definition, “the small of the back point”, had no correlation with the waist location. Most women’s concave points were found in the range between the under bust height and garment’s waist band height. There was no accurate definition of physical landmark method for the waist type H and obscure abdomen type. Therefore, it will be necessary to define landmarks based on body figure factors and body shapes. Algorithms in this study will be useful for body scan developers to enhance accuracy of the 3D scan data so that apparel manufacturers and researcher can develop consistent sizing systems for various body shapes. References Ashdown, S. and Dunne, L. (2006), “A study of automated custom fit: readiness of the technology for the apparel industry”, Clothing and Textiles Research Journal, Vol. 24 No. 2, pp. 121-36. Au, C.K. and Yuen, M.M.F. (1999), “Feature-based reverse engineering of mannequin for garment design”, Computer-aided Design, Vol. 31 No. 12, pp. 751-9. Azouz, B.Z., Shu, C and Mantel, A (2006), “Automatic location of anthropometric landmarks on 3D human models”, Third International Symposium on 3D Data Processing, Visualization and Transmission, 3DPVT, University of North Carolina, Chapel Hill, NC, pp. 750-7. Cho, E. and Sohn, H. (2001), “A study on the poor breast shapes for 20’s women”, Journal of Costume and Culture, Vol. 9 No. 1, pp. 11-18. Dekker, L., Douros, I., Buxton, B.F. and Treleaven, P. (1999), “Building symbolic information for 3D human body modelling from range data”, Proceedings of the Second International Conference on 3-D Digital Imaging and Modeling (3DM’99), Ottawa, pp. 388-97. Gordon, C.C., Bradtmiller, B., Clausen, C.E., Churchill, T, McConville, J.T. and Tebbetts, I. (1989), Anthropometric Survey of US Army Personnel: Methods & Summary Statistics. Natick/TR-89-044, US Army Natick Research Development and Engineering Center, Natick, MA. ISO 8559 (1989), “Garment construction and anthropometric surveys – body dimensions”, Reference no. 8559–1989, ISO, Geneva. Leong, I.-F., Fang, J.J. and Tsai, M.J. (2007), “Automatic body feature extraction from a marker-less scanned human body”, Computer-aided Design, Vol. 39 No. 7, pp. 568-82.

Park, E. and Sohn, H. (1996), “A comparative analysis of breast type in 20’s and 40’s women”, Journal of Korean Home Economics Association, Vol. 34 No. 2, pp. 85-97. Sohn, H. and Ko, T. (2000), “A study on the obese breast shape of 20’s women”, Journal of Costume and Culture, Vol. 8 No. 2, pp. 282-92. Wang, C.C.L., Chang, K.K. and Yuen, M.M.F. (2003), “From laser-scanner data to feature human model: a system based on fuzzy logic concept”, Computer-aided Design, Vol. 25 No. 3, pp. 241-53. Further reading ISO 20685 (2005), 3D Scanning Methodologies for Internationally Compatible Anthropometric Databases, ISO, Geneva. Size Korea (2004), 5th Size Korea National Sizing Survey in South Korea, available at: http://sizekorea.kats.go.kr Corresponding author Yunja Nam can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

Automatic Landmark Identification 357

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

IJCST 22,5

Effect of sewing on the drape of goat suede apparel leathers Kaliappa Krishnaraj, Palanisamy Thanikaivelan, Kavati Phebeaardn and Bangaru Chandrasekaran

358

Central Leather Research Institute, Chennai, India Received 12 August 2009 Revised 13 April 2010 Accepted 13 April 2010

Abstract Purpose – Drape is a property that affects the aesthetic appeal and functionality of materials used for clothing manufacture. When various cut components of apparel are assembled together, the drape behavior of the final garment could be affected compared to the base fabric. Goat suede leathers are widely used for making apparels. The purpose of this paper is to analyse the effect of sewing on the drape characteristics of goat suede leather. Design/methodology/approach – For this study commercially available goat suede leathers of Indian origin from five different firms were used. The bending length, drape coefficient and number of nodes were measured for goat suede leathers with different stitch and seam values and compared with the same measurements taken in plain leather. For bending length measurement, plain, single stitched and seamed samples were prepared. For drape coefficient measurement, plain, radially stitched and seamed namely half circle and quarter circle samples were prepared. Findings – The introduction of seam (0.5 and 1 cm allowance) on the goat suede leathers increases the bending length significantly thereby reducing the drape ability but the influence of single stitch on the bending length is negligible. On contrary, there is a significant increase in drape coefficient values for simple radial stitched as well as for seamed samples. The number of nodes generated reduced marginally upon the introduction of stitch or seam. Originality/value – The paper provides the information on the effect of sewing on the drape characteristics of goat suede apparel leathers. This is the first of its kind study on leather as clothing material. Keywords Clothing, Leather, India Paper type Research paper

Glossary Seam

: A line formed by the joining together of two separate pieces of the same or different materials. Stitch : A link or loop or knot made by single pass of a needle in sewing. Stitch length : The length of each complete stitch, usually measured in millimeters. Seam allowance : The amount of space between the edge of the fabric and the line of stitching.

International Journal of Clothing Science and Technology Vol. 22 No. 5, 2010 pp. 358-373 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011071820

Introduction Drape ability is a very important property for any clothing material. It is one of the properties which influence the aesthetic appearance of the clothing material and the overall beauty of the cloth. Chu et al. (1950) described drape as the way that a fabric hangs or falls over a three dimensional form in response to the force of gravity. Understanding drape behavior of clothing material is important for selection of suitable materials and for design and development of apparels. Over the last several decades starting from Peirce (1930), many researchers have developed methods and instruments for assessment of fabric drape. The cantilever bending tester developed by Peirce (1930) measures the two dimensional drape. Further, Chu et al. (1950) designed a drape meter

for three dimensional drape measurements. Later Cusick (1965) developed a drape meter based on similar principles using a planimeter to determine the area of the projected sample for 3D drape measurement. In a modification of this method, Cusick (1968) simplified the procedure of obtaining drape coefficient by using mass rather than area. Both Chu et al. and Cusick used the term drape coefficient (DC) to express the degree of there dimensional deformation of the fabric as a percentage. Recent studies on literature shows that number of nodes formed during drape formation is also used along with drape coefficient to represent the drape ability. More recent developments use digital camera and image analysis to determine the drape coefficient (Vangheluwe and Kiekens, 1993; Behera and Pattanayak, 2008). Stylios and Zhu (1997) introduced a new algorithm, which defines precisely the drape ability of a given fabric. Fabric mechanical properties have been used by Stylios and Wan (1999) for simulating the virtual 3D shape of the fabric, which produce a time-variable deformation of the virtual fabric drape. For making apparel, various components of the apparel has to be sewn together. Hence, it is possible that the sewing could affect the drape of the clothing material. Many researchers have investigated the effect of sewing on the textile fabric drape and concluded that there is a significant change in drape behavior due to the seams compared to the plain fabrics ( Jinlian et al., 1997; Kaushal et al., 2005). Clothing materials range from natural to synthetic fabrics. One of the oldest natural clothing materials of human kind which finds its use even recent time is the skin of the animal or leather. Leather is produced from individual skins and, therefore, differs from textiles, which are available in continuous length with nearly uniform properties. Goat suede leather is the thinnest among the apparel leathers and hence suitable for apparels requiring more fall, flexibility and textile like clothing. Studies on the drape measurements on apparel leathers are limited (Urbanija and Gersak, 2004; Krishnaraj et al., 2008, 2009). Krishnaraj et al. (2009) studied the relation between drape parameters and mechanical properties of goat suede leathers. They concluded that the flexural rigidity, formability, softness, thickness and weight have good correlation with drape coefficient as well as number of nodes. Literature survey reveals that there is no study carried out so for with respect to effect of sewing on the drape characteristics of apparel leathers. The number of components/panels for making leather apparels is generally more compared to textile apparels. One of the reasons for having more number of components in leather apparels is the restricted size of the leather. On an average the total number of components in a leather jacket is around ten. Hence, it is absolutely necessary to investigate the effect of seams on the drape characteristics of apparel leathers. In this study, the bending length, drape coefficient and number of nodes were measured for goat suede leathers with different kinds of stitches and seam values and correlated with the same measurements taken on a plain goat suede leather sample without any seam. Since the structure of the leather is highly complicated when compared to textile fabrics, it was decided to use simple and basic instruments like Peirce bending tester and Cusick drape meter for drape measurements in this study. Materials Commercially available goat suede leathers of Indian origin were procured from five different firms. Leathers were chosen from each firm with fairly uniform substance and size. All the samples for this study were cut from butt region of the leather.

Goat suede apparel

359

IJCST 22,5

360

Table I. Tensile properties of goat suede leathers from different firms

Table II. Description and symbolic representation of samples for bending length measurement

For bending length measurements, the samples were taken along as well as across back bone directions with seam allowance values of 0.5, 1 cm and also without any seam. For drape parameters measurement, three different types of samples were cut namely full circle, half circle and quarter circle. The tensile strength and per cent elongation values of the selected leathers are given in Table I. Methods, measurement and calculation For preparation of samples, the cut components have to be sewn together. Coats Astra 100 per cent staple spun polyester threads were used for sewing with Pfaff 1245 model single needle flat bed unison feed lockstitch sewing machine. Groz-Beckert titanium needle model 134-35-90 R point was used and the stitch length was maintained at 3 mm. After the sewing, the seams were press opened and gently hammered. For bending length measurements, the samples were designated as LPlain, LSS, LS0.5, LS1, CPlain, CSS, CS0.5 and CS1 as shown in Table II. For drape coefficient measurements, the samples were designated as FCPlain, FCSRS, FCDRS, HCS0.5, QCS0.5, HCS1 and QCS1 as shown in Table III. Half circle and quarter circle samples were attached with different seam allowance values to form full circle to enable drape measurement. The principle employed in bending length measurement is to measure a particular length of the fabric specimen of specified dimensions which when used as a cantilever

Firm1 Firm2 Firm3 Firm4 Firm5

Tensile strength (MPa)

Elongation (%)

19.7 ^ 1.5 22.8 ^ 0.8 21.0 ^ 0.9 23.4 ^ 1.2 26.1 ^ 1.6

44.4 ^ 1.0 45.5 ^ 1.7 48.3 ^ 1.5 38.7 ^ 0.9 55.5 ^ 2.1

Note: The values are average of ten samples along with standard deviation

Sample identification

Description

LPlain

Along backbone without any seam stitch

CPlain LSS CSS LS0.5

Across backbone without any seam stitch Along backbone, single stitch, lengthwise Across backbone, single stitch, lengthwise Along backbone, 0.5 cm seam allowance, single stitch, lengthwise

CS0.5

Across backbone 0.5 cm seam allowance, single stitch, lengthwise

LS1

Along backbone 1 cm seam allowance, single stitch, lengthwise Across backbone 1 cm seam allowance, single stitch, lengthwise

CS1

Symbolic representation

Sample identification Description

FCPlain

Symbolic representation

Goat suede apparel

Full circle sample without any stitch or seam

361 FCSRS

Full circle with single radial stitch (SRS) through diameter

FCDRS

Full circle with double radial stitch (DRS) 908 apart

HCS0.5

Two half circle samples joined by SRS with 0.5 cm seam allowance

QCS0.5

Four quarter circle samples joined by SRS with 0.5 cm seam allowance

HCS1

Two half circle samples joined by SRS with 1 cm seam allowance

QCS1

Four quarter circle samples joined by SRS with 1 cm seam allowance

bends to a constant angle under its own weight. Size of the test specimens were 25 £ 200 mm. Bending length was determined following Peirce’s method using SASMIRA stiffness tester according to Indian Standard (IS) 6490 (1971). Bending length equals half the length of rectangular strip of fabric that will bend under its own weight to an angle of 41.58. Low value of bending length indicates low stiffness which means better drape ability and a high value of bending length indicates high stiffness or poor drape ability. The drape coefficient was determined according to IS 8357 (1977) using Bombay Textile Research Association drape meter. The drape coefficient was expressed in percentage. The low value of DC indicates easy deformation (better drape ability) and

Table III. Description and symbolic representation of samples for drape measurement

IJCST 22,5

362

a high value of drape coefficient indicates less deformation (poor drape ability). A circular leather specimen of 25 cm diameter was sandwiched between two horizontal discs of smaller diameter (12.5 cm), and the unsupported annular ring of fabric was allowed to hang down under the action of gravity. A planar projection of the contour of the draped specimen was recorded on a light sensitive paper. The drape pattern obtained was cut along the outline and its area was determined gravimetrically. The drape coefficient was calculated (based on Cusick’s method) as a ratio of the projected area of the drape specimen to its theoretical maximum as given below: Drape coefficient ðDC%Þ ¼

ðw=WÞ 2 a £ 100 A2a

where: W ¼ Mass per unit area of the paper. w ¼ Mass of the drape pattern. a ¼ Area of circle of 12.5 cm diameter ¼ 122.8 cm2. A ¼ Area of circle of 25 cm diameter ¼ 491.1 cm2. The folds present in the contour of drape profile were referred to as nodes and the number of such nodes formed was also recorded for each sample. Softness of goat suede leathers was determined following IUP 36 test method using ST 300 digital leather softness tester. Softness was expressed in mm of leather deflection. The size of the reducing ring used in the softness tester was 20 mm. Softness values were measured in four positions and the mean values were calculated for each sample. Thickness was measured according to IS 5914 (1970). Thickness was measured in four positions and the mean value is calculated for each sample. Weight of the leather samples were measured using an electronic balance with two digit accuracy. The average values of softness, thickness and weight of the goat suede leathers from different firms are shown in Table IV. Results and discussions Bending length The bending length values of goat suede leathers procured form different firms cut along back bone direction are given in Table V and for samples cut across backbone direction are given in Table VI. It is observed that the value of bending length increases slightly or almost same for leathers with single stitch compared to plain samples both along as well as across back

Table IV. Softness, thickness and weight of goat suede leather samples from different firms

Firm1 Firm2 Firm3 Firm4 Firm5

Softness (mm)

Thickness (mm)

Weight per unit area (gm/dm2)

6.10 ^ 0.23 5.21 ^ 0.30 5.49 ^ 0.33 5.59 ^ 0.29 6.53 ^ 0.14

0.46 ^ 0.02 0.5 ^ 0.04 0.54 ^ 0.02 0.46 ^ 0.04 0.45 ^ 0.04

2.6 ^ 0.2 3.6 ^ 0.5 3.6 ^ 0.3 3.1 ^ 0.1 3.0 ^ 0.3

Note: The values are average of seven samples along with standard deviation

bone directions. Bending length values of along backbone leather samples for different seam conditions are plotted as shown in Figure 1. From the Table V and Figure 1, it is evident that the bending length values increases significantly by the introduction of a stitch/seam. The increase in the bending length from plain leather to the seam leather ranges from 1.05 to 2.1 cm for samples cut along back bone direction.

Goat suede apparel

363 Firm1 Firm2 Firm3 Firm4 Firm5

L Plain

LSS

LS 0.5

LS 1

1.74 2.53 2.46 2.31 1.90

1.88 2.50 2.29 2.34 2.08

2.84 4.21 3.84 3.90 2.95

2.90 4.31 3.89 4.41 3.01

C Plain

CSS

CS 0.5

CS 1

1.65 2.38 1.94 2.40 1.78

1.94 2.49 2.18 2.06 1.99

3.30 4.21 3.83 3.64 2.86

2.91 4.15 3.73 3.95 2.86

Note: The values are average of four samples

Firm1 Firm2 Firm3 Firm4 Firm5

Note: The values are average of four samples

Table V. Bending length values (cm) of goat suede leather samples cut along backbone direction

Table VI. Bending length values (cm) of goat suede leather samples cut across backbone direction

4.6 4.4 4.2 4.0 Bending length (cm)

3.8 3.6

Firm1 Firm2 Firm3 Firm4 Firm5

3.4 3.2 3.0 2.8 2.6 2.4 2.2 2.0 1.8 1.6 L Plain

LSS

LS0.5

LS1

Samples with different seam conditions – along backbone

Figure 1. Plot of bending length for different seam conditions of goat suede leather sample – along backbone

IJCST 22,5

364

The bending length values of different seam conditions for across back bone direction are plotted as shown in Figure 2. From the Figure 2, it is observed that for across back bone samples the increase in bending length is significant between plain and stitch/seam samples similar to samples cut along back bone direction and the increase ranges from 1.26 to 1.89 cm. In most cases, the increase in bending length is almost 100 per cent from plain to seamed sample. From Figures 1 and 2 it is observed that the increase in bending length follows almost uniform pattern for leathers from different firms. Drape coefficient Drape coefficient values of goat suede leathers with different seam conditions from five different firms are given in the Table VII. It is seen that the drape coefficient value is higher for all sewn/stitched leather compared to plain leather samples. Drape coefficient values of goat suede leathers from different firms for different seam conditions are plotted in Figure 3. It is noted that the drape coefficient values of leathers from Firm2 increases rapidly compared to leathers from all other firms. While leathers from Firm4 exhibited fairly less increase in drape coefficient values. 4.4 4.2 4.0 3.8 Bending length (cm)

3.6 3.4

Firm1 Firm2 Firm3 Firm4 Firm5

3.2 3.0 2.8 2.6 2.4 2.2 2.0

Figure 2. Plot of bending length for different seam conditions of goat suede leather sample – across backbone

Table VII. Drape coefficient (per cent) of goat suede leather with different stitch/seam conditions

1.8 1.6 C Plain

CSS

CS0.5

CS1

Samples with different seam conditions – across backbone

Firm1 Firm2 Firm3 Firm4 Firm5

FCPlain

FCSRS

FCDRS

HC0.5

QCS0.5

HCS1

QCS1

23.1 31.6 30.7 45.6 26.5

31.0 52.4 39.4 47.5 30.8

28.3 51.1 43.8 54.0 32.2

23.5 50.3 39.6 50.8 34.6

28.5 60.6 50.3 47.9 28.1

24.6 60.0 34.8 49.0 28.3

28.7 53.0 49.9 47.1 28.7

Note: The values are average of four samples

Goat suede apparel

65 60 55 Drape coefficient (%)

50 45

365

40 35 30

Firm1 Firm2 Firm3 Firm4 Firm5

25 20 15 10 5 0 FCPlain FCSRS FCDRS HCS0.5 QCS0.5 HCS1

QCS1

Samples with different stitch/seam conditions

The percentage increase in drape coefficient values due to introduction of stitch/seam in goat suede leathers is shown in Table VIII. From the table, it is noted that the increase in the value of drape coefficient is more in leathers from Firm2 (up to 92 per cent increase) and Firm3 (up to 64 per cent increase) leathers compared to other firms. This may be attributed to the difference in processing and bulk properties of leathers from Firms2 and 3 compared to leathers from other firms. From the Table IV, it is evident that leathers from Firms2 and 3 are heavier, thicker and less soft compared to leathers from other firms which results in more increase in drape coefficient values upon introduction of stitch/seam. Since simple stitch on the leather and sewing with seam allowance are entirely different, it was decided to compare them separately with plain leather. Drape coefficient values of leathers with SRS, DRS and plain (without stitch/seam) are plotted as shown in Figure 4. It is observed that the drape coefficient increases with SRS as well as DRS. In other words, the drape ability reduces with the introduction of stitches. In most cases, drape coefficient is increasing from SRS to DRS. The drape coefficient values of plain and half circle samples with different seam allowances are plotted in Figure 5. From the plot it is observed that for most cases (Firms3-5) the drape coefficient value increases for half circle samples with seam allowance 0.5 cm and decreases when the seam allowance is increased to 1 cm. There is no significant change in DC value for leathers from Firm1 for all seam conditions. The drape coefficient values of QCS0.5, QCS1 and plain leather samples are plotted as shown in Figure 6. It is seen that the drape coefficient value increases for QCS0.5 samples compared to the plain leather samples and then it remains almost constant for QCS1 samples except for Firm2 leather. From the Figures 4-6, it is observed that the drape coefficient values increases upon introduction of stitch or seam for leathers from different firms. The difference in the extent of increase between leathers from different firms may be

Figure 3. Plot of drape coefficient values of goat suede leather samples with different stitch/seam conditions

Table VIII. Increase in drape coefficient of goat suede leather due to stitch/seam 5.2 22 19.5 62 13.1 43 8.4 18 5.7 22

7.9 34 20.8 66 8.7 28 1.9 4 4.3 16

FCDRS-FCPlain

8.1 31

5.2 11

8.9 29

18.7 59

0.4 2

HCS0.5-FCPlain

1.6 6

2.3 5

19.6 64

29 92

5.4 23

QCS0.5-FCPlain

1.8 7

3.4 8

4.1 13

28.4 90

1.5 7

HCS1-FCPlain

366

Firm1 Value % change Firm2 Value % change Firm3 Value % change Firm4 Value % change Firm5 Value % change

FCSRS – FCPlain

2.2 8

1.5 3

19.2 63

21.4 68

5.6 24

QCS1-FCPlain

IJCST 22,5

Drape coefficient (%)

54 52 50 48 46 44 42 40 38 36 34 32 30 28 26 24 22

Goat suede apparel

Firm1 Firm2 Firm3 Firm4 Firm5

367

FCPlain

FCSRS

FCDRS

Full circle samples with different stitches

60

Drape coefficient (%)

55 50

Figure 4. Plot of drape coefficient of goat suede leather samples with simple stitches

Firm1 Firm2 Firm3 Firm4 Firm5

45 40 35 30 25 FCPlain

HCS0.5

HCS1

Half circle samples attached with different seams

attributed to the change in processing conditions adopted by different firms and the difference in the bulk properties of leathers. Numbers of nodes or folds generated during drape coefficient measurement for goat suede leathers from various firms are given in Table IX. The number of nodes and the drape coefficient values are inversely proportional (Krishnaraj et al., 2009). As expected, the number of nodes generated by FCSRS and FCDRS goat suede leathers

Figure 5. Plot of drape coefficient of goat suede leather; half circle samples with different seams

IJCST 22,5

65 Firm1 Firm2 Firm3 Firm4 Firm5

Drape coefficient (%)

60

368

55 50 45 40 35 30

Figure 6. Plot of drape coefficient of goat suede leather; quarter circle samples with different seams

Table IX. Number of nodes of goat suede leather with different stitch/seam conditions

25 Plain

QCS0.5

QCS1

Quarter circle samples attached with different seams

Firm1 Firm2 Firm3 Firm4 Firm5

FCPlain

FCSRS

FCDRS

HCS0.5

QCS0.5

HCS1

QCS1

9 6.5 7 6.5 9

8 6.5 6.5 6 8

8.5 6 6.5 6 6.5

9 5 6.5 5.5 7

8.5 5 6.5 6 8.5

9 6.5 6 6 9

8 5.5 7 6.5 8

Note: The values are average of four samples

samples are lower or equal to the number of nodes generated by FCPlain samples. It is also seen that the introduction of seam results in similar or reduced number of nodes in most cases. The number of nodes for different seam conditions is plotted as shown in Figure 7. Maximum change (decrease) in number of nodes was noted between plain leather and HCS0.5 for the Firm5 followed by Firm2. Number of nodes formed for FCSRS, FCDRS and FCPlain goat suede leather samples are plotted as shown in Figure 8. The number of nodes decreases for both FCSRS and FCDRS samples when compared with FCPlain sample in most of the cases. The number of nodes either decreases or remains constant for leathers from all the firms upon introduction of stitches except for Firm1 leathers. Numbers of nodes for half circle and quarter circle samples with different seam allowance values are plotted as shown in Figures 9 and 10, respectively. It is seen that the number of nodes decreases when 0.5 cm seam allowance is applied for both half circle and quarter circle samples. And in most of the cases when the seam allowance value is increased from 0.5 to 1 cm, the number of nodes slightly increases or remains the same. This trend is similar to the observation made during drape

Goat suede apparel

No. of nodes

8

369

6

4 Firm1 Firm2 Firm3 Firm4 Firm5

2

0 FCPlain FCSRS FCDRS HCS0.5 QCS0.5

HCS1

QCS1

No. of nodes

Samples with different stitch/seam conditions

9.2 9.0 8.8 8.6 8.4 8.2 8.0 7.8 7.6 7.4 7.2 7.0 6.8 6.6 6.4 6.2 6.0

Figure 7. Plot of number of nodes of goat suede leather samples with different stitch/seam conditions

Firm1 Firm2 Firm3 Firm4 Firm5

FCPlain

FCSRS

FCDRS

Full circle samples with different stitches

coefficient analysis. In general, the number of nodes decreases when a stitch/seam is introduced in goat suede leathers. In order to understand the influence of number of components on the drape parameters it was decided to plot the drape parameters for quarter circle and half circle

Figure 8. Plot of number of nodes of goat suede leather; full circle samples with different stitches

IJCST 22,5

11

10

9 No. of nodes

370

Firm1 Firm2 Firm3 Firm4 Firm5

8

7

6

Figure 9. Plot of number of nodes of goat suede leather; half circle samples with different seam allowance

5 FCPlain

HCS0.5

HCS1

Half circle samples with different seams

9.0 8.5

No. of nodes

8.0 7.5

Firm1 Firm2 Firm3 Firm4 Firm5

7.0 6.5 6.0

Figure 10. Plot of number of nodes of goat suede leather; quarter circle samples with different seam allowance

5.5 5.0 FCPlain

QCS0.5

QCS1

Quarter circle samples with different seams

samples for a fixed seam allowance. The drape coefficient of plain, HC and QC samples for 0.5 cm seam allowance is plotted as shown in Figure 11. The same plot for number of nodes is shown in Figure 12. It is noted that the drape coefficient value increases and the number of nodes decreases due to paneling. When the number of panel increases (HC to QC), the drape

Goat suede apparel

65 Firm1 Firm2 Firm3 Firm4 Firm5

60

Drape coefficient (%)

55

371

50 45 40 35 30 25 FCPlain

HCS0.5

QCS0.5

Figure 11. Plot of drape coefficient of goat suede leather; different paneled samples

QCS0.5

Figure 12. Plot of number of nodes of goat suede leather; different paneled samples

Different paneled samples Note: 0.5 cm seam

9.0 8.5

Firm1 Firm2 Firm3 Firm4 Firm5

No. of nodes

8.0 7.5 7.0 6.5 6.0 5.5 5.0

FCPlain

HCS0.5 Different paneled seams

Note: 0.5 cm seam

coefficient increases for most samples except for Firms4 and 5. While the number of node values remain almost constant for both HC and QC samples except for Firm5 leather. Since this behavior is not consistent, further detailed analysis is needed for clear understanding.

IJCST 22,5

372

Conclusions In this study, the effect of stitch and seam on the drape properties of the goat suede leathers have been analysed by measuring the bending length, drape coefficient and number of nodes. The bending length increases marginally when a single stitch is introduced in a plain leather sample. But the introduction of seam (0.5 and 1 cm allowance) increases the bending length significantly thereby reducing the drape ability. The drape coefficient values are increasing for single and double radial stitched samples compared to plain samples. As expected, the number of nodes is less for stitched samples compared to plain samples. The drape coefficient of 0.5 cm seam samples increases compared to plain samples, but further increase in seam allowance did not alter the drape coefficient values significantly. Further studies are needed for understanding the influence of number of panels on the drape parameters. References Behera, B.K. and Pattanayak, A.K. (2008), “Measurement and modeling of drape using digital image processing”, Indian Journal of Fibre and Textile Research, Vol. 33, pp. 230-8. Chu, C.C., Cummings, C.L. and Teixeria, N.A. (1950), “Mechanics of elastic performance of textile materials part V: a study of the factors affecting the drape of fabrics – the development of a drape meter”, Textile Research Journal, Vol. 20, pp. 539-48. Cusick, G.E. (1965), “The dependence of fabric drape on bending and shear stiffness”, Journal of the Textile Institute, Vol. 56, pp. T596-T606. Cusick, G.E. (1968), “The measurement of fabric drape”, Journal of the Textile Institute, Vol. 59, pp. 253-60. Indian Standard (IS) 6490 (1971), Method of Determination of Stiffness of Fabrics-cantilever Test, Bureau of Indian Standards, New Delhi. IS 5914 (1970), Method of Physical Testing of Leather, Bureau of Indian Standards, New Delhi. IS 8357 (1997), Method for Assessment of Fabric Drape, Bureau of Indian Standards, New Delhi. Jinlian, H., Siuping, C. and Ming-tak, L. (1997), “Effect of seams on fabric drape”, International Journal of Clothing Science and Technology, Vol. 9 No. 3, pp. 220-7. Kaushal, R.S., Behera, B.K., Roedel, H. and Andrea, S. (2005), “Effect of sewing and fusing of interlining on drape behavior of suiting fabrics”, International Journal of Clothing Science and Technology, Vol. 17 No. 2, pp. 75-90. Krishnaraj, K., Thanikaivelan, P. and Chandrasekaran, B. (2008), “Mechanical properties of sheep nappa influencing drape”, Journal of the American Leather Chemists Association, Vol. 103, pp. 215-21. Krishnaraj, K., Thanikaivelan, P. and Chandrasekaran, B. (2009), “Relation between drape and mechanical properties of goat suede garment leathers”, Journal of the Society of Leather Technologists and Chemists, Vol. 93, pp. 1-7. Peirce, F.T. (1930), “The handle of cloth as a measurable quantity”, Journal of the Textile Institute, Vol. 21, pp. T377-T416. Stylios, G.K. and Wan, T.R. (1999), “The concept of virtual measurement: 3D fabric drapeability”, International Journal of Clothing Science and Technology, Vol. 11 No. 1, pp. 10-18. Stylios, G.K. and Zhu, R. (1997), “The characterization of the static and dynamic drape of fabrics”, Journal of the Textile Institute, Vol. 88 No. 4, pp. 465-75.

Urbanija, V. and Gersak, J. (2004), “Impact of the mechanical properties of nappa clothing leather on the characteristics of its use”, Journal of the Society of Leather Technologists and Chemists, Vol. 88, pp. 181-90. Vangheluwe, L. and Kiekens, P. (1993), “Time dependence of the drape coefficient of fabrics”, International Journal of Clothing Science and Technology, Vol. 5 No. 5, pp. 5-8.

Goat suede apparel

Further reading IUP 36 (2000), “Measurement of leather softness”, Journal of the Society of Leather Technologists and Chemists, Vol. 84, pp. 377-9.

373

Corresponding author Bangaru Chandrasekaran can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0955-6222.htm

IJCST 22,5

New method for investigating the dynamic pressure behavior of compression garment

374

Yongrong Wang, Peihua Zhang, Xunwei Feng and Yuan Yao

Received 26 January 2010 Revised 24 April 2010 Accepted 24 April 2010

Key Laboratory of Textile Science and Technology, Ministry of Education, College of Textiles, Donghua University, Shanghai, Peoples Republic of China Abstract Purpose – The paper aims to develop a system and measuring method for investigating the dynamic pressure behavior of compression garments. Design/methodology/approach – The dynamic pressure behavior measurement, realized by use of the self-designed system, is a direct measuring method, which is based on a rigid hemisphere with five pressure sensors distributed on its surface. The dynamic pressure is measured over time under the process of fabric 3D deformation. The pressure distributions at the basic five sites are accepted as the measuring results. The dynamic stiffness index can be calculated from dynamic pressure profile and 3D deformation of compression garments. Findings – The measuring system records the pressure-time curve and pressure-deformation curve. The dynamic pressure stiffness index expresses the change in pressure owing to the change in elongation of compression fabrics. The pressure measuring system and the index provide much information in the field of compression garment assessment. Research limitations/implications – Another characteristic that was not mentioned but important is pressure hysteresis, which can give the information about pressure decay when fabrics undergoing repeated stretch and relaxation. The influence factors of hysteresis and its role in compression garments also requires further research. Originality/value – To determine and characterize the dynamic pressure behavior of compression garment under 3D deformation, this study develops a measuring system and defines a new index. The measuring system can be used in scientific research institutes and factories, contribute to optimize process parameters and quality control of compression garment. Keywords Dynamics, Pressure, Measurement, Garment industry Paper type Research paper

International Journal of Clothing Science and Technology Vol. 22 No. 5, 2010 pp. 374-383 q Emerald Group Publishing Limited 0955-6222 DOI 10.1108/09556221011071839

1. Introduction “Compression garment” is meant garment which applies pressure to specific areas of human body for operational reasons such as tight-fitting garments or girdles, or for medical reasons, such as the management or treatment of venous ulceration, deep vein thrombosis or burns (Dias et al., 2006). Compression garment is characterized by the pressure it exerts on the specific area of human body. In daily practice it is well known that the behavior of a compression garment can significant differ from others, the pressure changed due to the dimensional deformation of compression fabric, relate to its physical and mechanical properties, such as tensile, shearing, bending, and compression performances, etc. (Liu et al., 2005, 2007). The pressure function of the compression garment depends on the difference between minimum pressure and maximum pressure during body movement (Stolk et al., 2004). This difference is resulted from fabric

deformation and the pressure stiffness index of compression fabrics. Compression garment with a high pressure stiffness index, which benefits the consumers with a “compression effect” and comfortable sensation. The dynamic stiffness index for fabric is a new index of the dynamic pressure performance defined in this paper, which will characterize the dynamic pressure performance at different elongation level, calculated based on the dynamic pressure profile and 3D deformation of compression garments. For pressure to be effective in compression garments, a pressure measuring system is essential to test the amount of pressure actually provided under fabric 3D deformation. In the European Committee for Standardization (2001), the pressure values of a compression garment are measured indirectly in a laboratory setting. The tensile tension of the compression garment is measured under semi-static conditions. Laplace’s Law, which relates the pressure, the tensile tension and the curvature radius, was used to calculate the pressure of the compression garment. Lately, a variety of developed measuring techniques can be used to determine pressure performance directly under each device, however, no one system has been established or identified as the only or best way to measure these pressures. Nishimatsu et al. (1998) measured pressure of socks using an elastic optical fiber. Moreover, Yu et al. (2004) and Fan and Chan (2005) reported the development of a basic system to measure garment pressure based on a soft manikin. And several other studies have been conducted on designing new pressure measuring system based on air-pack type pressure sensor, Flexiforce pressure sensor (Martin et al., 2000; Teng et al., 2007; Kang et al., 2007). Among these pressure measuring systems, sensors are stuck on the surface of supporting devices, such as soft manikin (Yu et al. and Fan et al.), head manikin (Kang et al.) and human body (Teng et al.). The thickness of sensors and connector wires contribute to the measurement accuracy since they are easy to cause the additional fabric or garments deformation. Additionally, these dimensions of supporting devices are fixed, only static pressure can be tested. For the pressure measuring system developed in this paper, the outside of sensors are distributed on the surface of hemisphere, the other parts and the connector wires are inserted inside of the hemisphere, which contribute to the improved measurement stability and accuracy. Further, the hemisphere presses the fabric samples at a predetermined speed under control of displacement driving device, the pressure changes and fabric deformation are recorded, which provides the possibility of the dynamic pressure behavior investigation under the process of fabric 3D deformation, and the dynamic pressure and stiffness may be assessed. The aim of this study is to develop a new measuring system and simple method that can provide useful information concerning both the dynamic pressure and the stiffness of different compression fabrics. 2. Investigation methods 2.1 Test stand description The measuring system for monitoring the dynamic pressure behavior of compression fabric is a direct measuring method, which is base on a rigid hemisphere with five pressure sensors distributed on its surface. Figure 1 shows the bottom view of sensors distribution on the surface of hemisphere. The five sensors record the dynamic pressure value over time during the fabric 3D deformation, which are accepted as the measuring results. The measuring system is composed of four parts, including pressure monitor device, sample holding device, displacement driving device, and data acquisition device.

Dynamic pressure behavior 375

IJCST 22,5 Site 2

376 Site 1

Site 3 Site 5

Site 4

Figure 1. The bottom view of sensors distribution on the surface of hemisphere

Figure 2 shows the scheme and the photograph of the dynamic pressure measuring system. The basic structure and mechanism are as follows: . Pressure monitor device is composed of a rigid hemisphere and five highprecision pressure sensors, the sensors are distributed on the surface of hemisphere, monitor pressure changes during the hemisphere downward motion. . Sample holding device is composed of up-holding device and down-holding device for fixing samples in a certain position.

4

5 3

6 7

8

9 2

11

10

1 (a)

Figure 2. The dynamic pressure measuring system

(b)

Notes: (a) Schematic diagram; (b) photograph (1, support saddle; 2, bottom board; 3, framework; 4, center sill; 5, connecting bar; 6, rigid hemisphere with five distributed sensors; 7, up-holding device; 8, down-holding device; 9, operational channel; 10, mounting plate; 11, bottom plate)

.

.

Displacement driving device is composed of a digital-to-analog converter card, a voltage/frequency converter and a driving motor. The driving motor controls the hemisphere vertical motion at predetermined speed. The data acquisition device is composed of a signal processor, data collection software, and PC.

Dynamic pressure behavior 377

Technical parameters of the system are listed below: . pressure measurement range: 500 gf; . measurement resolution: 0.05 percent; . the diameter of sensor: 4 mm; . the diameter of hemisphere: 18 cm; and . the net diameter of hold device: 20 cm. 2.2 Investigation procedure To evaluate the dynamic pressure behavior of compression garment, first start the computer-driven system software for equipment self-test and standard initialization, then set measuring parameters (including the speed of vertical motion, maximum distance, etc.). The drive motor keeps the hemisphere moving in vertical direction according to the predetermined parameters. Pressure sensors detect pressure signal, transmitted to computer via signal amplifier, data acquisition, and processing module, finally the signal was recorded and displayed. In our experiment, the hemisphere moves vertically downward 60 mm then move upward until original position at a constant speed of 150 mm/min. Figure 3 shows the schematic diagram of data processing and acquisition. 3. Experiment 3.1 Material Five samples were knitted on machine as detailed in Table I. One kind of knitting structure was used as plain-knitted. There are four different contents of DOW XLA elastic fiber as 7-9, 11 percent, respectively.

Pressure sensor

Signal amplification

Data acquisition Data logging Print

Hemisphere (a) Notes: (a) Photograph of interface; (b) schematic diagram

Driving system (b)

Parameters setting

Computer

3.2 The physical and mechanical properties The physical and mechanical properties of samples were measured using a Kawabata evaluation system for fabric standard evaluation system and INSTRON tensile tester

Figure 3. The data processing and acquisition

IJCST 22,5

in laboratory with a controlled environment (temperature: 20 ^ 2 8C, relative humidity: 65 ^ 3 percent). Table II shows material physical and mechanical properties of DOW XLA elastic fabrics. Each sample was measured repeatedly three times for warp and weft direction, respectively. In order to minimize the shape instability of the knitted fabric, all samples were conditioned in a standard atmosphere for 24 hours prior to the formal testing.

378 4. Test result and discussion 4.1 The dynamic pressure distribution of a compression fabric Five samples were evaluated and similar pressure curves were recorded by the dynamic pressure measuring system. The pressure values are increasing as the hemisphere downward motion and reach the maximal value, then decreasing as the hemisphere upward motion. Figure 4 shows the dynamic pressure behavior over time for sample 2. There is a large difference between the dynamic behavior at site 5 and the other sites. The sensor 5 on the bottom of the surface, first contacts with fabric, and pressure instant increase during the hemisphere downward motion, around 15 s later, the other four sensors started to contact with fabric, each pressure value increases immediately. The pressure at site 5 is larger than at other four sites at the most deformation status.

Table I. Basic description of tested DOW XLA elastic fabrics

Samples

Fabric structure

1 2 3 4 5

Piece Piece Piece Piece Piece

solid single solid single solid single solid single solid single

Properties

Indices

Tensile

LT (%) G (gf/cm degree) 2HG (gf/cm) 2HG5 (gf/cm) B (gf cm2/cm) 2HB (gf cm/cm) LC WC (gf cm/cm) RC (%) MIU MMD SMD (um) TO (mm) W (mg/cm2)

Shearing Bending Compression Surface Weight and thickness Table II. Material physical and mechanical properties of DOW XLA elastic fabrics

dyed dyed dyed dyed dyed

Fiber content jersey jersey jersey jersey jersey

7% lastol and 93% cotton 8% lastol and 92%cotton 11% lastol and 89% cotton 9% lastol, 55% cotton and 36% modal 11% lastol, 53% cotton and 36% modal

Sample 1

Sample 2

Sample 3

Sample 4

Sample 5

85.47 1.78 3.50 3.80 0.25 0.31 0.36 0.47 55.30 2.49 2.65 1.12 1.16 210

86.67 1.82 4.10 5.00 0.11 0.14 0.38 0.48 58.30 2.15 1.10 1.25 1.08 185

92.38 1.11 1.70 2.00 0.09 0.09 0.41 0.53 58.50 2.56 0.96 1.13 1.12 155

82.66 1.60 2.10 4.10 0.17 0.17 0.44 0.48 58.30 2.42 0.99 1.02 1.05 190

88.10 1.33 1.60 2.40 0.53 0.11 0.29 0.46 39.10 2.40 0.88 1.19 1.23 165

Notes: LT, linearity of extension curve; G, shear stiffness; 2HG, shear hysteresis (0.58); 2HG5, shear hysteresis (58); B, bending stiffness; 2HB, bending hysteresis; LC, linearity of compression curve; WC, compression energy; RC, compression resilience; MIU, coefficient of friction; MMD, mean deviation of MIU; SMD, geometrical roughness; TO, thickness at 0.5 gf/cm2 pressure; W, weight per unit area

Dynamic pressure behavior

35

Pressure (Kpa)

30 Site 1 Site 2 Site 3 Site 4 Site 5

25 20 15 10

379

5 0

0

4

8

12

16

20

24 28 Time (s)

32

36

40

44

48

Figure 4. The dynamic pressure behavior over time for sample 2

Figure 5 shows the dynamic pressure behavior over elongation for sample 2. The pressure over elongation curves at site 1-4 are similar, because they are on the same position level, and they contact with fabric at the same time theoretically, however slightly different pressure value occurred among these four sites. The knitted structure of stitches, free slide of successive stitches and local threads expansions contribute to the pressure variation. With the hemisphere upward motion, the pressure attenuation is faster than increasing when its downward motion, the tensile and recovery properties of XLA fabrics are the potential reasons. 4.2 The dynamic pressure behavior of different compression fabric The pressure and elongation recordings gave a hysteresis curve relating the dynamic pressure behavior and the deformation of five compression fabrics. Figures 6 and 7 show the dynamic pressure over elongation behaviors at sites 5 and 3, respectively. At the maximum deformation, the pressure values of five samples followed by, at site 5, 38.24, 33.23, 13.80, 24.84, 19.58 kpa; at site 3, 25.20, 23.35, 9.65, 16.98, 12.64 kpa. During the measurement, the pressure is increasing with the increment of 3D deformation, it is obvious that five fabrics present different pressure values while at the same elongation, the physical and mechanical properties should be taken into account for it. Among the sample 1-3, at the same elongation level, the smaller pressure value with the larger contents of elastic fiber, larger contents of elastic fiber could improve the elasticity, which result in the gentle pressure performance. Compare the samples 3 and 5, the pressure is higher with the additional Modal fiber. 35 Site 1 Site 2 Site 3 Site 4 Site 5

Pressure (Kpa)

30 25 20 15 10 5 0

0

5

10 Elongation (%)

15

20

Figure 5. The dynamic pressure behavior over elongation for sample 2

IJCST 22,5 Pressure (Kpa)

380

40

Figure 6. The dynamic pressure behavior over elongation at site 5

35

Sample 1

30

Sample 2

25

Sample 3

20

Sample 4

15

Sample 5

10 5 0 0

5

10 Elongation (%)

15

20

10 Elongation (%)

15

20

30 Sample 1

Pressure (Kpa)

25

Figure 7. The dynamic pressure behavior over elongation at site 3

Sample 2

20

Sample 3

15

Sample 4

10

Sample 5

5 0

0

5

During the downward motion, three phases are included for site 5: (1) pressure value increases rapidly within 0.5 percent elongation; (2) pressure value increases almost linearly with the elongation from 0.5 to 19 percent; and (3) the relative constant pressure value with the elongation 19-20 percent. The measuring set, sensor accuracy should be taken into account for the first phase and third phase. For the second phase, clothing pressure is formed by fabric deformation, which is dependent on its stretch properties, so the dynamic pressure performance is related with the tensile, recovery properties within its elasticity rang. During the upward motion, three phases are included for site 5: (1) A stable pressure value or a slightly decrease with the elongation 20-18 percent, the main reason for this, inertia continues the increased deformation trend when the hemisphere reach the downward maximum and started to back, also the inner friction among threads hinder the fabric into a narrow shape. (2) With the elongation from 18 to 10 percent, a sharp decrease occurred in dynamic pressure behavior. The tensile properties and the relative speed between fabric recovery and the hemisphere upward should be taken into consideration. First, the elasticity of fabric is composed of elastic filament and stitch structure of fabric, apart from elasticity, elastic filament also shows the

so-called phenomenon of hysteresis, which is the loss in recovered linear length of an elastic product after which has been subjected to stretching and relaxation. Second, the inner friction occurs during the stitches slipping, it hinders the fabric into a narrow shape, causes the recovery speed of fabric lower than the hemisphere upward motion’s. (3) From 10 to 0 percent, the pressure value decreases to zero. The plastic deformation of fabric, the relative speed between hemisphere and fabric cause the separation between them, so the pressure decreases to zero. Further, the dynamic pressure behavior under different downward and upward speeds should be identified.

Dynamic pressure behavior 381

4.3 A new index for determination dynamic behavior of compression fabric In order to quantify the dynamic behavior of the compression fabric we introduce a new index: the dynamic pressure stiffness index. Stiffness is defined as the increment in pressure when the elongation increased by 1 percent. The stiffness of compression garment is important for their function. The higher the stiffness is, the more difficult it is to put on and pull off the compression garment, but at the same time it is more effective in keeping fitting, reducing refluxes, and improving ambulatory venous hypertension. The compression garments that are acceptable to the consumers should have a correct balance between comfort and effectiveness. To determine the stiffness, three different records were taken. First, the pressure was recorded for an elongation that was 0.5 percent smaller than the right size. Second, the pressure for elongation that was 0.5 percent larger than the right size was measured. The stiffness then was calculated by the difference value. Figure 8 shows the dynamic stiffness index of five compression fabrics at different elongation level. It is obvious that the stiffness increase with the increasing elongation of five compression fabrics. Different stiffnesses were presented, among samples 1, 2, 3, the smaller stiffness with the larger contents of elastic fiber at the same elongation level. Larger contents of elastic fiber could improve the elasticity, which result in the gentle pressure increment. Compare the samples 3 and 5, the stiffness was higher with the additional Modal fiber. Also, the internal frictional forces play a relatively important role in the small elongation variations. The stitches of the compression fabric must slide along each as the fabric stretching, giving rise to internal frictional forces that oppose this expansion. In the opposite direction, when the deformation of fabric is decreasing, the elastic 4

Stiffness

2 0 Sample 1 Sample 2 Sample 3 Sample 4 Sample 5

–2 –4 –6 –8

0

4

8 12 Elongation (%)

16

20

Figure 8. The dynamic stiffness index at different elongation level

IJCST 22,5

threads pull the compression fabric into a narrow shape. Again, the internal frictional forces between the stitches oppose these elastic forces. In stretching this knit fabric we have to overcome not only the elastic forces but also the high friction forces among fibers, threads, and stitches. These internal frictional forces cause hysteresis in the pressure-elongation curve as shown in Figures 6 and 7.

382

5. Conclusion To determine the dynamic pressure behavior of compression garment, this study develops a system for measuring the dynamic pressure behavior under fabric 3D deformation, introduces a new index: the dynamic pressure stiffness index. The new measuring system and index provide much needed information in the field of compression garment assessment. Another characteristic that was not mentioned but important is pressure hysteresis, which can give information about pressure decay when fabrics undergoing repeated stretch and relaxation. The influence factors of hysteresis and its role in compression garments will be the scope of our further research. References Dias, T., Cooke, W., Fernando, A., Jayawarna, D. and Chaudhury, N.H. (2006), “Pressure garment”, US Patent, US7043329B2, USPTO, Alexandria, VA. European Committee for Standardization (CEN) (2001), No-Active Medical Devices. Working Group 2 ENV 12718: European Prestandard “Medical Compression Hosiery”, CEN TC 205, CEN, Brussels. Fan, J. and Chan, A.P. (2005), “Prediction of girdle’s pressure on human body from the pressure measurement on a dummy”, International Journal of Clothing Science and Technology, Vol. 17 No. 1, pp. 6-12. Kang, T.J., Park, C.H., Jun, Y.M. and Jung, J. (2007), “Development of a tool to evaluate the comfort of a baseball cap from objective pressure measurement(I) – holding power and pressure distribution”, Textile Research Journal, Vol. 77 No. 9, pp. 653-60. Liu, R., Kwok, Y.L., Li, Y., Lao, T.T. and Zhang, X. (2005), “Effects of material properties and fabric structure characteristics of graduated compression stockings (GCS) on the skin pressure distributions”, Fibers and Polymers, Vol. 6 No. 4, pp. 322-31. Liu, R., Kwok, Y.L., Li, Y., Lao, T.T. and Zhang, X. (2007), “Quantitative assessment of relationship between pressure performances and material mechanical properties of medical graduated compression stockings”, Journal of Applied Polymer Science, Vol. 104 No. 1, pp. 601-10. Martin, F.P., Satsue, H. and Duncan, B. (2000), “Evaluation of a sensor for low interface pressure applications”, Medical Engineering & Physics, Vol. 22 No. 9, pp. 657-63. Nishimatsu, T., Ohmura, K., Sekiguchi, S., Toba, E.J. and Shoh, K.J. (1998), “Comfort pressure evaluation of men’s socks using an elastic optical fiber”, Textile Research Journal, Vol. 68 No. 6, pp. 435-40. Stolk, R., Wegen van der-franken, C.P. and Neumann, H.A. (2004), “A method for measuring the dynamic behavior of medical compression hosiery during walking”, Dermatologic Surgery, Vol. 30 No. 5, pp. 729-36. Teng, T.L., Chou, K.T. and Lin, C.H. (2007), “Design and implementation of pressure measurement system for pressure garments”, Information Technology Journal, Vol. 6 No. 3, pp. 359-62. Yu, W.N., Fan, J.T., Qian, X.M. and Tao, X.M. (2004), “A soft mannequin for the evaluation of pressure garments on human body”, Sen-I Gakkaishi, Vol. 60 No. 2, pp. 57-64.

Further reading Park, C.H., Jun, Y.M., Kang, T.J. and Kim, J.H. (2007), “Development of a tool to measure the pressure comfort of a cap (II)-by the analysis of correlation between objective pressure and subjective wearing sensation”, Textile Research Journal, Vol. 77 No. 7, pp. 520-7. Corresponding author Peihua Zhang can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

Dynamic pressure behavior 383

International Journal of Clothing Science and Technology

ISSN 0955-6222 Volume 22 Number 6 2010

International textile and clothing research register Editor-in-Chief Professor George K. Stylios

Access this journal online __________________________

2

Editorial advisory board ___________________________

3

Editorial __________________________________________

4

Research register __________________________________

6

Research index by institution_______________________

90

Research index by country _________________________

91

Research index by subject__________________________

92

Research index by principal investigator ____________

94

CONTENTS

This journal is a member of and subscribes to the principles of the Committee on Publication Ethics

New for 2010 † As part of Emerald’s commitment to highlighting the impact of research, authors will, from this volume, be able to flag up implications for research, practice and/or society. Our structured abstracts format aims to pinpoint further for the reader, the utility of the research in question.

EDITORIAL ADVISORY BOARD Professor Jaffar Amirbayat Amirkabir University of Technology, Tehran, Iran Professor H.J. Barndt Philadelphia College of Textiles & Science, Philadelphia, USA Professor Mario De Araujo Minho University, Portugal Professor Dexiu Fan China Textile University, Shanghai, China Professor Jintu Fan Hong Kong Polytechnic University, Hong Kong Professor P. Grosberg Shankar College of Textile Technology and Fashion, Israel Professor Carl A. Lawrence University of Leeds, UK

Professor Trevor J. Little North Carolina State University, USA Professor David Lloyd University of Bradford, UK Professor Masako Niwa Nara Women’s University, Japan

Editorial advisory board

3

Professor Isaac Porat School of Textiles, University of Manchester, UK Professor Ron Postle The University of New South Wales, Australia Professor Rosham Shishoo Swedish Institute for Fibre and Polymer Research, Mo¨lndal, Sweden Professor Paul Taylor University of Newcastle, Newcastle upon Tyne, UK

Professor Gerald A.V. Leaf Heriot-Watt University (Hon), UK

International Journal of Clothing Science and Technology Vol. 22 No. 6, 2010 p. 3 # Emerald Group Publishing Limited 0955-6222

IJCST 22,6

4

International Journal of Clothing Science and Technology Vol. 22 No. 6, 2010 pp. 4-5 Emerald Group Publishing Limited 0955-6222

Editorial The International Textiles and Clothing Research Register championing the research efforts of the community The International Textile and Clothing Research Register (ITCRR) is now in its 15th year of publishing the research efforts of our community. You can see the breadth of activity, the large funding involved, the hot topics, and the research players in the field of textiles and clothing research. By so doing ITCRR provides a platform of participation and dissemination to those working in our discipline and avoids duplication of effort. As you see in this new edition, textiles and clothing research and practice is increasing in volume, in quality and in diversity, all good news for all of us involved in the field. We try to capture as many projects as possible, but I understand that there may be projects that have not been registered. I will welcome them in our next issue and I invite them to send me their project details anytime during the year for the next ITCRR issue. Research, development, and innovation can, without doubt, give us more wisdom, enable our industries to become more competitive, and contribute to our quality of life, even more so today with the current economic turbulence. I believe that registering research projects will provide the due credit to originators of the research and contribute to the future development of this field. Groups of expertise can be identified in this manner, repetition and re-invention can be avoided leading to best utilisation of time and funding for faster and better-directed research. Textiles and clothing originate from the physiological need to protect ourselves from the environment. New challenges however are already upon us with nanotextiles, nanofibres, nanocoatings, with multifunctional and smart textiles and clothing, and with wearable electronics. Special International Journal of Clothing Science and Technology (IJCST) issues are being planned for promoting technical, functional textiles and clothing. We continue to welcome contributions from textile and clothing aesthetics, design and fashion highlighting our belief that design and technology go hand-in-hand. IJCST was set up 22 years ago as a platform for the promotion of scientific and technical research at an international level. Our original statement that the manufacture of clothing needs to change to more technologically advanced forms of production and retailing still stands; however, IJCST has evolved further by also providing opportunities in the new research areas of nanotextiles, smart textiles and clothing and in wearable electronics. The journal, now fully indexed in SCI continues with its authoritative style to accredit original technical research, adhering to our refereeing processes however difficult these may prove at times. IJCST will be instrumental in continuing to support conferences and meetings from around the world in its effort to promote the science and technology of clothing. I would like to thank our research community and those authors in particular that have contributed to this volume, our editorial board for their continuous support, and our colleagues who have acted in a refereeing capacity and have given us their free time and q Professor George K. Stylios

expertise to progress our research efforts. I take the liberty to list some of those names below (apologies in advance if anyone has accidentally been omitted from this list): . Professor Paul Taylor, University of Newcastle (Emeritus) . Professor Isaac Porat, UMIST . Professor R.H. Wardman, Heriot Watt University . Professor R. Christie, Heriot Watt University . Dr Taoruan Wan, University of Bradford . Professor David Lloyd, University of Bradford . Professor G.A.V. Leaf, Heriot-Watt University (Visiting) . Dr David Brook, University of Leeds . Dr Jaffer Amirbayat, University of Manchester . Professor Jintu Fan, Hong Kong Polytechnic University . Dr X. Chen, University of Manchester . Dr Jelka Gersak, University of Maribor . Dr Hua Lin, Nottingham University . Dr Dan Mei Sun, Heriot Watt University . Dr Lisa McIntyre, Heriot Watt University . Dr T. Wan, Heriot Watt University . L. Luo, Heriot Watt University Correspondence address: Heriot-Watt University, School of Textiles, Netherdale, Galashiels, Selkirkshire, TD1 3HF, Scotland, UK. E-mail: [email protected] and [email protected] George K. Stylios Editor-in-Chief

Editorial

5

IJCST 22,6

Research register Belfast, Northern Ireland UK

6

University of Ulster, School of Art and Design, University of Ulster at Belfast, York Street, Belfast BT15 1ED, Northern Ireland. Tel: +028 90267231; Fax: 028 90267356; E-mail: [email protected] Principal investigator(s): Dr Julie Soden Research staff: Mrs Donna Campbell, Dr Graeme Stewart

3D Reinforced natural fibre woven preforms and eco-composites Academic partners

Industrial partners

Engineering Composites AVL loom USA Research Centre, University of Ulster Jordanstown, Collaborative: Prof. R. Wool, Centre Composite Materials (CCM) University of Delaware, Newark USA, Dr K. Kirwan, Dr S. Maggs, University of Warwick Innovative Manufacturing Research Centre Project start date: November 2007 Project end date: November 2009 Project budget: £391,440 Source of support: AHRC Research Grants Scheme Keywords: 3D Woven, Technical textiles, Engineering composites, VARTM As a cross-discipline investigation, this project successfully navigated the boundaries and synergies that exist between technical textile design and engineering composite domains to produce a wide range of natural fibre 3D woven natural fibre perform and composite parts (NFC’s). This early assessment established the benefits and challenges these materials face, and their future potential. The restriction in increased uptake of NFC’s thus far, has been due to the natural variability and inconsistency of the materials, the limited mechanical performance of nonwoven and plied lay-up preform assemblies, the commercial availability and proven performance of a naturally derived bioresin to form 100% recyclable composites, and cost. The project pioneered the development of using both natural fibres and naturallyderived synthetically produced fibres constructed into woven preforms using advanced digital Dobby and Jacquard technology. The project focused on the design and assemblage of 3D woven textile preforms, tackling the complex design parameters associated with obtaining desired loomstate fibre volume fractions and aesthetic surface

qualities within the material. It also highlighted production and scale-up issues in relation to production volume. The engineering team and collaborators processed these materials using VARTM technology using both standard thermosetting epoxies and bio-derived resins comprised from epoxidised Soy-bean and Furan resin derived from hemicellulose sugars. Adopting various impregnation strategies using VARTM processing technology produced a range of composite specimens, which underwent mechanical test programs which assessed flexural strength and damage resistance.

Project aims and objectives Weave design: to design, manufacture and demonstrate that a range of 3D and multiple layer reinforced woven preforms using naturally derived fibres can combine technical load-bearing properties with multifunctional design characteristics to produce a range of fibre reinforced composite structures for a variety of targeted applications. To assess the benefits, constraints and parameters of both advanced Dobby and Jacquard Weave technology with a view for both prototype generation and production volume requirements. Resin processing: to mould and impregnate fabrics using vacuum assisted resin transfer moulding (VARTM) and infusion techniques into modular composite parts. To optimise resin impregnation processes for 3D natural fibre woven fabrics with thermosetting epoxy and plant-derived bio-resins. To draw comparisons between resin types in terms of compatibility with the fabric article and processing methods chosen. Material testing: to establish initial mechanical properties for load-bearing natural fibre composites under laboratory test conditions. Project: to raise overall profile of natural fibre and Eco-composites, and to filter these through appropriate networks for further debate and collaboration.

Research deliverables (academic and industrial) .

Range of 40+ woven preforms and vacuum injected composite samples for full test programmes in flax, viscose rayon and other natural fibres.

.

Mechanical Test Data. Prototype examples have emerged for architectural, construction, infrastructure and industrial applications. Samples included woven composite wall panels, reinforcements for civil engineering, shaped, expandable, scaffold and containment reinforcements targeted at infrastructural/construction uses and aerodynamic parts for the Worlds First Sustainable Racing Car, (in conjunction with the University of Warwick Innovative Manufacturing Research Centre).

.

Publications and outputs J.A., Soden, G.F.J., Stewart, “Natural fibre composites with 3D reinforcement for new application areas” Critical Edition, Journal Biobased Materials and Bioenergy, pending publication June 2010. J.A., Soden, G.F.J., Stewart, “Natural fibre composites with 3D reinforcement for new application areas” Keynote Address, Natural Fibres 2009, Materials for a Low Carbon Future, Institute of Materials (IOM3), Carlton House London 14-16 December 2009. J.A., Soden, G.F.J., Stewart, D., Campbell, A., McIlhagger “Manufacture and testing of 3D woven natural fibre composites” ICCM-17, 17th International Conference on Composite Materials, 27-31 July 2009, Edinburgh, UK Session 3D Textiles and Composites, 31 July: D1:14, CD, no pagination.

Research register

7

IJCST 22,6

8

Edinburgh, Scotland, UK Heriot-Watt University, School of Engineering and Physical Sciences, Riccarton, Edinburgh, Scotland, UK, EH14 4AS. Tel: 00 44 131 451 3034; Fax: 00 44 131 451 3473; E-mail: [email protected]; [email protected] Principal investigator(s): Professor J.I.B. Wilson and Dr R.R. Mather Research staff: Ms. A.H.N. Lind

Solar cells in textiles Academic partners

Industrial partners

Project start date: 2001 Project end date: 2010 Project budget: Source of support: Keywords: Thin film silicon, Solar energy, Photovoltaics We are developing thin-film silicon solar cells on low cost textile substrates, using chemical vapour deposition (CVD) technology, based on previous thin-film diamond expertise. The CVD technology employs a proprietary microwave plasma system (developed at Heriot-Watt University) with silane/hydrogen/dopant gas mixtures to produce the sequence of layers that forms the active part of these cells. We have shown that relatively low deposition temperatures of 200 degrees C and the active plasma conditions of the process do not affect our textile substrates, whether of woven or nonwoven construction. In addition, solutions have been determined to the problem of providing reliable electrical contacts over fibrous, flexible substrates, together with a conventional transparent conducting oxide as the top contact in the cell “sandwich” structure. Effective “first barrier” encapsulation may also use our deposition technology.

Project aims and objectives Flexible solar cells for a variety of applications:, e.g. building fac¸ades, use in remote areas, emergency use in disaster relief, camping/leisure industry, portable chargers.

Research deliverables (academic and industrial) Working prototype.

Publications and outputs “Solar textiles: production and distribution of electricity coming from solar radiation. Applications” in Intelligent Textiles and Clothing, Ed. H. Mattila, Woodhead Publishing Limited, Cambridge, 2006. “Textiles make solar cells that are flexible and lightweight”, Technical Textiles International, December 2002, pp. 5-6.

Edinburgh, Scotland, UK Heriot-Watt University, School of Engineering and Physical Sciences, Riccarton, Edinburgh, Scotland, UK, EH14 4AS. Tel: 00 44 131 451 3034; Fax: 00 44 131 451 3473; E-mail: [email protected]; [email protected] Principal investigator(s): Professor J.I.B. Wilson and Dr R.R. Mather Research staff: Ms. A.H.N. Lind

Solar cells in textiles Academic partners

Industrial partners

Project start date: 2001 Project end date: Ongoing Project budget: N/A Source of support: N/A Keywords: Thin film silicon, Solar energy, Photovoltaics We are developing thin-film silicon solar cells on low cost textile substrates, using chemical vapour deposition (CVD) technology, based on previous thin-film diamond expertise. The CVD technology employs a proprietary microwave plasma system (developed at Heriot-Watt University) with silane/hydrogen/dopant gas mixtures to produce the sequence of layers that forms the active part of these cells. We have shown that relatively low deposition temperatures of 200 degrees C and the active plasma conditions of the process do not affect our textile substrates, whether of woven or non-woven construction. In addition, solutions have been determined to the problem of providing reliable electrical contacts over fibrous, flexible substrates, together with a conventional transparent conducting oxide as the top contact in the cell “sandwich” structure. Effective “first barrier” encapsulation may also use our deposition technology.

Project aims and objectives Textile based solar cells for a variety of applications: e.g. building fac¸ades, use in remote areas, emergency use in disaster relief, agricultural applications, camping/leisure industry, portable chargers.

Research deliverables (academic and industrial) Working prototype. Publications and outputs “Solar textiles” in Polymer Electronics – A Flexible Technology, Eds. F. Gardiner and E. Carter, iSmithers, Shawbury, 2009. “Solar textiles: production and distribution of electricity coming from solar radiation. Applications” in Intelligent Textiles and Clothing, Ed. H. Mattila, Woodhead Publishing Limited, Cambridge, 2006. “Textiles make solar cells that are flexible and lightweight”, Technical Textiles International, December 2002, pp 5-6.

Research register

9

IJCST 22,6

10

Galashiels, UK Heriot-Watt University, School of Textiles and Design, Netherdale, Galashiels, Scotland, TD1 3HF. Tel: +44 (0) 1896 892140; Fax: 44 (0) 1896 756701; E-mail: [email protected] Principal investigator(s): R.H. Wardman

Verification of algorithm for the numerical specification of standard depths of colour Academic partners

Industrial partners

None None Project start date: May 2008 Project end date: May 2011 Project budget: £12000 Source of support: Society of Dyers and Colourists Keywords: Standard depths, Colour, ISO, Textiles Standard depths of colour are used to determine the fastness properties of dyed materials. The existing standard depths were established over fifty years ago and there is general agreement hat they are not of uniform depth. The project involves the visual assessment of fabric samples dyed to standard depths of shade defined by an algorithm developed by the PI. The visual assessments will be carried out by observers in at least five different countries and their results used to verify the algorithm. The algorithm will then be written into a new proposal for an ISO standard that defines standard depths of colour.

Project aims and objectives To prepare samples according to the algorithm previously published and verify its accuracy by visual assessments of a panel of colourists in five different countries.

Research deliverables (academic and industrial) A new ISO standard for the numerical specification of standard depths of colour. Publications and outputs C.C., Chen, R.H., Wardman and K.J., Smith, “The mapping of a surface of constant visual depth in CIELAB colour space”, Coloration Technol., (2002), 118, 281 R.H., Wardman, S., Islam and K.J., Smith, “Proposal for a numerical definition of standard depths”, Coloration Technol., (2006), 122, 350

Galashiels, UK Heriot-Watt University, School of Textiles and Design, Netherdale, Galashiels, Scotland, TD1 3HF. Tel: +44 (0) 1896 892140; Fax: +44 (0) 1896 756701; E-mail: [email protected] Principal investigator(s): R.M. Christie and R.H. Wardman Research staff: R. Shah

Digital fast patterned microdisposal of fluids for multifunctional protective textiles Academic partners

Industrial partners

University of Manchester, Hogeschool Gent, University of Lodz, University of Twente

Ten Cate, Grado Zero Espace, B&B Corporate Knitwear, Guantenor S.L., J. Sarens, NV, IRIS DP S.r.L, JPC S.P. zoo, Liebaert, Skalmantas, SKA Polska s.p.zoo, Vexed Generation Ltd, Xennia Technology Ltd, D’Appolonia SpA, Lamberti SpA, Xaar PLC, Saxion Project end date: May 2010

Project start date: May 2006 Project budget: e12.6m Source of support: EU FP6 Keywords: Inkjet printing, Textiles

To develop breakthrough technology based on digitally microdisposing fluids on textiles enabling high-speed protective functionalisation, continuous processing and customised production. Digital microdisposal has the ability of exact localisation and patterning of functionalities in multilayer textile substrates integrating advanced thermo and hydro regulation, sensorics, actuating and controlled release functions, based on nanotechnology and multifunctional materials.

Project aims and objectives To realise a generic functionalisation technology for making multifunctional protective textiles on the basis of localised and multilayered compounds. To use low temperature and low water processessing by inkjet application of functional compounds. Develop innovations in mechatronics and micro-fluids, nanotechnology and multifunctional materials as well as in micro-nano metrology. Set new standards of performance in personal protective equipment.

Research register

11

IJCST 22,6

Research deliverables (academic and industrial) On-line process for the microdisposal of active fluids onto textile fabrics. Monitoring system integrating different inspection techniques. Publications and outputs None in the public domain.

12

Galashiels and other site partners, UK Research Institute for Flexibel Materials, School of Textiles and Design, Heriot Watt University, Galashiels, Selkirkshire, TD1 3 HF. Tel: +01896 892135; Fax: 01896 758965; E-mail: [email protected] Principal investigator(s): Prof. G.K. Stylios, RIFleX, Heriot Watt University, Dr K. Lee, Unilever Research (Industrial), Dr R. Potluri, Manchester University and Prof. Long, Nottingham University Research staff: L. Luo plus others in partner universities and industrial partners

Multi-scale integrated modelling for high performance flexible materials Academic partners

Industrial partners

University of Nottingham and Manchester University

Unilever, OCF Plc, TechniTex Faraday Ltd, Crode International Plc, ScotWeave Ltd, Airbags International Ltd, Moxon of Huddersfield Ltd, Carrington Carrer and Workwear Ltd

Project start date: 1 January 2007 Project end date: 31 December 2010 Project budget: £1.7 Million Source of support: Department of Trade and Industry DTI Keywords: Modelling, Yarn, Fabric, Garment, Hagh performance This is a flagship proposal for the UK. It is based on integrating micro, meso and macro scale structure/property and deformation models for high performance flexible materials. The outputs will be industry targeted solutions for predicting the properties and behaviour of high performance flexible materials in deformed states during usage including garments. The proposal stems from the modelling achievements of the three academic partners, combining their complimentary work, and after integration, adapting them for industry use, with Unilever, the lead partner, and 12 companies covering diverse applications of the outputs, attempting to represent this sector. OCF and Scotweave will help commercialising the output in the form of a software product or licence. The main aim of the project is to develop the models such that they can be used primarily by the high performance textiles and garment industry, but also by other industries dealing with flexible materials.

The key strength of the proposal is that for the first time an attempt is made to model the behaviour of flexible materials in a 3D manner, taking into account the dynamic changes of performance related properties with physical changes during use: . Bridging the gap between (a)industry needs for predictive and development tools and (b) academic modelling efforts on two levels: structure/property micro/meso scale and macro-scale whole flexible structures such as high performance clothing. .

Modelling of overall properties and performance of simple deformed textile structures such as draped/creased/folded fabrics, predicting the dynamic changes of performance related to physical deformations. This will be an innovative engineering tool for industries involved in designing, developing and manufacturing flexible materials, initially targeted at fabrics, but with applications to paper and thin films.

Modelling simple high performance whole garments based on structural/ geometrical parameters of the flexible materials, to predict the changes in whole garment performance caused by yarn and fabric changes. No model that can perform this function for the high performance clothing industry is yet available. This aspect of the work will be a powerful tool for the high performance clothing and garment industry (high performance medical wear, protective clothing, sportswear, etc.). The project will lead to two way modelling: predicting properties and performance of deformed textile materials and whole garments from structural, mechanical and geometrical parameters, or vice versa, i.e. generating structural, mechanical and geometrical requirements for specific end-product characteristics. Focusing on the high performance clothing sector, the work will provide a first-in-its-kind tool for design, development, engineering and optimisation of high performance flexible materials and exploited by a consortium of diverse companies, three of which are technology providers and the rest users of the technology. Different mechanisms (e.g. interfacial modification) can be exploited in the manufacture, modification, cleaning and care of high performance multi-component textiles. This will permit rational design of these materials and associated products, reducing the need for experimental development and testing. This project will push the boundaries in multi-scale modelling by building on leading edge expertise in fibre to yarn scale modelling at Manchester University, yarn to textile scale modelling at Nottingham University, and textile to whole flexible structures and clothing at Heriot-Watt University. By developing and interfacing these areas closely coupled tools will allow completely predictive models to be developed. It will also be the first time that modelling techniques have been used to understand the effects of interfacial properties on textile performance, leading to increasing innovation in high performance textile products. .

Project aims and objectives .

.

Combining the strong modelling expertise of the four academic partners to create multi-scale integrated models for predicting structural properties and performance of flexible materials. Optimising the integrated models for industry use, focusing in the first instance at high performance clothing and technical textiles.

Research register

13

IJCST 22,6

14

.

Increasing time and cost efficiency of industry in product design and fabrication, starting with the high performance clothing industry.

The objectives address industry needs for new and improved product design and fabrication aids for high performance materials, e.g. in view of the London 2012 Olympics (high performance sportswear, anti-terrorism protective clothing, etc.). They will enable the prediction of properties and performance before manufacturing, hence accelerating the development stage, reducing costs and increasing competitiveness.

Research deliverables (academic and industrial) A Virtual Testing capability will be developed and transferred into industry, reducing the amount of experimentation required for new product development (new textiles, treatments and laundry products designed and manufactured) and thereby reducing the associated costs and waste generation (improving sustainability of production). Direct beneficiaries of this work will be manufacturers of apparel, textile, technical and allied industries. Indirect beneficiaries will include manufacturers of carpets, non-wovens, composites, paper and structural materials, as well as retailing and software (after commercialising): . Bridging the gap between (a) industry needs for predictive and development tools and (b) academic modelling efforts on two levels: structure/property micro/meso scale and macro-scale whole flexible structures such as high performance clothing. .

Modelling of overall properties and performance of simple deformed textile structures such as draped/creased/folded fabrics, predicting the dynamic changes of performance related to physical deformations. This will be an innovative engineering tool for industries involved in designing, developing and manufacturing flexible materials, initially targeted at fabrics, but with applications to paper and thin films.

.

Modelling simple high performance whole garments based on structural/ geometrical parameters of the flexible materials, to predict the changes in whole garment performance caused by yarn and fabric changes. No model that can perform this function for the high performance clothing industry is yet available. This aspect of the work will be a powerful tool for the high performance clothing and garment industry (high performance medical wear, protective clothing, sportswear, etc.).

Publications and outputs Too early.

Galashiels, Scotland, UK Research Institute for Flexibel Materials, School of Textiles and Design, Heriot Watt University, Galashiels, Selkirkshire, TD1 3 HF. Tel: 01896 892135; Fax: 01896 758965; E-mail: [email protected] Principal investigator(s): Prof. G.K. Stylios Research staff: X. Zhao

Integration of CFD and CAE for design and performance assessment of protective clothing Academic partners

Industrial partners

None

Tilsatec Ltd, TechniTex Faraday Ltd, Camira Fabrics Ltd, St Jame’s University Hospital, Remploy Ltd, Pil Membranes Ltd, Altair Engineering Ltd Project end date: 31 May 2010

Project start date: 1 June 2007 Project budget: £600,000 Source of support: Engineering and Physical Sience Research Council EPSRC and Department of Trade and Industry DTI Keywords: CFD, CAE, Protective fabric, Garment, Apparel, Modelling high performance This collaborative proposal aims at improving and developing new textiles for protective clothing by integrating Computational Fluid Dynamics (CFD) and Computer Aided Engineering (CAE). A new industrial tool for predicting the diffusion of chemical and/or biological (CB) agents through multilayer, non-homogenous flexible porous materials such as fabrics and whole garments will also be established. Multi-layer textile structures (flat and shaped) and simple garments will be modelled, using equations of mass, heat and momentum balance, integrated with human computational representation. The outcome will be optimisation of current commercial fabrics/garments and developing new protective clothing whilst, at the same time offering a new objective tool for product designers, engineers and developers to predict and evaluate performance of CB protective products.

Project aims and objectives .

.

Develop and integrate CAE/CFD for modelling of a clothed human to predict the performance of textiles and garments in CB protective applications, linking with fabric design and manufacturing. Establish an objective measure protocol for CB protection end uses.

.

Improve protection performance of (PPE) and (PC) through new product development for extreme conditions.

.

Develop new materials by better understanding of the complex interactions between the flow of CB agents and textile/materials properties.

Research deliverables (academic and industrial) With a worldwide focus on CB agents, the project is timely for industry, considering the legislative demands for public health, safety and security. Growing concerns from government, the public and private sector about threats of terrorist attacks or epidemic outbreaks have led to increased performance and evaluation requirements. Increasing emphases on security for international events (e.g. London 2012) are also drivers for the above innovations. The proposed project addresses these issues by development key technologies for immediate and long-term use. The consortium experience can achieve the project objectives and lead to a major breakthrough in PC.

Research register

15

IJCST 22,6

Publications and outputs Too early.

Galashiels, Scotland, UK 16

Heriot-Watt University, RIFleX, School of Textiles and Design, Netherdale, Galashiels TD1 3HF, United Kingdom. Tel: +44 1896 89 2135; Fax: +44 1896 75 8965; E-mail: [email protected] Principal investigator(s): Prof. George K. Stylios Research staff: Mohammad Mahfuzur Rahman Chowdhury

Investigating nano fibre production by the electrospinning process Academic partners

Industrial partners

None None Project start date: July 2004 Project end date: December 2010 Project budget: N/A Source of support: N/A Keywords: Electrospinning, Electrospinning process, Parameters, Polymer, Nanofibre application Electrospinning is a unique way to produce novel polymer nanofibres with diameter typically in the range of 10 nm to 500 nm. Using this process, the polymer nanofibres can be made from a variety of polymer solutions or melt to produce fibres for a wide range of applications. Electrospinning occurs when the electrical force at the surface of a polymer solution or melt overcomes the surface tension and causes an electrically charged jet to be ejected. When the jet dries or solidifies, an electrically charged fibre remains. This charged fibre can be directed or collected or accelerated by electrical forces, then collected in sheets or other geometrical forms. This research project is an investigation of the electrospinning process and the effect of process variables on orientation, crystallinity, microstructure and mechanical properties of the nanofibres produced. Some of the polymeric parameters investigated are polymer type, solvent type, molecular weight, solution properties, viscosity, conductivity and surface tension. In the case of process parameters, the electric potential, flow rate, concentration, distance between capillary and collection screen, ambient parameters are important.

Project aims and objectives .

To investigate process-structure-property relationships in polymer fibres with nanosize diameters produced by electrospinning.

.

To investigate the morphology and properties of the polymer nanofibres.

.

To produce fibres at uniform diameters.

Research deliverables (academic and industrial) . .

Nanofibres of uniform diameter. Defined mechanical and physical properties.

.

Process-structure-property relationships.

.

Detailed understanding of the electrospinning process. Nanofibres suitable for applications such as air filtration, protective clothing, fibre reinforced support, and Biomedical.

.

Publications and outputs “Nano fibre and its medical application”, Poster presentation in Research in support of Medicine, Health and Safety Conference, Heriot-Watt university, Scotland, UK.

Galashiels, Scotland, UK Heriot-Watt University, RIFleX, School of Textiles and Design, Netherdale, Galashiels TD1 3HF, UK. Tel: +44 1896 89 2135; Fax: +44 1896 75 8965; E-mail: [email protected] Principal investigator(s): Prof. George K. Stylios Research staff: Liang Luo

Interactive wireless and smart fabrics for textiles and clothing Academic partners

Industrial partners

None None Project start date: September 2002 Project end date: December 2009 Project budget: N/A Source of support: Worshipful Company of Weavers Keywords: Smart, Interactive, Textiles, Garment, Clothing, Sensors, Wireless The last few years have witnessed an increased interest in wearable technologies, smart fabrics and interactive garments. This has come about by certain technological innovations n the areas of sensor-based fabrics, micro devices, wire and wireless networks. In terms of textiles, most of current developments are towards the fashion markets and have resulted in glorifying garments as gimmicky gadgets. However, some efforts are also being directed in using the technology for improving the quality of life, or even for life saving purposes. Examples of such uses can be found in the military, healthcare, fire fighting, etc. This research project investigates new interdisciplinary technologies in fabrics, sensors and wireless computing, for the development of a prototype interactive garment for monitoring various functions of the wearer.

Research register

17

IJCST 22,6

18

Project aims and objectives The general aim of the project is to develop technologies for use in interactive garments, which can provide monitoring functions for various applications such as the clinical or healthcare sector. More specifically, objectives are: .

.

Develop suitable wireless sensors for various measurements, including ECG, temperature, breathing, skin conductivity, mobility and movement, humidity, positioning, etc. Develop a Personal Area Network and a Wireless Communication Centre.

.

Optimise suitable wireless technologies such as Bluetooth to enable communication between sensors and a central processing unit.

.

Conceptualise a smart multilayer fabric. Integrate technologies.

.

Research deliverables (academic and industrial) .

Wireless sensors for physiological and other measurements.

.

Wireless communication centre for relaying information between sensors, wearers, central processing unit and Internet.

.

Conceptual multilayer fabric suitable for interactive garments.

Publications and outputs Stylios, G.K. and Luo L. “A SMART wireless vest system for patient rehabilitation”, Wearable Electronic and Smart Textiles Seminar, Leeds, UK, June 11, 2004. Stylios, G.K. and Luo, L. “Investigating an Interactive Wireless Textile System for SMART clothing”, 1st International Textile Design and Engineering Conference (INTEDEC 2003), Fibrous Assemblies at the Design and Engineering Interface, Edinburgh, UK, September 22-24, 2003. Stylios, G.K. and Luo, L. “The Concept of Interactive, Wireless, Smart Fabrics for Textiles and Clothing”, 4th International Conference, Innovation and Modelling of Clothing Engineering Processes – IMCEP 2003, Maribor, Slovenia, October 9-11, 2003. Stylios, G.K., Luo, L., Chan, Y.Y.F. and Lam Po Tang, S. “The Concept of Smart Textiles at the Design/Technology Interface”, 5th International Istanbul Textile Conference, Recent Advances and Innovations in Textile and Clothing, Istanbul, Turkey, May 19-21, 2005.

Gent, Belgium Universiteit Gent – Ghent University, Vakgroep Textielkunde – Department of Textiles, Technologiepark 907, 9052 Zwijnaarde (Gent), Belgium. Tel: + 32 9 264 5419; Fax: + 32 9 2645831; E-mail: [email protected]

SysTex – coordination action for enhancing the breakthrough of intelligent textile systems (e-textiles and wearable microsystems) Academic partners

Research register

Industrial partners Philips Research, Multitel, Smartex, Anne Demoor bvba, Plastic Electronics, IHofmann

Ghent University, Institut Francais du Textile et de l’Habillement, IMEC, CNR-INFM, University of Pisa, Commissariat a l’Energie Atomique Project start date: 1 May 2008 Project end date: 30 April 2011 Project budget: 800.000 Source of support: FP7 programme Keywords: Intelligent textiles, E-textiles, Wearable electronics, Breakthrough

Wearable electronics embedded in or transformed into textile systems are a new generation of products that contribute to economy as well as to society. SysTex wants to bring partners involved in European projects in this area together in order to group the results of numerous efforts that are currently going on. It wants to expand to national level and to merge textiles and organic electronics. Inter-project agreements must enable a higher level of exchange of knowledge and materials between linked projects. Information on technical and non-technical aspects of RTD and commercialisation of intelligent textile systems will be collected and made available through a web based tool. Training materials will be collected as well as demonstrators that can be used for specialists as well as for a wider public. The project wants to become a single point of contact for all matters related to intelligent textile systems, linking existing initiatives and completing their activities.

Project aims and objectives SysTex aims at developing a framework for current and future actions in research, education and technology transfer in the field of e-textiles and wearable micro systems/electronics in Europe to support the textile industry in the most efficient and effective way to transform into a dynamic, innovative, knowledge-driven competitive and sustainable sector. The objectives are to create an extensive and detailed road map of current and possible future technological developments in e-textiles, to organise and facilitate inter-project contacts, to organise training and exchanges and to identify needs, breakthroughs and bottlenecks. The information this will provide will be spread to the appropriate target groups using appropriate tools. The coordination action will also provide a platform for intelligent textiles, for the partners involved in European projects but also for interested companies and users. A road map will be built based on available road maps. An experienced legal counsellor will analyse contracts, co-operation agreements, etc. and prepare agreements that enable exchange of information and materials between relevant projects without harming protection and exploitation of results. The work programme will include:

19

IJCST 22,6

20

.

analysis of the available and ongoing research activities in e-textiles and wearable microsystems, this will cover activities that should contribute to the progress of smart textile systems. The information should provide an insight in project objectives, status and contacts;

.

production of web tools for disseminating the results of this inventory and analysis;

.

organisation of contacts and exchange of materials, people and expertise;

.

building a joint road map;

.

joint exploitation, dissemination and training of all member parties; and

.

providing legal support for enabling communication and exchanges without hindering protection and exploitation.

Research deliverables (academic and industrial) Collection of detailed information on relevant activities in the area of intelligent textile systems on all levels, including technical as well as non technical aspects. Widely accessible website with user friendly database with full information on activities on intelligent textile systems. Reports on training and education opportunities and collection of information on training material like demonstrators including conditions of use. Organisation of guest lectures, student projects, the annual SysTex Student Award. Distribution of information through the platform like reports on state of the art technology to improve efficiency of research efforts and exploitation of research efforts. Contribution to policy plans of EC and local funding bodies. Templates for inter-project agreements in form of contracts and legal agreements. Increased openness and exchanges between projects and hence faster progress of work.

Publications and outputs The expected outcomes are: .

paper and digital reports/newsletters containing state-of-the-art of research efforts as well as arguments for building future strategies;

.

databases and information systems;

.

inter-project agreements enabling active exchange of project results;

.

organisation of and participation at events;

.

training and education activities;

.

raising awareness in all respective target groups;

.

targeted lectures and training tools;

.

materials like powerpoints, flyer for dissemination;

.

alignment of research; and

.

enhancement of research progress.

Zwijnaarde (Gent), Belgium Ghent University, Department of Textiles, Technologiepark 907, B-9052 Zwijnaarde (Gent), Belgium. Tel: +32 9 264 57 35; Fax: +32 9 264 58 46; E-mail: [email protected] Principal investigator(s): Prof. Dr Ir. Lieva Van Langenhove Research staff: Dr ir. Vincent Nierstrasz ([email protected])

NO BUG: Novel release system and bio-based utilities for insect repellent textiles and garments Academic partners

Industrial partners

Project start date: 15 October 2009 Project end date: 14 October 2013 Project budget: e 523827 Source of support: EUROPEAN COMMISSION, FP7-NMP 2008-SME-2 Keywords: In several applications of professional textiles and clothes, mosquito repellency is an important issue. Two major problems arise: repellents currently in use are harmful, resistance to conventional repellents increases, the lifetime of release systems is too short. Solving these two problems are the main goals of the No Bug project. Novel biorepellents will be considered and evaluated as well as two release systems (multilayer coating and textile bioaggregates) in order to repel mosquitoes causing malaria or dengue. Novel release concepts are multilayer coatings and in situ release of the active compounds. Targeted prototypes are textiles for health workers and bed nets (mosquitoes). The project will study what are the best conditions of use of the biorepellents and how to integrate them in the textile products. Testing, exploitation and dissemination will be an active part of the work.

Project aims and objectives Not available

Research deliverables (academic and industrial) Not available Publications and outputs See https://biblio.ugent.be/input?func¼search; http://textiles.ugent.be/

Zwijnaarde (Gent), Belgium Ghent University, Department of Textiles, Technologiepark 907, B-9052 Zwijnaarde (Gent), Belgium. Tel: +32 9 264 57 35; Fax: +32 9 264 58 46; E-mail: [email protected]

Research register

21

IJCST 22,6

22

Principal investigator(s): Prof. Dr Ir. Lieva Van Langenhove Research staff: Dr Simona Vasile ([email protected])

Multirapier – multirapier technology for woven 3-D fabrics Academic partners

Industrial partners

Project start date: 1 October 2009 Project budget: 101925 Source of support: IWT, Tetra Keywords: N/A

Project end date: 30 September 2011

Market demand for three-dimensional fabrics has recently shown considerable growth. Although many of the 3-D fabrics used are knits, nonwovens or braids, woven 3-D structures can be a superior alternative for certain applications. Multirapier weaving is not yet fully exploited in the construction of woven 3-D fabrics, as most woven 3-D fabrics are presently produced on either looms with single weft insertion or on three-dimensional orthogonal weaving (e.g. 3WEAVEw of 3TEX). Multirapier technology, as applied in face-to-face weaving is both technologically and economically a good alternative for producing 3-D woven structures. Face-to-face looms are simultaneously forming multiple sheds and inserting multiple wefts and enable production of fabrics of up to 5 m width, which is a major advantage in many technical markets. The conversion of face-to-face looms into use as 3D looms is a very important milestone in the development of woven 3-D fabrics. A relatively large capacity of this type of weaving is available in Flanders but the traditional markets, such as upholstery and carpets are steeply declining. This IWTTETRA research project aims at converting existing technology to the technical textiles market. Decorative and technical fabric production techniques, however, are very different and conversion from one to the other requires technical modifications. Aramids, glass or carbon are often used in technical textiles and the correct use of these yarns is a further technological challenge. Their stress-strain behaviour, abrasion on machine parts will certainly involve machine adaptations. As machine modifications are expected, due to the fabric construction and raw material used, the consortium of the ”Multirapier” project includes a face-to-face loom manufacturer. The project will further benefit of the expertise of three RTD partners: the Departments of Textiles of Ghent University and the University College of Ghent as well as the Institut fu¨rTextiltechnik (ITA), Aachen (DE) as partner in the ERASME project with the same topic. SMEs and large companies from Belgium and Germany interested in woven 3-D fabrics as supplier/buyer are further completing the project consortium.

Project aims and objectives Not available.

Research deliverables (academic and industrial) Not available. Publications and outputs See https://biblio.ugent.be/input?func¼search; http://textiles.ugent.be/

Zwijnaarde (Gent), Belgium Ghent University, Department of Textiles, Technologiepark 907, B-9052 Zwijnaarde (Gent), Belgium. Tel: +32 9 264 57 35; Fax: +32 9 264 58 46; E-mail: [email protected] Principal investigator(s): Prof. Dr Ir. Karen De Clerck Research staff: Ing. Lieve Van Landuyt

BIOCOT II: innovative strategies and assays for bioengineered cotton fibres with improved processing and end-user properties (Bayer) Academic partners

Industrial partners

Project start date: 1 November 2009 Project budget: 36,996,989 Source of support: IWT Keywords:

Project end date: 31 October 2012

Cotton is the most important natural fibre: it represents about one third of the total world fibre consumption. Despite the intensive and long-lasting use of cotton fibre for textile applications, several steps of the cotton fibre processing are still inefficient or require large amounts of harsh chemicals. Most of the progress in improving these processes and adding new end-user characteristics to the fibre results from new or modified chemical and enzymatic treatments. Little progress is being made by improving the cotton fibre itself. The development of traits in cotton through genetic engineering is a lengthy and costly process that requires a careful selection of the target traits and approaches to be tested. This project focuses on the first step in this process and determines, within a 2-years time scope, what modifications of the cotton fibre are required to achieve improved functionality in 3 fields: . reactivity and dyeing; . intrinsic wrinkle resistance; and . flame retardancy. It is also important to see whether the introduced modifications do not affect the basic properties of the cotton fibre in a negative way. Therefore, after imparting changes to the fibres, tests were performed with currently available standard cotton tests to get an idea about the general characteristics of the fibres. The tests that are currently available are designed to be performed on fabrics and require at least 100 g to several kilos of material. Therefore, one of the challenges of this project is to develop and optimize small scale tests which are applicable to extremely low amounts of fibre material.

Project aims and objectives Not available.

Research register

23

IJCST 22,6

Research deliverables (academic and industrial) Not available. Publications and outputs See https://biblio.ugent.be/input?func¼search; http://textiles.ugent.be/

24

Zwijnaarde (Gent), Belgium Ghent University, Department of Textiles, Technologiepark 907, B-9052 Zwijnaarde (Gent), Belgium. Tel: +32 9 264 57 35; Fax: +32 9 264 58 46; E-mail: [email protected] Principal investigator(s): Prof. Dr Ir. Karen De Clerck ([email protected])

Advanced characterization of fibre morphology Many of the current and future projects within the Department of Textiles involve some means of characterization of the fibre morphology. The morphology of fibres is in general very different from that of bulk polymer materials due to the high degree of orientation in the fibres. Also fibres often require separate dedicated sampling techniques due to their specific structure both at macro and at micro level. Therefore, a considerable effort has been spent within the Department of Textiles to develop new or optimise existent techniques for the characterization of fibre morphology. The analytical tools used are diverse with the focus being on thermal analysis (thermo-mechanical analysis, differential scanning calorimetry and modulated differential scanning calorimetry), spectroscopy (Fourier transform infrared spectroscopy and microscopy, DRIFTS, ATR, Fourier transform Raman spectroscopy and UV-VIS-NIR spectroscopy) as well as on microscopy (confocal laser scanning microscopy).

Project aims and objectives Not available.

Research deliverables (academic and industrial) Not available. Publications and outputs See https://biblio.ugent.be/input?func¼search; http://textiles.ugent.be/

Zwijnaarde (Gent), Belgium Ghent University, Department of Textiles, Technologiepark 907, B-9052 Zwijnaarde (Gent), Belgium. Tel: +32 9 264 57 35; Fax: +32 9 264 58 46; E-mail: [email protected] Principal investigator(s): Dr Philippe Westbroek ([email protected]), Prof. Dr Ir. Karen De Clerck ([email protected])

Bio-based functional materials from engineered self-assembling peptides (EU-STREP-FP6-2003-NMP-TI-3) This project aims at advancing the science and technology of sustainable and functional materials. Specifically it targets innovative nano-coatings for plastics, metals and ceramic objects, exploiting the self-assembly capabilities of short (25) amino-acid sequences (= peptides) in industrially relevant applications. Self-assembly is a method of spontaneous organization of molecules into higher order structures and defined by a set of boundary conditions (e.g. pH, T, etc). This addresses the need for water-based coating solutions beyond plastics and explores principles of nature for supramolecular structure formation at various surfaces. These ambitions require a work programme addressing the challenges facing peptide-based nanocoatings in terms of their biotechnological engineering at a cost of The contribution of the Department of Textiles is situated on the level of producing peptide-functionalized nano-dimensioned fibres. This results in opening new application possibilities and insight in the behaviour of electrospinning methods during the electrospinning of (bio)polymers and subsequent coating procedures. An indirect impact of this development is the availability of functionalized nano-fibres for the textile industry. Ghent University delivered proof-of-principle for some peptide combinations. However, peptides are short chain molecules and therefore difficult to electrospin.

Project aims and objectives Not available.

Research deliverables (academic and industrial) Not available. Publications and outputs See https://biblio.ugent.be/input?func ¼ search; http://textiles.ugent.be/

Zwijnaarde (Gent), Belgium Ghent University, Department of Textiles, Technologiepark 907, B-9052 Zwijnaarde (Gent), Belgium. Tel: +32 9 264 57 35; Fax: +32 9 264 58 46; E-mail: [email protected]

Research register

25

IJCST 22,6

Principal investigator(s): Prof. Dr Ir. Karen De Clerck ([email protected]), Dr Philippe Westbroek ([email protected]) Research staff: ir. Sander De Vrieze ([email protected])

26

Nanofibres for filter applications: PhD work Electrospinning is a booming technique to produce nonwoven structures from a polymer solution. These nonwoven structures consist of nanofibres. Nanofibres are fibres having a diameter between 1 and 1000 nanometers. Nanofibre nonwovens are nonwovens with very specific properties. These properties can be optimised by a thorough study of the process and the process parameters. The most important parameters are flow rate, used polymer, used solvent, applied voltage, mass percentage of polymer, distance between collector and surface,. . . The obtained structures are applicable in both air and water filtration. According to the parameters, different filters with different cut-off values are obtained. The cut-off value of a filter is the maximum size of the particles going through the filter. This value is directly linked with the diameter of the nanofibres and the tortuosity of the nonwoven. Different polymers are investigated for their electrospinnability and the possible application as a filter. Polymers like cellulose acetate, polysulfone, poly(ethylene-covinyl acetate) are tested. Cellulose acetate nanofibres seem to be the most promising polymers for further research.

Project aims and objectives Not available.

Research deliverables (academic and industrial) Not available. Publications and outputs See https://biblio.ugent.be/input?func ¼ search; http://textiles.ugent.be/

Zwijnaarde (Gent), Belgium Ghent University, Department of Textiles, Technologiepark 907, B-9052 Zwijnaarde (Gent), Belgium. Tel: +32 9 264 57 35; Fax: +32 9 264 58 46; E-mail: [email protected] Principal investigator(s): Prof. Dr Ir. Karen De Clerck ([email protected]) ¨ zgu¨r Ceylan ([email protected]) Research staff: O

Study of cotton fibres modified and developed for high-value applications: PhD work Today’s cotton fibres have developed over the last centuries, with the fibres being longer and stronger than a few centuries ago. Many of these improvements can be attributed to

continuous research and advanced breeding projects. Although quite some work has been done to optimise the mechanical properties, a possible improvement of the intrinsic chemical properties has been lacking behind. In this PhD work, it is investigated what aspects of the chemical behaviour of the cotton fibres would benefit from intrinsic improvements. This is done by relating the fibre properties to demanding end-user applications. Therefore, various methods are to be established to allow the characterisation of the aimed traits on small scale fibrous samples and moreover relate them to large-scale end-user tests. This PhD is performed within the programme of the BIOCOT project.

Project aims and objectives Not available.

Research deliverables (academic and industrial) Not available. Publications and outputs See https://biblio.ugent.be/input?func¼search; http://textiles.ugent.be/

Zwijnaarde (Gent), Belgium Ghent University, Department of Textiles, Technologiepark 907, B-9052 Zwijnaarde (Gent), Belgium. Tel: +32 9 264 57 35; Fax: +32 9 264 58 46; E-mail: [email protected] Principal investigator(s): Dr ir. Philippe Westbroek Research staff: ir. Bert De Schoenmaker ([email protected])

Natural fibre composites with a matrix of a renewable resource: PhD work Composites have many advantages compared to other materials like metals and ceramics. The main advantage of high performance composites consists of their high specific strength and modulus. In order for the transportation industry to diminish its fuel consumption, cars, planes, etc. need to be lightened. This can be done by substituting metals by composites. Nowadays, about 25 weight percent of the airbus A380 consists of composites. The Boeing 787 Dreamliner is pushing the envelope with a total composite fraction of 50% by weight. The main research work of this PhD is to study the possibilities of using resins of renewable resources in engineering composites. Currently, most composites are made out of glass, carbon or aramid fibre and a fossil-based resin. Production, usage and disposal of such composites still have a great impact on the environment. It is better to use green composites, which have a smaller or no impact on the environment. So-called biocomposites are very often CO2-neutral and biodegradable. In this research work, flax is chosen as reinforcement. Possible resins are PLA, PHAs, natural epoxies, polyfurans, . . .

Research register

27

IJCST 22,6

Project aims and objectives Not available.

Research deliverables (academic and industrial) Not available.

28

Publications and outputs See https://biblio.ugent.be/input?func¼ search; http://textiles.ugent.be/

Zwijnaarde (Gent), Belgium Ghent University, Department of Textiles, Technologiepark 907, B-9052 Zwijnaarde (Gent), Belgium. Tel: +32 9 264 57 35; Fax: +32 9 264 58 46; E-mail: [email protected] Principal investigator(s): Prof. Dr Ir. Lieva Van Langenhove Research staff: Prof. Dr Ir. Lieva Van Langenhove

LIDWINE – Multifunctionalized medical textiles for wound (e.g. decubitus) prevention and improved wound healing (“Lidwine”) Academic partners

Industrial partners

Project start date: 1 September 2006 Project end date: 31 August 2010 Project budget: e543.752 Source of support: EUROPEAN COMMISSION – FP6 – IP – SME, FP6 – INTEGRATED PROJECTS – SME Keywords: Decubitus, Electrotherapy, Nanotechnology Several techniques will be developed to prevent decubitus and to stimulate its healing. Examples are passive and active antibacterial action, materials with reduced surface friction, massage and electrotherapy. The latter is the specific task of UGent. The materials will be developed in textile structures. The technologies used will be nanoparticles, encapsulation, brush coatings.

Project aims and objectives Not available.

Research deliverables (academic and industrial) Not available. Publications and outputs See https://biblio.ugent.be/input?func¼search; http://textiles.ugent.be/docs/AnnualReports/2008.pdf

Zwijnaarde (Gent), Belgium Ghent University, Department of Textiles, Technologiepark 907, B-9052 Zwijnaarde (Gent), Belgium. Tel: +32 9 264 57 35; Fax: +32 9 264 58 46; E-mail: [email protected] Principal investigator(s): Prof. Dr Ir. Lieva Van Langenhove Research staff: Prof. Dr Ir. Lieva Van Langenhove

COMPAS – COMPUTER BASED EVALUATION OF ASPECT CHANGE OF CARPETS BY WEAR Academic partners

Industrial partners

Project start date: 1 September 2006 Project end date: 31 August 2009 Project budget: e415.790,32 Source of support: IWT, VIS Keywords: Carpet wear, Evaluation, Image processing The aim of the project is to develop an objective method to be able to quantitatively measure carpet wear in a univocal and accurate way. To this end, images recorded with a colour CCD camera are being processed. New algorithms developed and in development in the framework of other projects will be tested for the evaluation of carpet wear. A considerable obstacle in the research regarding automatic evaluation of carpets is the lack of an extensive basic library of images of good as well as bad samples, and for a large variety of qualities. Establishing such a library is part of the project as well.

Project aims and objectives Not available.

Research deliverables (academic and industrial) Not available. Publications and outputs See https://biblio.ugent.be/input?func¼search; http://textiles.ugent.be/docs/AnnualReports/2008.pdf

Zwijnaarde (Gent), Belgium Ghent University, Department of Textiles, Technologiepark 907, B-9052 Zwijnaarde (Gent), Belgium. Tel: +32 9 264 57 35; Fax: +32 9 264 58 46; E-mail: [email protected]

Research register

29

IJCST 22,6

Principal investigator(s): Prof. Dr Ir. Lieva Van Langenhove Research staff: Prof. Dr Ir. Lieva Van Langenhove

30

Systex: coordination action for enhancing the breakthrough of intelligent textile systems (e-textiles and wearable microsystems) Academic partners

Industrial partners

Project start date: 1 May 2008 Project end date: 30 April 2011 Project budget: e800.000 Source of support: EUROPEAN COMMISSION, FP7 – IST – Coordination and Support Action Keywords: Smart textiles, E-textiles, Wearable micro systems, Wearable electronics The project aims at creating a framework for current and future research activities, technology transfer and education in the area of wearable textile systems and e-textiles. This results in enhancing the cooperation between various entities that contribute to the development and commercialization of the textiles of the future. Technical and non technical information on relevant projects is collected.

Project aims and objectives Not available.

Research deliverables (academic and industrial) Not available. Publications and outputs See https://biblio.ugent.be/input?func¼search; http://textiles.ugent.be/docs/AnnualReports/2008.pdf

Zwijnaarde (Gent), Belgium Ghent University, Department of Textiles, Technologiepark 907, B-9052 Zwijnaarde (Gent), Belgium. Tel: +32 9 264 57 35; Fax: +32 9 264 58 46; E-mail: [email protected] Principal investigator(s): Prof. Dr P. Kiekens Research staff: ir. Els Van der Burght

MUDRA LEARNING NETWORK Academic partners

Industrial partners

Project start date: 1 February 2008 Project budget: e43.394,73

Project end date: 30 September 2009

Source of support: VLAAMSE GEMEENSCHAP, SAMENWERKINGSPROJECTEN VLAANDEREN/CENTRAAL – EN OOST-EUROPA Keywords: Textiles, Networking, Innovation MUDRA Learning Network is a network project of Flemish business and educational partners with their Croatian and Slovenian counterparts. The mentorship methodology offers the framework to a large group of SMEs and entrepreneurs to exchange expertise and to professionalise their management. The know-how from the universities strengthens the technological capacity and helps these companies to innovate. The target group: textile and design-related companies.

Project aims and objectives Not available.

Research deliverables (academic and industrial) Not available. Publications and outputs See https://biblio.ugent.be/input?func¼search; http://textiles.ugent.be/docs/AnnualReports/2008.pdf

Zwijnaarde (Gent), Belgium Ghent University, Department of Textiles, Technologiepark 907, B-9052 Zwijnaarde (Gent), Belgium. Tel: +32 9 264 57 35; Fax: +32 9 264 58 46; E-mail: [email protected] Principal investigator(s): Prof. Dr ir. Karen De Clerck Research staff: Prof. Dr ir. Karen De Clerck

Research into new “sensor materials”: pH-sensitive colorants in textile materials Academic partners

Industrial partners

Project start date: 1 January 2009 Project end date: 31 December 2012 Project budget: e172.000 Source of support: UNIVERSITEIT GENT, BIJZONDER ONDERZOEKSFONDS 2008 Keywords: PH-sensitive dyes, Textiles, Spectroscopy, Microscopy The aim of the project is to obtain a better understanding of the interactions between pHsensitive dyes and textiles. Both a macroscopic spectral analysis and a general microscopic evaluation will be performed. The dye-fiber interactions, the local distribution, the impulse sensitivity and spectral variations as a function of time and place will be looked at.

Research register

31

IJCST 22,6

Project aims and objectives Not available

Research deliverables (academic and industrial) Not available

32

Publications and outputs See https://biblio.ugent.be/input?func¼search; http://textiles.ugent.be/docs/AnnualReports/2008.pdf

Zwijnaarde (Gent), Belgium Ghent University, Department of Textiles, Technologiepark 907, B-9052 Zwijnaarde (Gent), Belgium. Tel: +32 9 264 57 35; Fax: +32 9 264 58 46; E-mail: [email protected] Principal investigator(s): Prof. Dr P. Kiekens Research staff: Prof. Dr Paul KIEKENS, Prof. Dr Ir. K. De Clerck

FRONT – FLAME RETARDANT ON TEXTILE Academic partners

Industrial partners

Project start date: 1 November 2008 Project end date: 31 October 2010 Project budget: e141.000 Source of support: EUROPEAN COMMISSION, FP7 Keywords: Textiles, Flame retardants, Nanoclays The project aim is to introduce finishing products in the European textile market, to produce textile fabrics resistant to fire with high performance and quality, as requested from evolution of legislation and from customer attention. Moreover, flame retardant finishing on textiles would achieve a multifunctional textile composition having not only fire resistance properties, but also an additional functionality. The project proposed by SMEs engaged in the field of polymer processing will contribute to meet needs and to strengthen the competitiveness of the EU manufacturers.

Project aims and objectives Not available.

Research deliverables (academic and industrial) Not available. Publications and outputs See https://biblio.ugent.be/input?func¼search; http://textiles.ugent.be/docs/AnnualReports/2008.pdf

Zwijnaarde (Gent), Belgium Ghent University, Department of Textiles, Technologiepark 907, B-9052 Zwijnaarde (Gent), Belgium. Tel: +32 9 264 57 35; Fax: +32 9 264 58 46; E-mail: [email protected] Principal investigator(s): Prof. Dr Ir. Karen De Clerck Research staff: Prof. Dr Ir. Karen De Clerck

Tapijtfabriek ALFA: water absorbing capacity of hollow fibres for filling yarn in artificial turf Academic partners

Industrial partners

Project start date: 1 March 2008 Project end date: 28 February 2009 Project budget: e 28.050 Source of support: IWT, KMO-innovatiestudie Type 3 Keywords: Hollow fibre – water absorption – PP This project was conducted together with ALPHA-carpets. The capacity of water absorption of hollow PP-fibres was examined and improved, in order to optimize the quality of actual products such as bath maths and towels, but also to be able to bring new products on the market in the future. In the framework of this project, new methods have been written to quantify the water absorption and the drying of hollow fibres. More understanding was gained in the effect of the extrusion parameters on the fibre diameter and the capacity of water absorption of hollow fibres.

Project aims and objectives Not available.

Research deliverables (academic and industrial) Not available. Publications and outputs See https://biblio.ugent.be/input?func¼search; http://textiles.ugent.be/docs/AnnualReports/2008.pdf

Zwijnaarde (Gent), Belgium Ghent University, Department of Textiles, Technologiepark 907, B-9052 Zwijnaarde (Gent), Belgium. Tel: +32 9 264 57 35; Fax: ++32 9 264 58 46; E-mail: [email protected] Principal investigator(s): Prof. Dr P. Kiekens Research staff: Johanna Louwagie; Johanna Louwagie

Research register

33

IJCST 22,6

TRITex: TRANSFER OF RESEARCH AND INNOVATIONS IN TEXTILES Academic partners

34

Industrial partners

Project start date: 1 January 2009 Project end date: 31 December 2012 Project budget: e 490.948,42 Source of support: INTERREG IV, EFRO Keywords: Research, Innovation, Textiles Ensait and Ghent University – Department of textiles want to establish common and complementary actions in order to strengthen the cross border cooperation. 4 actions are foreseen: multilateral research programmes, development of modules for distance learning, valorisation of digital training modules at industrial partners, organisation of seminars.

Project aims and objectives Not available.

Research deliverables (academic and industrial) Not available. Publications and outputs See https://biblio.ugent.be/input?func¼search; http://textiles.ugent.be/docs/AnnualReports/2008.pdf

Zwijnaarde (Gent), Belgium Ghent University, Department of Textiles, Technologiepark 907, B-9052 Zwijnaarde (Gent), Belgium. Tel: +32 9 264 57 35; Fax: +32 9 264 58 46; E-mail: [email protected] Principal investigator(s): Prof. Dr P. Kiekens Research staff: Dr Ir. Vincent Nierstrasz

BIOTIC: Biotechnical functionalization of (bio)polymeric textile surfaces Academic partners

Industrial partners

Project start date: 1 April 2008 Project end date: 31 March 2010 Project budget: e 223.288,66 Source of support: EUROPEAN COMMISSION, PEOPLE MARIE CURIE ACTIONS (Intra-European Fellowships (IEF)) Keywords: Biotechnology, Enzymes, Grafting, Functionalisation, Surface modification, Textiles, Biopolymer, Polymer, Nano-structuring

The aim is to functionalise textile materials using biotechnology. The research will be based on a concerted multi-disciplinary approach, thereby creating the possibility to produce functionalised materials with unique properties and functionalities. The research will focus on enzymatic grafting of functional groups on textile fibres, and specific enzymatic surface modification to obtain functional nano-structured surfaces.

Project aims and objectives

35

Not available.

Research deliverables (academic and industrial) Not available. Publications and outputs See https://biblio.ugent.be/input?func¼search; http://textiles.ugent.be/docs/AnnualReports/2008.pdf

Zwijnaarde (Gent), Belgium Ghent University, Department of Textiles, Technologiepark 907, B-9052 Zwijnaarde (Gent), Belgium. Tel: +32 9 264 57 35; Fax: ++32 9 264 58 46; E-mail: [email protected] Principal investigator(s): Prof. Dr P. Kiekens Research staff: ir. Els Van der Burght

SNAP – production of imidised styrene-malein acid (SMI) nanoparticles and surface interactions with different types of substrates (TOPCHIM) Academic partners

Industrial partners

Project start date: 1 September 2008 Project end date: 31 August 2011 Project budget: e 249.525 Source of support: IWT, Onderzoeksproject Keywords: SMI nanoparticles, Imidisation, Surface properties This project aims for a better understanding of the chemistry and physics at the nano level of SMI nanoparticles, which are created by imidisation of copolymer styrene-maleic anhydride (SMA). The main aspect is to broaden insight into the physical characteristics of the nanoparticles such as shape, size and uniformity. Interaction with minerals and substances from renewable sources is another focus.

Project aims and objectives Not available.

Research deliverables (academic and industrial) Not available.

Research register

IJCST 22,6

36

Publications and outputs See https://biblio.ugent.be/input?func¼search; http://textiles.ugent.be/docs/AnnualReports/2008.pdf

Zwijnaarde (Gent), Belgium Ghent University, Department of Textiles, Technologiepark 907, B-9052 Zwijnaarde (Gent), Belgium. Tel: +32 9 264 57 35; Fax: +32 9 264 58 46; E-mail: [email protected] Principal investigator(s): Prof. Dr Ir. Lieva Van Langenhove Research staff: Ing. Johanna Louwagie; Ing. Johanna Louwagie

Bexco: Arctic ropes – research into the dynamic behaviour of synthetic ropes in extremely cold circumstances Academic partners

Industrial partners

Project start date: 1 July 2008 Project end date: 31 October 2009 Project budget: e 21.499,46 Source of support: IWT, KMO-innovatiestudie type 3 Keywords: Dynamic behaviour, Synthetic ropes The purpose of the study is to obtain insight in the phenomena that occur in frozen ropes that are under cyclic load an in the effect of the phenomena on their mechanical properties, fatigue and life span of the rope. The influence of ice will be studied on the macro- and microscopic level on ropes after cyclic loading in frozen conditions.

Project aims and objectives Not available.

Research deliverables (academic and industrial) Not available. Publications and outputs See https://biblio.ugent.be/input?func¼search; http://textiles.ugent.be/docs/AnnualReports/ 2008.pdf

Zwijnaarde (Gent), Belgium Ghent University, Department of Textiles, Technologiepark 907, B-9052 Zwijnaarde (Gent), Belgium. Tel: +32 9 264 57 35; Fax: +32 9 264 58 46; E-mail: [email protected]

Principal investigator(s): Prof. Dr Ir. Lieva Van Langenhove Research staff: Simona Vasile

Research register

NIRIS: new insertion rules for a new insertion system (Picanol) Academic partners

Industrial partners

Project start date: 1 December 2008 Project end date: 30 November 2011 Project budget: e 840.813,28 Source of support: IWT, Onderzoeksproject Keywords: Weft preparation system, Air jet loom, Speed increase The aim of this project is to ultimately bring a new weft preparation system on the market. This should allow to insert the weft yarn faster on an air jet loom. In concrete figures, a speed increase on the loom of 10 to 15% is set as goal.

Project aims and objectives Not available.

Research deliverables (academic and industrial) Not available. Publications and outputs See https://biblio.ugent.be/input?func¼search; http://textiles.ugent.be/docs/AnnualReports/2008.pdf

Zwijnaarde (Gent), Belgium Ghent University, Department of Textiles, Technologiepark 907, B-9052 Zwijnaarde (Gent), Belgium. Tel: +32 9 264 57 35; Fax: +32 9 264 58 46; E-mail: [email protected] Principal investigator(s): Prof. Dr Ir. Karen De Clerck Research staff: Dr Philippe Westbroek

Advanced water filtration with nanofibres (Hogeschool West-Vlaanderen) Academic partners

Industrial partners

Project start date: 1 October 2008 Project end date: 30 September 2010 Project budget: e 103.000 Source of support: IWT, TETRA-fonds Keywords: Waterfiltration, Nanofibers, MBR

37

IJCST 22,6

The goal of the project is the evaluation of nanofiber nonwovens produced by electrospinning for usage in advanced waterfiltration. This project is a continuation of a previous project around nanofibers. The focus is put on the research in the MBRtechnology.

Project aims and objectives

38

Not available.

Research deliverables (academic and industrial) Not available. Publications and outputs See https://biblio.ugent.be/input?func¼search; http://textiles.ugent.be/docs/AnnualReports/2008.pdf

Zwijnaarde (Gent), Belgium Ghent University, Department of Textiles, Technologiepark 907, B-9052 Zwijnaarde (Gent), Belgium. Tel: +32 9 264 57 35; Fax: +32 9 264 58 46; E-mail: [email protected] Principal investigator(s): Prof. Dr Ir. Lieva Van langenhove; Prof. Vanfleteren, Prof. Leman Research staff: Hertleer Carla; Hertleer Carla

UGent Mobile Textiles: a travelling compact, attractive and interactive platform demonstrating the multidisciplinary research at Ghent University that contributes to the textiles of the future Academic partners

Industrial partners

Project start date: 1 January 2009 Project end date: 31 December 2011 Project budget: e 95.100 Source of support: UNIVERSITEIT GENT, Werkgroep Wetenschapscommunicatie en – popularisering: Projecten Wetenschap en maatschappij Keywords: Smart textiles, Demonstrators, Textile antennas, Music, Stretchable electronics Textiles of the future are smart. This emerging research area has a large economical potential and a huge social relevancy. Four departments of Ghent University join forces to develop demonstrators that exhibit research on smart textiles at UGent. In a mobile stand, the visitor is given the opportunity to acquaint with these new technologies.

Project aims and objectives Not available.

Research deliverables (academic and industrial) Not available.

Research register

Publications and outputs See https://biblio.ugent.be/input?func¼search; http://textiles.ugent.be/docs/AnnualReports/2008.pdf

39

Hong Kong, China The Hong Kong Polytechnic University, QT715, Institute of Textiles and Clothing, Hung Hom, Kowloon, Hong Kong. Tel: +00852-27666490; Fax: 00852-27542521; E-mail: [email protected] Principal investigator(s): Prof. Tao Xiaoming Research staff: Hua Tao, Zhu Bo, Wang Yangyong, Li Xinsheng, Sun Shaomin, Shu Lin, Yi Weijin, Chow Michelle, Li Qiao

Fabric sensors for three dimensional surface pressure mapping Academic partners

Industrial partners

The Hong Kong Research Institute of Textiles and Apparel Limited

Sun Hing Industries Holding Limited, Hong Kong, Wide Source Technology Group Limited, Hong Kong, TAL Apparel Limited, Hong Kong Project end date: 31 August 2010

Project start date: 1 December 2007 Project budget: HK$8,025,400 Source of support: Innovation and Technology Commission, HKSAR, (ITP/034/07TP) Keywords: Fabric pressure sensor Fabric pressure sensors are ideal candidates for measuring pressure on three dimensional surfaces, which is promising for applications in functional wear and building maintenance. The project aims to further develop pilot production technologies of fabric sensors for measuring pressure on three dimensional complex surfaces. The project will also explore applications of such fabric sensors. The technology will create a new market of high valueadded products for traditional textile and apparel industries.

Project aims and objectives .

To optimize the materials systems for the fabric sensors meeting the application requirements.

.

To optimize the structural and aesthetic design for the sensors and packages based on electro-mechanical analysis.

.

To develop pilot fabrication technology and equipment for the fabric sensors. To explore various packaging production technologies for the fabric sensors.

.

IJCST 22,6

.

.

40

To establish a testing protocol for the sensors, regarding to directional sensitivity, effects of temperature, humidity, UV and water, fatigue, aging, measurement reliability. To develop prototype fabric pressure sensors for functional wear and building maintenance.

Research deliverables (academic and industrial) .

The optimised design and material systems for fabric pressure sensors in terms of performance and cost.

.

Pilot fabrication technology and equipment for the fabric pressure sensors.

.

Packaging method and production technology of the fabric pressure sensors. A testing protocol for performance and reliability of the fabric pressure sensors.

. .

Product prototypes based on the fabric pressure sensors.

Publications and outputs Lin Shu, Tao Hua, Yangyong Wang, Qiao Li, David Dagan Feng and Xiaoming Tao, “In-Shoe Plantar Pressure Measurement and Analysis System based on Fabric Pressure Sensing Array”, IEEE Transactions on Information Technology in Biomedicine, Vol. 14 No. 3, 2010. Tao, X.M., Wang, Y.Y., Hua, T., Zhu, B. and Li, Q., Soft Pressure Sensing Device, US Patent: US 12/712, 123, 2010. Zhu, B. and Tao X.M., “Mechanical analysis and foundation design of soft pressure sensor”, The 86th Textile Institute World Conference Proceedings, Hong Kong, November, 2008.

Hong Kong, China The Hong Kong Polytechnic University, QT715, Institute of Textiles and Clothing, Hung Hom, Kowloon, Hong Kong. Tel: +00852-27666490; Fax: 00852-27542521; E-mail: [email protected] Principal investigator(s): Prof. Tao Xiaoming Research staff: Wang Guangfeng, Zhang Zhifeng, Zheng Wei, Ying Diqing, Zhang Chi, Hui Chi-Yuen, Mai Kaiying, Shen Jing

Small sized fiber sensors Academic partners

Industrial partners

The Hong Kong Research Institute of Textiles and Apparel Limited

Hong Kong Tak Ying Trading Company, Pool Heng Company Limited, Best Technology Company Limited, Esquel Enterprise Limited Project end date: 31 December 2010

Project start date: 1 March 2008 Project budget: HK$5,463,600

Source of support: ITF project Keywords: POF, Sensor, FBG, Large deformation In applications such as medical devices, industry, robotics and wearable electronics, the size of sensors is a very important parameter. Fibers of a few microns in diameter are ideal candidates. Under the Guangdong-Hong Kong Technology Cooperation Funding Scheme, the present project is aimed to develop interactive intelligent textile material. Electrically conductive fibers and fiber assemblies and polymeric optical fiber will be developed for the small sized fibers sensors. The sensors to be developed should have high environment stability, long service life, reasonable material and production cost. The proposed project will investigate and develop design and fabrication technologies, as well as the package of single cell of fiber sensors, and explore their applications in industries.

Project aims and objectives .

To select and fabricate materials for the small size fiber sensors.

.

To make appropriate fibers. To develop the structural design for the small size fiber sensors.

. . .

To develop fabrication technology and equipment for the fiber sensors. To explore package technologies for the fiber sensors.

Research deliverables (academic and industrial) .

The optimized design and materials for the small size fiber sensors in term of performance and cost.

.

Fabrication technology and equipment for the small size fiber sensors. Package method and production technology of the small size fiber sensors.

. . .

A test protocol for performance and reliability of the small size fiber sensors. The theoretical model for the single cell of the fiber sensors.

Publications and outputs Xiaoming Tao, Wei Zheng, Xiao-hong Sun, Guangfeng Wang, Chi-Yuen Hui, Fiber loop strain sensor for large deformation, US Patent Application No. 61/212,451.

Hong Kong, China The Hong Kong Polytechnic University, QT715, Institute of Textiles and Clothing, Hung Hom, Kowloon, Hong Kong. Tel: 00852-27666490; Fax: 00852-27542521; E-mail: [email protected] Principal investigator(s): Prof. Tao Xiaoming Research staff: Zhu Bo, Hua Tao, Wang Yangyong, etc.

Research register

41

IJCST 22,6

In-situ multi-parameter evaluation system for smart protective apparel under high-speed impact Academic partners

42

Industrial partners

Shenzhen NanHua Electronic Technology Nanjing University of Science and Co., Ltd, Joint Sensor Instruments Technology, Suzhou Institute of (Hong Kong) Limited, TAL Apparel Nano-tech and Nano-bionics, Chinese Limited Academy of Science Project start date: 2 August 2010 Project end date: 31 July 2012 Source of support: The Innovation and Technology Fund, Hong Kong Keywords: In-situ, Multi-parameter, Smart, Impact The modern society has an increasing demand for smart protective apparel against impact from contact sports, traffic, and ballistic force, etc. However, so far there has been no measuring and sensing technology available for flexible textile materials with large deformation. Based on the previous achievements by the applicants, the present project is aimed to develop a built-in and in-situ multi-parameter evaluation system to be integrated with protective apparel. By real-time obtaining and analysis of spatial and temporal distributions of strain and pressure inside protective apparel during high-speed impact loading, the system can provide information and intelligence to help train sportsmen and reduce possible injuries to human body, to timely detect injury on human body and wirelessly transmit the situation to related surgeons, and to evaluate performances of protective textiles and apparel. On one hand, the project targets developing a platform technology of smart protective apparel with a reasonable cost and ensured performance quality. On the other hand, the technology will also fill a gap in the field of high-speed impact protection as well as corresponding product market.

Project aims and objectives .

To measure and confirm the dynamic response of the fabric strain and pressure sensors according to the requirement of high-speed impact applications.

.

To conduct mathematical modeling of the fabric sensors and the evaluation system according to the results of dynamic tests.

.

To explore and optimize the configuration, connection, and materials for the integration of the evaluation system with smart protective apparel.

.

To develop pilot fabrication technology and equipment for the evaluation system as well as smart protective apparel.

.

To establish testing protocols for the fabric sensors, the evaluation system, and the smart protective apparel.

.

To develop prototypes of the evaluation system, and make necessary modifications according to real impact test.

.

To design and construct prototypes of smart protective apparel with integrated evaluation system, and evaluate the related performances.

.

To conduct data processing/analysis, and provide relevant information of impact, injury situation on human body, or technical specifications of protective apparel.

Research register

Research deliverables (academic and industrial) .

Pilot fabrication technology and equipment for the smart protective apparel and the evaluation system.

.

Integration and packaging technology of the evaluation system for the smart protective apparel.

.

Testing protocol for the performances of the evaluation system and the smart protective apparel. Application prototypes of smart protective apparel integrated with the evaluation system with acceptable performances and reasonable cost. The methods of signal acquisition, storage, transmission and data analysis, as well as the evaluation methods of impact information, injury of human body, and performances of protective apparel.

.

.

Publications and outputs Seminars/exhibitions – to disseminate and publicize the findings of the project to the textile and apparel industry. Papers/booklets/patents – to publish papers, present the know-how to interested parties, and patent the technology/products. Internet – to provide a readily-available platform for the easy access of the public. Trial prototypes – to provide prototypes to interested companies/customers for trial purpose. Commercialization – to transfer the technologies to the industry for mass production.

Ithaca, New York, USA Cornell University, Department of Fiber Science and Apparel Design, 243 MVR Hall. Tel: +607-255-1929; Fax: 607-255-1093; E-mail: [email protected] Principal investigator(s): Hwa Kyung Song, Susan P. Ashdown Research staff: Catherine Devine

CATEGORIZATION OF LOWER BODY SHAPES BASED ON MULTI VIEW ANALYSIS, AND DEVELOPMENT OF AUTOMATED CUSTOM-MADE PANTS DRIVEN BY BODY SHAPE Academic partners

Industrial partners

Project start date: 2009 January Project budget: Source of support:

Project end date: 2010 December

43

IJCST 22,6

44

Keywords: Automated made-to-measure system, Sizing, Body shape analysis, 3D body scanning The objectives of this study are to categorize lower body shapes and postures based on multi view analysis, and to improve pants fit of automated custom-fitted patterns using base patterns that are balanced and corrected for each body shape. As methods for body classification, principal component analysis and cluster analysis will be conducted based on 2682 females aged 18 to 35 from SizeUSA data, a national anthropometric survey data (2003). Once the body shapes are identified, base pants patterns for each body shape group will be developed, using as a starting point a pattern from a current apparel company (Land’s End). For each body shape group, four subjects with hip girths in the 50th percentile range (between 38” and 42”) will be recruited for refinement and fitting of the base patterns. Custom-made pants will be fitted on subjects examining and correcting the fit of the pants in four postures (standing, sitting on a chair, walking, and stepping), and the alterations will be transferred to the original patterns. By comparing pants patterns within each group and among groups, and by comparing each subject’s measurements to the measurements of the central subject of each body shape group, one subject from each group will be selected to represent the group as a whole, and to be the fit model for the base pattern for this group. After a final fitting, the final block patterns for each group will be confirmed. Subjects will fill out a questionnaire with general demographic questions, questions about body satisfaction, body shape perception and your fit satisfaction with ready-to-wear pants. The hypothesis of this study is that optimal customization could occur if customization starts from the most correctly shaped and balanced garment for each customer’s figure type. As a validation process, additional ten subjects for each group will be recruited to compare appearance and fit satisfaction between two pairs of pants, one created from an automated custom-made system using a single base pattern (Standard customization method) and the other created from the same automated custom-made system using the base patterns created for the different body shapes (Body shape driven customization method). Subjects’ sizes will not be controlled at this stage except that all subjects will have less than 48.57” hip girth measurement (90th percentile of the hip girth distribution in SizeUSA). Subjects will answer questions about body cathexis, body shape perception and satisfaction with ready-to-wear fit. Additionally, they will assess the fit of two sets of pants after four postures and choose which of the two pair of pants fits best. Expert fit judges will also perform visual analysis of the fit of the pants, using photographs or avi files of the subjects, and will rank the fit of the two pair of pants for each subject in a blind test.

Project aims and objectives The objectives of this study are to categorize lower body shapes and postures based on multi view analysis, and to improve pants fit of automated custom-fitted patterns using base patterns that are balanced and corrected for each body shape.

Research deliverables (academic and industrial) Papers will be published in peer reviewed journals.

Publications and outputs Song, H.K. and Ashdown, S.P., “Categorization of lower body shapes based on multiple view analysis”, Textile Research Journal (accepted). Song, H.K. and Ashdown, S.P., “Investigating the validity of visual fit assessment from 3D scans”, Clothing and Textiles Research Journal (in press).

45

I˙zmir, TURKEY Dokuz Eylu¨l University, Department of Textile Engineering, Tınaztepe Campus, 35160 Buca/Izmir/TURKEY. Tel: +90-232-4127211; Fax: +90-232-4127210; E-mail: [email protected] Principal investigator(s): Dr Vildan SULAR Research staff: Gonca BALCI

Factors affecting yarn friction Academic partners

Research register

Industrial partners

Project start date: May 2008 Project end date: May 2010 Project budget: 47000EURO Source of support: Dokuz Eylul University Keywords: Yarn friction, Normal load, Contact area, Friction coefficient, Friction force Yarn friction properties is one of the important properties affecting production stages such as yarn production, fabric and garment formation and also the importance of friction properties continues till the end of the life of a textile product. Sometimes friction property is needed although it is undesirable for some processes. Yarn-to-different surfaces (metal, ceramic, etc.) friction, friction between and within yarns play a great role in winding, weaving, knitting and also sewing. Furthermore, the magnitude of friction of textile materials affects most of textile processing. For this reason to investigate yarn friction properties is very important. In this research, frictional properties of yarns produced having different structural properties (fibre composition, linear density, twist, production system) and different test parameters (normal load, contact area, etc.) will be investigated. Friction coefficients and frictional forces of the yarns will be compared for different conditions and factors affecting these properties will be examined in a systematic way.

Project aims and objectives Friction is an important property affecting quality, productivity and performance of a product and it is also important for textiles from fiber to finished product. Although there are a lot of researches on frictional properties of textiles about the nature of friction, its impact on textile processing and its role in determining yarn and fabric properties, studies are still going on this topic because there are a lot of parameters affecting friction and to get under control these properties is difficult. This research aims

IJCST 22,6

to determine yarn frictional properties of the yarns having different characteristics and especially yarns made of new fibers are planned to examine. In the context of this research, also test parameters will be examined.

Research deliverables (academic and industrial)

46

A large database containing different yarn characteristics and yarn frictional properties will be developed at the end of study. This database may be useful for researchers and also friction results may be useful for yarn producers, woven or knitted fabric producers. Relationships between yarn parameters and friction results will be examined and furthermore frictional properties of new fibers and comparison between different yarn types will be given for different test conditions. Publications and outputs The research has began on May 2008.

I˙ZMI˙R, TURKEY Dokuz Eylu¨l University, Department of Textile Engineering, Tınaztepe Campus, 35160 Buca/Izmir/TURKEY. Tel: +0090 232 4127211; Fax: 0090 232 4127210; E-mail: [email protected] Principal investigator(s): AYSUN CI˙RELI˙ AKS¸I˙T ¨ SE ¨ GE KO Research staff: BENGI˙ KUTLU, NURHAN ONAR, MU

USAGE OF PLASMA TECHNOLOGY FOR DEVELOPMENT OF CONDUCTIVITY PROPERTIES OF TEXTILE MATERIALS Academic partners

Industrial partners

Project start date: May 2008 Project end date: May 2010 Project budget: e 40000 ¨ NI˙ERSITY (BAP) Source of support: T.C. DOKUZ EYLU¨L U Keywords: Plasma polymerization, Textile, Conductivity The surface structure of fibres is very important in processing and use, since friction, abrasion, wetting, adhesion, adsorption and penetration phenomena as well as antistatic behaviour are involved. In order to obtain textile materials with a desired property, the fibre surface is often modified with polymer, inorganic or hybrid organic/inorganic layers before use. The demand for electrically conductive fibres and textiles has increased in recent years because of applications as antistatic materials, sensors, materials for electromagnetic shielding and biomedical use. However, an ideal method for modification remains to be found for the preparation of stable conductive textiles.

We want to give conductive feature on the textile materials with the use of plasma technology. In plasma processing technology, it is well established that exposure to plasmas generated in inert gases and/or reactive gases can clean the surface of materials and modify their characteristics, particularly their surface energy. Active species from the plasma bombard and/or react with monolayers on the surface of materials and change their surface properties either temporarily or permanently. Such work includes metals or polymers of interest in many industries and less commonly, textiles. Plasma technology applied to the treatment of textiles has developed markedly during the past decade, due to its potential environmental and energy conservation benefits, in developing high-performance materials for the world market and. In practice, the surface properties of natural and synthetic fibres or filaments can be modified using plasma treatment. This can lead to processes such as polymerisation, grafting, crosslinking, etc. with concomitant effects on wetting and wicking, dyeing, printing, surface adhesion, electrical conductivity and other characteristics of interest in the textile industry. Since adhesion is a surface-dependent property, mediated at a molecular scale, plasma technology can effectively achieve modification of this near-surface region without affecting the bulk properties of the materials of interest. Like polyaniline, polypyrrole conductive chemicals which compose a conductive thin film on textile’s surface is covered in this project. For furnishing conductivity on textiles this project needs plasma machine which work with low pressure, have RF (radio frequency) plasma (10-100 W and 30-360 s working power and time) with heat system (25-95
C).

Project aims and objectives Aim of this project is, composing conductive thin film on textile’s surface with conductive chemicals like polyaniline, polypyrrole by plasma technology.

Research deliverables (academic and industrial) To obtain conductivity on textile materials.

I˙ZMI˙R, TURKEY Dokuz Eylu¨l University, Department of Textile Engineering, Tınaztepe Campus, 35160 Buca/Izmir/TURKEY. Tel: +0090 232 4127211; Fax: 0090 232 4127210; E-mail: [email protected] Principal investigator(s): AYSUN CI˙RELI˙ AKS¸I˙T ¨ SE ¨ GE KO Research staff: NURHAN ONAR, UMI˙T HALI˙S ERDOAN, MU

PRODUCING CONDUCTIVE FIBRE AND DEVELOPING OF THEIR PROPERTIES Academic partners

Industrial partners

Project start date: 07/2008 Project budget: e 30000

Project end date: 07/2010

Research register

47

IJCST 22,6

48

Source of support: T.C. DOKUZ EYLUL UNIVERSITY (BAP, Scientific Research Project) Keywords: Fibre, Textile, Conductivity The multifunctional textiles for leisure clothing are required while progressing of technology. Increasing of functional properties of textile materials provided useable of textile materials in various areas. For example, intelligent textiles intensively attract the interest in the world. Conductivity textiles are the part of intelligent textiles. Conductive textiles can be used from data transferring to electromagnetic shielding, antistatic properties, heating element and sensors, so on. In the project, we will produce the powder of polyaniline and polypyrrole polymer as conductive polymer. Then we will add the powders to the melting during spinning. What’s more, we will coat the fabric with polyaniline and polypyrrole film.

Project aims and objectives In the project, we aimed to produce conductive textiles by coating with conductive polymers by chemical oxidative process and by melt spinning the textile fibers by doping of conductive polymer powders produced by chemical oxidative polymerization. To the aim, we will produce polyaniline, polypyrrole polymers. What’s more, we will characterized the samples by using FTIR, DTA-TG, XRD and SEM.

Research deliverables (academic and industrial) To occur conductivity on textile materials by chemical oxidative polymerisation method. Publications and outputs Aks ¸ it, A.C., Onar, N., Ebeoglugil, M.F., Kayatekin, I., C¸elik, E., O¨zdemir, I., Nurhan Onar, M., Electromagnetic and Electrical properties of coated Cotton fabric with Barium Ferrite doped Polyaniline Film, APP-2008-06-1917, 2008, Journal of Applied Polymer Science (submitted). ¨ zdemir, I., Conductivity and Magnetic Onar, N., Aks ¸ it, A., Ebeoglugil, M.F., Birlik, I., C¸elik, E., O properties of coated fabrics with Barium Ferrite doped Aniline Solution, III. International Technical Textiles Congress, 1-2 December 2007, Istanbul Fair Center, Yesilkoy/Istanbul, pp. 198-206 (oral presentation). Onar, N., Aksit, A., Avgin, I., Celik, E., Ebeoglugil, M.F., Kayatekin, I. and Ozdemir, I., “Magnetic properties of coated fabrics with Barium Ferrite doped Silica Sol”, 10th International Conference and Exhibition of the European Ceramic Society, June 17-21, 2007, Berlin (oral presentation).

IZMIR, TURKEY Dokuz Eylu¨l University, Department of Textile Engineering, Tınaztepe Campus, 35160 Buca/Izmir/TURKEY. Tel: +902324127211; Fax: +902324127210; E-mail: [email protected]

Principal investigator(s): Prof. Dr Arif Kurbak Research staff: Tuba Alpyıldız

Research register

Studies on textile composites Academic partners

Industrial partners

Project start date: December 2007 Project end date: December 2010 Project budget: e30000 Source of support: Dokuz Eylul University Keywords: Knit, Composite, Impact, Reinforcement Textile preforms are to be investigated and improved as reinforcements in the structural composite materials. Among the textile preforms, knitted fabrics can be formed, with considerably low costs, into almost every possible shape by making use of their extensional deformability offering the advantage that a more homogeneous fibre content is achieved over the entire surface of the part, and also at points of strong curvature. In the composite materials reinforced by knitted fabrics, fibre orientation distribution in the composite is determined by the knit structure and does not change significantly during the production of the composite. In this study different knitted structures will be investigated and the structure with sufficient mechanical properties will be indicated.

Project aims and objectives The major aim of this project is to carry on the studies of the improvements of the textile preforms and to manufacture a knitted reinforcement which is light in weight and has adequate resistance against impact.

Buca/Izmir, TURKEY Dokuz Eylu¨l University, Department of Textile Engineering, Tınaztepe Campus, 35160 Buca/Izmir/TURKEY. Tel: 00902324127211; Fax: 00902324127210; E-mail: [email protected] Principal investigator(s): Prof. Dr Ays¸ e OKUR Research staff: Res. Ass. Musa KILIC¸

Analyses of unevenness and hairiness on blended yarns Academic partners

Industrial partners

Project start date: May 2008 Project end date: May 2010 Project budget: 52000 Euro Source of support: Dokuz Eylu¨l University Keywords: Blended yarns, Hairiness, Unevenness, Modal, Tencel, Promodal

49

IJCST 22,6

50

Yarns made of regerated cellulosic fibres such as modal, tencel, promodal and blends of these with cotton have wider use especially in recent years. However, when the literature is checked, it is found that there are not adequate scientific researches on unevenness and hairiness of these kinds of yarns. So, in this project it is aimed to analyze the unevenness and hairiness of blended yarns made of cotton/modal, cotton/tencel and cotton/promodal. In the project, the effect of linear density and twist, the effect of blend type and the effect of spinning system on hairiness and unevenness will be analyzed by using yarns made of 100% cotton, 67%-33%, 50%-50%, 33%-67% cotton-regenerated cellulosic fibre and 100% regenerated cellulosic fibre. Also, performance properties of fabrics made of these yarns will be analyzed within the context of the project.

Project aims and objectives In this project it is aimed to analyze the unevenness and hairiness of the yarns statistically, determine the relationships between unevenness-hairiness and offer a more realistic formula to calculate the limit irregularity of these blended yarns. It is hoped that the results of the project which has a fairly wide experimental study will be useful for practical uses.

Research deliverables (academic and industrial) The effect of blend ratio, linear density, twist, blend type, spinning system on hairiness and unevenness of blended yarns will be observed at the end of the study. Also, relationships between hairiness and unevenness of these yarns will be analyzed. Within the context of the project, performance properties of the fabrics made of these yarns will be analysed. A new formula is aimed to be derived from the statistical analyses for the limit irregularity of the blended yarns. Furthermore, different measuring principles for the yarn hairiness will be compared. It is thought that this project will contribute to the further studies on hairiness and unevenness of blended yarns made of regenerated cellulosic fibres with cotton.

Loughborough, UK Environmental Ergonomics Research Centre, Loughborough Design School, Loughborough University, Loughborough, UK LE11 3TU. Tel: 01509 223031; Fax: 01509 223940; E-mail: [email protected] Principal investigator(s): Prof. George Havenith Research staff: Sarah Davey, Victoria Richmond, Katy Griggs

Protective responsive outer shell for people in industrial environments (PROSPIE) Academic partners

Industrial partners

TNO, EMPA, Lithuanian Textile Institute Foritas; JSC Pakaita; I.O.C.P Spa; Capzo; Humanikin; Ergonsim; Merford Cabins; IFAK; D’Appolonia Spa;

VanHoutte Consulting; Bel-Confect; Palemono Keramika AB. Project end date: November 30, 2012

Project start date: Dec 1, 2009 Project budget: 3.66 million euro Source of support: European Union (Nanotechnologies, Materials and New production). Keywords: Protective clothing, Physiological load, Sensors, Cooling systems In the Prospie-project a new generation of personal protective equipment (PPE) will be developed and produced. The special feature of the PPE will be a dynamic cooling system that prevents the worker to become hyperthermic. Although sweat evaporation is an excellent cooling mechanism for work in the heat, this system is compromised when working in protective clothing. The body temperature rises and consequently the vigilance and task performance decrease. Eventually the worker has to abandon his task due to incompensable heat strain. Prospie aims to supply the worker with personal protective equipment that enables him or her to work longer in protective clothing with less discomfort. Innovative cooling methods, like forced ventilation, phase change materials and encapsulated endothermic salts, will be integrated with protective clothing. Sensors in the suit will measure relevant physiological data, such as skin temperature, heat flux and heart rate, to assess the thermal status of the worker, and the environmental conditions (temperature, relative humidity). The physiological signals will be used in an algorithm that will generate a warning signal when a certain safety threshold is surpassed. Data will also be transferred to industrial safety systems in order to alert rescue workers if needed. The operational benefit of prototypes of the suit will be determined in a controlled setting as well as in the industry where protective suits are indispensable. The results will be disseminated to standardization organizations, the industry and public procurement organizations. A training program will be made that focuses on the acceptability of the system by SME’s and end-users. Although the system aims to contain the newest technology, human factors and practical usability including for instance ease of cleaning are leading in the design of the prototypes.

Project aims and objectives To develop a protective clothing system to the production stage, incorporating physiological sensing and advanced cooling support.

Research deliverables (academic and industrial) A clothing concept based on existing technologies that is ready for mass production. Publications and outputs www.prospie.eu; http://cordis.europa.eu (search for PROSPIE).

Research register

51

IJCST 22,6

52

Maribor, Slovenia University of Maribor, Faculty of Mechanical Engineering, Smetanova ulica 17, SI-2000 Maribor, Slovenia. Tel: +386 2 220 7960; Fax: +386 2 220 7996; E-mail: [email protected] Principal investigator(s): Prof. Dr Sc. Jelka Gersˇak Research staff: Research Unit Clothing Engineering

Clothing engineering and textile materials Academic partners

Industrial partners

N/A N/A Project start date: 1 January 2009 Project end date: 31 December 2012 Project budget: 85.000 ECU for 209 Source of support: Slovenian Research Agency Keywords: Clothing, Fabric, Fabric mechanics, Behaviour, Comfort, Prediction The research work of the programme group, which is designed as an upgrade of theoretical and applied achievements of the earlier research program “Clothing and textile materials engineering”, combine three closely related areas of importance for the development of clothing science and technology. These areas are: .

study the draping behavior of woven fabrics;

.

survey of draping of garments, their fit and 3D dynamic clothing simulation;

.

study and development of the model for evaluation of thermophysiological and psychological (wellbeing); and

.

comfort at garment wearing.

Within the first area, the studies related to complex deformation of fabrics and their asymmetric draping behaviour have been carried out. Based on extensive research it was established that particular fabrics had complex three-dimensional shape and different number of folds that can be odd or even. This unique form of resulting folds can be attributed to anisotropic properties of textile materials depending on the fibre type, kind, structures and construction of the yarn, fabric construction parameters, as well as on finishing and mechanical properties that affect attributes of draping. In the area of research related to draping of garments, their fit and 3D dynamic clothing simulation, we focused on the research of complex deformation at dynamic conditions, where we studied the influence of acting external forces on the change of geometry of the deformation, which is connected to the change of potential energy related to the fabric’s deformed shape and kinetic energy, as well as to parameters derived from anisotropic properties of the material and its nonlinear response. The third area of research has been directed to the study of the relationship between material properties and different physiological parameters of thermal comfort in wearing of clothing with the aim of creating a suitable database of thermo-physiological

comfort of the wearer. On the basis of extensive research on the relationship between material properties of the materials built-in the clothing and change the physiological parameters of the test persons during wearing of analyzed models in a warm environment, we found out that the material properties of woven fabrics have a direct influence on the thermo-physiological wearing comfort. The resulting correlation is statistically confirmed in a higher ambient temperature, in the case of lower velocity of air movement, while at higher velocity of air movement, there is no statistically reliable correlation. The resulting variations in the levels of accumulated amounts of sweat in tested garments depending on the amount of evaporated sweat are the consequence of the effect of climate-dependent material properties of textile materials related to the velocity of air movement in a warm environment and thus connected heat-regulating activity of the analyzed models, which directly affects the assurance of thermophysiological comfort in garment wearing in warm environment.

Project aims and objectives The main aims and objectives of the research programme “Clothing engineering and textile materials” are as follows: . .

to define the elastic behaviour of complex textile structures; evaluation of fabric behaviour at draping in a form of complex deformation;

.

to define of reasons for asymetrical behaviour of complex textile structures;

.

development of the material model and simulation of complex textile structures; design of the data base of factors related to the level of the appearance quality of business clothing;

.

.

design of the model for definition of thermal resistance of one and multi-layer clothing systems;

.

to set-up the characterisation of parameters related to thermophysiological comfort;

.

design of the data base related to subjective estimation of thermal comfort; and design of the model for predicting the physiological and psychological comfort at garment wearing.

.

Research deliverables (academic and industrial) The results of research programme will serve for setting-up the engineering concept for “knowledge-based products”. An important contribution to further development of the science is expected also from gained theoretical cognitions from the field related to the study of the relationship between the matter properties of textile materials and heat transfer and thermophysiological and ergonomic comfort of a human being wearing different kinds of garments and thereof developed theories.

Research register

53

IJCST 22,6

54

Publications and outputs C¸ay, A., Vassiliadis, S., Gersˇak, J. and Provatidis, C. (2009), “Parametric correlation of the measurement signals from automated and manual KES-F”, Proceedings of the 9th AUTEX Conference. Izmir, Ege University, Engineering Faculty, Department of Textile Engineering, pp. 1374-7. Fatkic´, E. and Gersˇak, J. (2009), “Influence of elasthan monofilament on structural parameters of jersey weft knit”, 2nd Scientific-professional symposium Textile science and economy, 23 January, Zagreb, Book of Proceedings, University of Zagreb, Faculty of Textile Technology, Zagreb, pp. 105-8. Go¨ktepe, F., Halasz, M., Tamas, P., Go¨ktepe, O¨., Gersˇak, J., Al-Gaadi, B. and O¨zdemir, D. (2010), “Twist direction and yarn type effect on draping properties”, 41st International Symposium on Novelties in Textiles, Symposium proceedings. Ljubljana: Faculty of Natural Sciences and Engineering, Department of Textiles, pp. 171-7. Grujic´, D., Gersˇak, J. and Ristic´, M. (2010), “Effect of physical and sorption properties of the amount of the absorbed sweat in the garment”, Tekstil, Vol. 59 No. 3, pp. 68-79. Tokmak, O., Berkalp, O. Berk and Gersˇak, J. (2010), “Investigation of the mechanics and performance of woven fabrics using objective evaluation techniques. Part 1, The relationship between FAST, KES-F and Cusick’s drape-meter parameters”. Fibres Text. East. Eur., Vol. 18 No. 2(79), pp. 55-9. Zavec Pavlinic´, D. and Gersˇak, J. (2009), “Predicting garment appearance quality”, The Open Textile Journal, Vol. 2, pp. 29-38.

Newark, Delaware, USA University of Delaware, 201 Alison Hall West, Newark, DE 19716, USA. Tel: 1-302-831-6124; Fax: 1-302-831-6081; E-mail: [email protected] Principal investigator(s): Huantian Cao, Rita Chang, Jennifer McCord, Jenna Shaw, Heather Starner, Jo Kallal

Change without buying: an application of adaptable design in apparel Academic partners

Industrial partners

Lauren Heine, Lauren Heine Group LLC Project start date: August 15, 2009 Project end date: August 14, 2010 Project budget: $10,000 Source of support: US Environmental Protection Agency Keywords: Material conservation, Environmentally conscious manufacturing, Inherently benign materials and chemicals, Reuse Excess consumption of apparel is driven by the apparel industry to offer more styles at lower prices in shorter time and the consumers’ desire to change fashion. Environmental problems such as pollution, hazardous waste, and natural resource depletion are related to excess apparel consumption. According to University of Cambridge report, on average, to produce 1 kg of textile and clothing output, about 0.6 kg

of oil equivalent primary energy and 60 kg of water are used, and about 2 kg of CO2 equivalent, 45 kg of waste water, and 1 kg of solid waste are generated. The goal of this project is to demonstrate a strategy of collaboration between apparel industry and consumers to decrease discard, increase utilization, retard fast-fashion, and promote longer wear of garments. We will implement adaptable design in apparel and demonstrate that adaptable apparel will meet consumers’ needs to change while reduce overall production and consumption. In this project, the target users are female college students and the apparel adaptability focuses on function, fit, and style.

Project aims and objectives The purpose of this project is to apply adaptable design in apparel and demonstrate that this strategy will allow the apparel industry to make a profit with better design and high quality product rather than large quantity and low quality; and will meet consumers’ desire to change without buying. Our objectives include: designing and producing adaptable apparel for female college students by using environmentally friendly materials; evaluating the adaptability, consumers’ acceptance, and cost of our design and product; revising the design based on evaluation results and developing educational tools.

Research deliverables (academic and industrial) Design and research are in progress.

Ohtsu-city, Japan SCI-TEX, 12-15, hanazono-cho, Ohtsu-city, 520-0222 Japan. Tel: 81-77-572-3332; Fax: 81-77-572-3332; E-mail: [email protected] Principal investigator(s): Tatsuki Matsuo

Propagation of knowledge on new textile science and technology Academic partners

Industrial partners

Project start date: Project end date: on going Project budget: N/A Source of support: N/A Keywords: Advanced technical textiles, Knowledge propagation The importance of advanced technical textiles has increased in the textile industry of developed countries. In addition, R&D on nano-technologies and electric-textiles are now intensively carried out. In this situation, propagation of knowledge on new textile science and technology must be meaningful.

Research register

55

IJCST 22,6

This project is being conducted individually by T. Matsuo through symposium lectures, journal articles and monographic books.

Ohtsu, Japan 56

SCI-TEX, 12-15 Hanazono-cho, Ohtsu, 520-0222 Japan. Tel: 077-572-3332; Fax: 077-572-3332; E-mail: [email protected] Principal investigator(s): Tatsuki Matsuo

Potential and limitation in fibrous configurational functions as materials Academic partners

Industrial partners

Project start date: 2010, August Project end date: 2011, February Project budget: Source of support: Keywords: Configurational function of fibrous material, Its potential and limitation Recently fibers have been intensively applied to technical textiles as functional materials. Further, several kinds of new fibrous materials such as carbon nano-tube, organic and inorganic nano-fibers, and metallic nano-wires have been intensively developed. The configurational functions of fibrous materials are consisted of the following four elements: (1) They are flexible (pliable). (2) They can have high ability in their axial transmission in such matters as mechanical load, heat conduction, electric conduction, optical light, liquid, particles and ion. (3) They have comparatively high extended specific surface area. (4) They have technological easiness in transformability into fiber assembly structures such as threads, fabrics, nonwovens paper and 3D structures. In this study, how effectively these configurational functions are used in several technical textile applications are investigated. What kinds of limitations as compared with the other kinds of material forms are there in the application of fibrous materials is also investigated. Finally, the potential in the new applications of fibrous materials in the near future are discussed.

Project aims and objectives To try to clarify the potential and limitation of fibrous configurational functions as compared with the other kinds of materials such as bilk and particle. Publications and outputs The results of this study will be presented the 39th Textile Research Symposium held at New Delhi in 2010 December.

Pisa, Italy University of Pisa, Via Diotisalvi, 2, 56126 Pisa, Italy. Tel: +39 0502217053; Fax: +39 0502217051; E-mail: [email protected] Principal investigator(s): Prof. Danilo De Rossi Research staff: Prof. Bruno Neri; Ing. Alessandro Tognetti; Ing. Enzo Pasquale Scilingo; Ing. Federico Carpi; Ing. Antonio Lanata`

PROETEX: protection e-textiles: micronanostructured fibre systems for emergency-disaster wear Academic partners

Industrial partners

Consiglio Nazionale delle Ricerche – INFM, Technical University of Lodz, Ghent University – Department of Textiles, University of PISA, Dublin City University, Institut National des Sciences Applique´es de Lyon

Smartex srl, Milior, Sofileta SAS, Thuasne, Commissariat a` l’Energie Atomique – “CEA”, CSEM Centre Suisse d’Electronique et de Microtechnique SA, Sensor Technology and Devices Ltd, Steiger S.A., Philips GmbH, Zweigniederlassung Forschungslaboratorien, Ciba Spezialita¨tenchemie AG, Diadora Invicta SpA, iXscient Ltd, Zarlink Semiconductor Limited, Brunet-Lion SAS, Brigade de Sappeurs Pompiers de Paris, European Centre for Research and Training in Earthquake Engineering, Direction de la De´fense et de la Se´curite´ Civiles Project start date: February 2006 Project end date: January 2010 Project budget: University of Pisa: e780,443; Total: e 12,792,242 (requested: e 8.100.000) Source of support: European Commission ProeTEX will develop integrated smart wearables for emergency disaster intervention personnel, improving their safety, coordination and efficiency and for injured civilians, optimising their survival management. This core application area, which is of significant societal importance in itself, will drive a wide range of key technology developments, building on current and past EU and national projects and the commercial activities of partners, to create micro-nano-engineering smart textile systems – integrated systems (fabrics, wearable garments) using specifically fibrebased micronano technologies. These are capable of being combined into diverse products addressing this core application area but also a wide range of other markets from extreme sports, through healthcare to transportation maintenance and building workers. The industrial partners can address these markets.

Research register

57

IJCST 22,6

58

Fiber systems can integrate sensors, actuators, conductors, power management, and the emergency disaster personnel smart garment will, within a wireless ambient planning and managing environment, progressively enhance and integrate fiber systems for: .

continuous monitoring of life signs (biopotentials, breathing movement, cardiac sounds);

.

continuous monitoring biosensors (sweat, dehydration, electrolytes, stress indicators);

.

pose and activity monitoring;

.

low power local wireless communications, including integrated fiber antennae;

.

active visibility enhancement, light emitting fibers;

.

internal temperature monitoring using fiber sensors;

.

external chemical detection, including toxic gases and vapours; and

.

power generation – photovoltaic and thermoelectric and power storage.

The technological base developed will concentrate on smart fibers/e-textiles, but the IP will combine these where appropriate with “conventional” microsystems (such as accelerometers, gyros, microcontrollers and wireless chips).

Project aims and objectives The central IP goal is to develop an integrated set of functional garments for emergency disaster personnel, such as firefighters and paramedics, plus systems for injured civilians. These will be produced using both enhanced and novel fibre based micronanosystems, whose development will extend the state of the art in this area. The project will roll out a sequence of progressively more capable integrated wearable systems for emergency disaster intervention personnel and injured civilians. Thus, overall the IP will: .

Progress the fundamentals of fibre-based sensor, processing, communications and power management systems.

.

Integrate these fibre-based capabilities into functional knitted or woven wearable garments.

.

Produce fully capable integrated communicating, sensor wearables, using additional “conventional” systems where necessary.

.

Test their match of user needs and requirements in a lab-based setting.

.

Demonstrate their function in a real-world application in a number of field trials.

Scientific objectives (1) Develop a multifunctional garment integrating an increasingly ambitious set of sensors and energy harvesting and storage which is reliable, robust, easy to wear and capable of manufacture. (2) Into this garment: Design, test and integrate a bioelectrical heart rate monitor into whole skin contact garment interface; Design, test and Integrate a cardiac sound monitor; Integrate sensor breathing monitor and ensure that signal conditioning and processing results in successful way.

(3) Develop fibre and new textile based technological solutions, with reliable functionality, capable of integration into wearable garments covering the following set of technological area capabilities: . Monitor bioelectrical potential. .

Sensing breath movements.

.

Sensing posture and movement. Biochemical sensing, specifically determination of dehydration status.

. . .

Sensing core temperature. Acting as local communications antennae.

.

Sensing external toxic gases/chemicals, including CO.

.

Generating local energy using thermoelectric generation. Generating local energy using photovoltaic processes.

.

Storing energy using Li-Ion textile batteries. Technical objectives .

.

.

.

.

Develop and adapt textile manufacturing processes to these new active fibres and layers (weaving, knitting, coating, laminating) but also innovate in terms of clothes conception to optimise the assembly step regarding interconnection needs for e-textile garment. Develop and test a multifunctional (inner and outer) garment integrating an increasingly ambitious set of sensors and energy harvesting and storage which is reliable, robust, easy to wear and capable of manufacture for both intervention people and injured civilians. The inner and out garment will include an adapted set of functionalities based on the developed technologies. As example first inner garment could integrate bioelectrical heart rate monitor, cardiac sound monitor, strain sensor breathing monitor inner temperature measurement and ensure that signal conditioning and processing results in successful and robust physiological monitoring. Energy generated by the heat (thermoelectricity) and the movement (piezoelectricity) of the of the wearer. Outer garment will typically include toxic gas measurement, external temperature; motion and position monitoring, data transmission system, energy could be provided by photovoltaic external layer and textile Li-Ion batteries. Realize field trial of the instrumented garment for technological validation.

Research deliverables (academic and industrial) The key deliverables will be: .

An inner garment for emergency disaster personnel, monitoring the health of the user through vital signs, biochemical parameters, activity and posture, generating and storing own power and communicating locally with other wearables and relaying through (D).

Research register

59

IJCST 22,6

.

An outer garment for emergency disaster personnel, measuring potential environmental insults (temperature, CO, other toxic gases), sensing posture and movement of the wearer and offering improved visibility, generating and storing its own power communicating locally with other wearables and relaying through (D).

.

An under-garment jerkin or chest band for injured civilians (closely related to (A)) monitoring their health, generating and storing its own power and communicating locally, relaying information via (E). Victims monitoring measures will include: body temperature; cardiac pulse; respiration rate; ECG; percutaneous CO saturation; percutaneous O2.

60 .

.

.

A portable unit for the emergency disaster personnel, communicating with A, B and C, but offering additional “conventional” microsystem, providing both local and long range communication (acting as a relay for A, B and C), including some specific sensors not easily integrated into (B), plus accelerometers, gyros and GPS to enable high accuracy position and movement determination. This device should allow data entry and displays/alarms. A portable unit for injured civilians, to include data relay capability and INS/GPS but no data entry or display. Some kind of integrated alarm or indicator to give the overall civilians health status. A simple user input, such as panic button, may be required.

Seoul, Republic of Korea A-108, iFashion Technology Center, Col. of Eng., Konkuk University, Gwang Jin Gu, Seoul, Korea, 143-701. Tel: +82-2-17-310-6317; Fax: +82-2-452-7865; E-mail: [email protected] Principal investigator(s): InHwan Sul

Real-time cloth simulation using particle based method and Pluecker coordinates Academic partners

Industrial partners

Project start date: 1 January 2009 Project end date: Project budget: Source of support: Keywords: Cloth simulation, Particle based method, Collision detection Particle based method is more suitable method for cloth simulation than finite element method because of its numerical stability and speed. But due to the complex nature of cloth, it is not easy to simulate the cloth in real-time. There are two components to

achieve real-time in cloth simulation. The first one is the calculation of velocities(i.e. drape engine) and the second one is the collision detection(collision engine). We use various techniques which are currently known to speed up the calculation of the velocities such as using block symmetry of matrices or modal-analysis based bending force approximation. And for the collision detection, we approximated the meshes to a k-DOP’s and found the intersection among velocity vectors and k-DOP’s. Pluecker coordinate was used for fast ray-BOX intersection test.

Project aims and objectives The objective is to make a versatile drape engine which can simulate cloths in any environment.

Research deliverables (academic and industrial) The technique can be readily used in 3D apparel CAD systems and modeling of textile process. Moreover, it can be applied to any other scientific research such as a medical simulation where calculation speed is more important factor than mechanical accuracy. Publications and outputs This work is based on the previous works: In Hwan Sul, “Fast Cloth Collision Detection Using Collision Matrix”, International Journal of Clothing Science and Technology, submitted (2008). In Hwan Sul and Tae Jin Kang, “Improvement of drape simulation speed using constrained fabric collision”, International Journal of Clothing Science and Technology, Vol. 16, pp. 43-50 (2004). In Hwan Sul, Sung Min Kim, Yong-Seung Chi and Tae Jin Kang, “Simulation of Cusick Drapemeter using particle based modeling: stability analysis of Explicit Integration Methods”, Textile Research Journal, Vol. 76 No. 9, pp. 712-19 (2006).

Sliven, Bulgaria College of Sliven (TU – Sofia), 59 “Bourgasko chaussee”, 8800 Sliven. Tel: 00359 44 667710; Fax: 00359 44 667505; E-mail: [email protected] Principal investigator(s): Ivelin Rahnev Research staff: 4 university lecturers from TU-Sofia, College of Sliven

Technology optimization of sirospun cotton yarns Academic partners

Industrial partners

Department of Textile Materials and “E.Miroglio” AD – Sliven, www.emiroglio. Design at the University of Maribo, com Slovenia Project start date: 1 May 2010 Project end date: 28 February 2011 Project budget: None Source of support: Indirectly sponsored by “E.Miroglio” AD – Sliven Keywords: Sirospun, Cotton, Yarns

Research register

61

IJCST 22,6

62

Manufacturing of Sirospun cotton yarns is relatively rare phenomenon in the spinning technology. Principal difficulty derives from the small length of the cotton fibers, which don’t manage to support the particular “spinning triangle” of the Sirospun technics. Another reason for the little by volume manufacturing of Sirospun cotton yarns is the mean trend to prepare weaving warps from single cotton yarns, which be sized in consequence. In the cases when the final product requires a fabric with resistance to wear increased, twisted threads are irreplaceable and the Sirospun cotton yarns become indispensables. Subject of the present investigation is a single ring-spun yarn, from cotton fibers, produced by the Sirospun method. Project includes following stages: (1) Preliminary computations concerning the fibrous composition and the torsion structure of the yarn. (2) Study of the technology capacity of a carded spinning mill for cotton ring-spun yarn. (3) Planning and carrying out an experiment to receive cotton Sirospun yarns. (4) Tests of the samples and laboratory data treatment. (5) Determining of the optimum machine adjustments to produce cotton Sirospun yarns with properties desired. Fibrous composition of the threads designed – Tt 20x2, 100-Cotton, Sirospun; consists of middle – length cotton with fineness Tt 1.67 dtex and staple length – 34.0 mm. Machine equipment to carry out the experimental work is a carding line to process cotton and cotton-type fibrous materials. Single yarns are received on ring spinning frame “Marzoli NSF2”. Mean factors of the technology optimization are the machine adjustments of the distance between the roving spins in the drafting zone and the spinning twists. Purpose of the threads designed is the weaving warp, so, their inherent properties are described by the mechanical resistance at extension, the general irregularity and the stability of the peripheral layer. In order to determine these qualitative indicators, the samples will pass through dynamometric tests, structural analysis and microscopic visualization. After statistical treatment of the laboratory data and regressive analysis, the technic file of the technology prototype will be conceived.

Project aims and objectives Goal of the present work is the obtaining of cotton Sirospun yarn with linear density 20x2, 100-Cotton with optimum properties, destined to weaving warp threads. Attaining this aim passes the following tasks: .

Methodology investigation (fibrous raw materials, equipment capacity and previous results).

.

Experimental work.

.

Receiving of the optimization model and drawing up of a technic file of the linear textile product: Tt 20x2, 100-Cotton, Sirospun.

Research deliverables (academic and industrial) Academic results consist in the investigation of the rheology properties and the structure of the cotton Sirospun yarns.

University results consist in the formulation of final diploma projects and doctoral thesis. Practical results consist in the technic file to reproduce the cotton Sirospun yarn and the patent application. Publications and outputs For our university team this is the beginning of a perspective investigation work. Results of this thorough research on the stages of the project will be described in 3 autonomous publications. A part of them will be reported to the annual conferences of AUTEX, the second part will be represented to the local university seminars and the third part will be offered to newspapers specialized of textile industry and clothing.

Tainan city, Taiwan, R.O.C. Dept of Fashion Design, Tainan University of Technology, 529 Jhongjheng Rd., Yongkang, Tainan 71002, Taiwan, R.O.C. Tel: +886-6-2532106 Ext. 356; Fax: +886-6-2435369; E-mail: [email protected] Principal investigator(s): Miao-Tzu Lin

Benefit analysis of textile integrated photovoltaics with different apparel parts and Azimuth Angle Academic partners

Industrial partners

None None Project start date: Year 2009 Project end date: Year 2010 Project budget: USD 10,000 Source of support: Keywords: Apparel, Solar cell, Textile integrated photovoltaics (TIPV) Textile integrated photovoltaics (TIPV) supplies power in electric garments, in order to fit in with the environmental protection and power saving demand of portable devices, is the most advanced way of providing mobile electricity. The way expects to grow more economic benefits in textile and garment industry. Power efficiency variance depends on the TIPV user azimuth angle change. Because of the solar cell area proportion change of the lighting area to the shadow area influences power efficiency. This study measured the whole lighting area power and the whole shadow area power to calculate the power when azimuth angle change. In order to confirm the calculating result, special azimuth angle power measuring are necessary. This study used the commercial a-Si flexible solar cell with hook and loop tape for removable function to get the solar cell power in clothing and calculating the average power and the power per day. The result may be suitable for

Research register

63

IJCST 22,6

designer to design clothing, such as solar cell amount calculating, fashion design and function demand.

Project aims and objectives

64

This study measured the whole lighting area power and the whole shadow area power to calculate the power when azimuth angle change. In order to confirm the calculating result, special azimuth angle power measuring are necessary. This study used the commercial a-Si flexible solar cell with hook and loop tape for removable function to get the solar cell power in clothing and calculating the average power and the power per day. The result may be suitable for designer to design clothing, such as solar cell amount calculating, fashion design and function demand.

Wuxi, China Jiangnan University, Lihu Road 1800, Wuxi, China, 214122. Tel: +86-510-85327307; Fax: +86-510-85327307; E-mail: [email protected] Principal investigator(s): Liu Jihong Research staff: Qian Kun, Panruru, Yang Ruihua

Mechanical property and application of “8” shape 3D woven enhancing composite Academic partners

Industrial partners

Jiangsu Information College Nanjing Composite Company of China Project start date: 1 January 2008 Project end date: 31 December 2011 Project budget: $20,000 Source of support: Nanjing Composite Company of China Keywords: 3D woven enhancing fabric, Composites, Binder yarn, Mechanical Property, Model, “8” Shape Three-dimensional (3D) woven enhancing fabric and its composite was produced on a modified rapier loom. Weaving parameters were studied and restructuring method was researched. After that mechanical property including tension, compress, and so on will be modelled an research. The results express that the property has relationship with the direction of fabric and layers.

Project aims and objectives .

Research on the parameters of producing woven fabric.

.

Build a model and research on mechanical properties.

Research deliverables (academic and industrial) .

A reconstruction loom for producing “8” shape woven fabric.

.

A kind of method for producing composite of “8” shape woven fabric.

Publications and outputs Weaving Thickness Parameters of “8” Shape 3D Woven Enhancing Fabric. Mechanical Property and Model of “8” Shape 3D Woven Enhancing Composite.

Zagreb, Croatia

65

Faculty of Textile Technology, University of Zagreb, Prilaz baruna Filipovic´a 30, HR-10000 Zagreb, Croatia. Tel: +385 1 37 12 552; Fax: +385 1 37 12 599; E-mail: [email protected] Principal investigator(s): Assist. Prof. Zˇeljko Sˇomodi, PhD Research staff: Assist. Prof. Ana Kunsˇtek, PhD, Slavica Bogovic´, MSc, Anica Hursa, MSc Igor Petrunic´, MSc, Assist. Prof. Simona Jevsˇnik, PhD, Daniela Zavec-Pavlinic´, PhD

Computational modelling in engineering analysis of textiles and garment Academic partners

Research register

Industrial partners

Kamensko d.d., Zagreb University of Maribor, Faculty of Mechanical Engineering, Maribor, Slovenia Project start date: 1 January 2007 Project end date: 31 December 2011 Project budget: N/A Source of support: Ministry of Science, Education and Sports, Republic of Croatia Keywords: Clothing technology, Numerical methods, Optimisation, Reinforcements on clothing The intention of this project is to give a contribution to advanced application of the methods of engineering analysis in the field of textiles and garment. This goal will be achieved by introduction, adaptation, elaboration and application of up-todate computational methods in the analyses of problems relevant for the field of textile and clothing engineering. Considering the existing experience and an overview of questions and problems actual for the engineering science in the field, the research is to be undertaken in a number of areas, such as: optimal design of structural reinforcements in garment based on the finite element analysis; three parameter model of tensile nonlinearity of textiles; computational evaluation of post-buckling stable state in prediction and simulation of fabric drape; general numerical solution of thin plate bending with application to optimal grip geometry in automated work piece manipulation. Depending on the timing and realization of these researches, there is a possibility of opening further research areas from the field of computational modelling in mechanics of textiles and garment, including the spatial modelling and design of clothing items. The methods of research to be applied primarily consist of derivation and elaboration of numerical models suitable for application in the problems under

IJCST 22,6

66

consideration, and the development and application of computer programmes based on these models. At the same time, the plan is to acquire and apply some of the existing software applicable in the problems to be considered, as well as to prepare and conduct experimental verification of results obtained by computations.

Project aims and objectives Aim and scope of the proposed research is to give a contribution in the improvement of the level of engineering and technological know-how in the field of textile and garment. The research is expected to result in computer programmes or engineering data collected in tables, diagrams, etc. that will be useful for the problem solution in the area of expertise covered by the research. The knowledge and methods developed by the research will be on offer for the interested subjects, from Croatia or elsewhere, primarily from the branch of textiles and garment. It can also be expected that after some time the collected knowledge and methods will be included in the teaching process at the Faculty of Textile Technology, primarily as parts of the subjects at the doctoral or diploma levels.

Research deliverables (academic and industrial) The principal user of the results will be Faculty of Textile Technology, as the institution for production and transfer of knowledge in the field of textile and garment. Further users will be the firms, institutions and individuals from among the designers and manufacturers of textile and garment, who already have the co-operation with the faculty, or shall have that co-operation in the future. The results and findings of the research shall be offered to these subjects by means of the Centre for development and transfer of textile and clothing technology and fashion design, as a unit in the structure of the faculty. The specific applications are expected in the expert analyses related to engineering in preparation of production processes in which the problems from the field of research appear. Publications and outputs Akrap-Kotevski, V. and Kunsˇtek, A. Writing Ability after Brain Damage, Proceedings of 3rd International Ergonomics Conference, Ergonomics 2007, Mijovic´, Budimir (Ed.), Zagreb, Croatian Society of Ergonomics, 2007, 279-85. Hursa, A., Rogale, D. and Sˇomodi, Zˇ “Application of numerical methods in the textile and clothing technology”, Tekstil, Vol. 55 No. 12, pp. 613-23 (2006). Sˇomodi, Zˇ., Hursa, A., and Rogale, D. “A minimisation algorithm with application to optimal design of reinforcements in textiles and garments”, Internationl Journal of Clothing Science and Technology, Vol. 19 Nos 3/4, pp. 159-66 (2007). Sˇomodi, Zˇ., Hursa, A., Rolich, T. and Rogale, D. Numerical analysis and optimisation of mechanical reinforcement on clothing, Book of Proceedings of the 1st meeting of Croatian Society of Mechanics, Cˇanadija, Marko (Ed.), Rijeka, Croatian Society of Mechanics, 2007, 173-8.

Zagreb, Croatia Faculty of Textile Technology, Prilaz baruna Filipovica 30, 10 000 Zagreb, Croatia. Tel: +385 1 37 12 577; Fax: +385 1 37 12 533; E-mail: [email protected] Principal investigator(s): Prof. Zenun Skenderi, PhD Research staff: Prof. Miroslav Srdjak, PhD, Prof. Momir Nikolic´, PhD, Prof. Alka Mihelic´-Bogdanic´, PhD, Bozˇo Tomic´, MSc, Vesna Marija Potocˇic´ Matkovic´, MSc, Ivana Salopek, MSc, Dragana Kopitar, BSc

Multifunctional technical nonwoven and knitted textiles, composites and yarns Academic partners

Industrial partners

N/A

Cˇateks d.d. Cˇakovec, Croatia, Regeneracija non-woven and carpets j.s.c., Zabok, Croatia Project end date: 1 January 2010

Project start date: 1 January 2007 Project budget:– Source of support: Ministry of Science, Education and Sport, Republic of Croatia Keywords: Yarns, Knitted and nonwoven fabrics, Structures, Properties, Thermal and vater-vapour resistance properties Further dislocation of the textile production from developed countries into Asia is a basic characteristic for the world textile industry today. In the field of technical textiles profound resistance is felt against relocation. An increase in the production of technical textiles is recorded due to a permanent expansion of the application range. It is used in: transportation, industry, medicine, hygiene, household, garment industry, agriculture, fishing trade, civil engineering, sport, safety, ecology, etc. Nonwovens make the most significant contribution to the development of technical textiles. Over last decades the technology of nonwovens production has experienced a rapid development, and the production of late years has registered an annual increase of approx. 10%. A significant application range for technical textiles or geotextiles is civil engineering, in particular road building. In addition to woven, knitted and similar structures, nonwovens play a predominant role with a share of approx. 75% in 2005. The most important functions of geotextiles are: separation of weak soil, reinforcement of soil or elements of building structures, filtration and drainage. Geotextile properties are: stability, uniform structure, small thickness, high strength and stretching, porosity, small surface mass and water permeability. Various applications require a more or less marked particular structure and characteristic. The first part of the project will deal with various structures and properties of technical textiles based on nonwoven and knitted structures, in particular on geotextiles. Moreover, manufacturing technologies of technical textiles and knitted

Research register

67

IJCST 22,6

68

materials as well as their controlling parameters will be discussed. Conventional technologies such as: spinning, weaving, knitting and clothing technology will probably not withstand the competitiveness coming from Asia. Besides, relocation of the manufacture of man-made fibres into the Far East is taking place. It is undoubtedly the case that only those disposing of raw materials and enough knowledge to produce and sell high-quality products will have the chances of survival on the market. The investigation of possibilities of manufacturing from coarser sorts of wool which have similar fineness as domestic wool and the investigation of their possible use for products such as carpets and several articles of clothing will be within the scope of this project. The limit of fibre spinnability, typical stress-strain curves, yarn behavior in cyclic examinations of elongation properties, surface friction and yarn hairiness.

Project aims and objectives There are two dominant reasons why the field of technical textiles is dealt with in the project: .

Intensive development of technical textiles because of an increase in the application (technical textiles on average of approx. 5.5%, nonwoven fabric approx. 10%).

.

Resistance to the relocation of the manufacture of technical textiles to the Far East.

As a result the manufacture of technical textiles (quantity) in the developed countries accounts for more than 40% of the total production of textiles. The state of the Croatian textile industry should be noted where the classic textile industry has almost vanished. Two companies Regeneracija and ateks manufacture technical textiles which have their markets. Regeneracija and ateks confirmed their participation in the project. Moreover, the use of domestic wool for various products of higher value is a challenge and obligation. However, today domestic wool is sold as a raw material, and in some regions it is not bought off which is an ecological problem. Based on the above mentioned facts, it is reasonable to deal with the subject matter of technical textiles and yarns and products, respectively, such as carpets within the scope of the project. In this way production of higher-quality products is promoted which is the purpose of this investigation. The aims of the investigation are as follows: (1) Definition of the interdependence, process parameters and physical-mechanical as well as other relevant properties: . technical nonwoven textile; . technical textile based on nonwoven fabric coated with polyurethane (PUR); and . technical textile based on knitted fabric from PET and PA coated with polyurethane (PUR) and other technical textiles based on knitted fabric. (2) Definition of the interdependence of raw materials, process parameters and physical-mechanical yarn properties, primarily wool, and the behavior of carpets in dynamic investigations (new instrument required for the project). (3) It will be attempted in case of interest of Regeneracija or other interested parties to investigate thermal properties of wool insulation materials.

Research deliverables (academic and industrial) Obtaining the new understandings of: . Thermal and vapor resistance properties of the knitted and nonwoven fabrics. .

Spun yarns, primarily coarser wool yarns: procedures of manufacturing and defined the controlling parameters of the processes.

.

Structure and properties of the coarser, industrially manufactured wool yarns should be emphasized that can be similar to the yarn spun from domestic wool as far as their properties.

.

Influence of fibres parameters (fineness, length, . . .) on limit of spinnability and spun yarn properties.

.

Possibilities of manufacturing carpets from coarser wool fibres will be investigate as well as compressibility of carpets on the new instrument purchased from funds of the project.

.

Processes of manufacturing: technical nonwoven fabrics intended for use in civil engineering, technical nonwoven fabric coated with polyurethane (PUR), technical textiles based on knitted fabric coated with polyurethane (PUR), as well as their structures and properties.

Publications and outputs Kopitar, D. and Skenderi, Z. “Prsteni i trkacˇi – glavni elementi prstenaste predilice”, Tekstil, Vol. 55, pp. 501-43 (2006). Potocˇic´, M., Vesna, M. and Salopek, I. Computer Assisted Study of Knitted Structures. Proceedings Vol. IV. CE Computers in Education (2007), pp. 99-102. Salopek, I. and Skenderi, Z. “Thermophysiological comfort of knitted fabrics in moderate and hot environment”, Proceedings of the 3rd International Ergonomics Conference, Mijovic´, Budimir (ur.). Zagreb: Croatian Society of Ergonomics, 2007, pp. 287-93. Salopek, I. Skenderi, Z. and Srdjak, M. Stoffgriff – ein Aspekt des Tragekomforts von Strickware. Melliand Textilberichte, Vol. 88, pp. 426-8 (2007).

Zagreb, Croatia University of Zagreb, Faculty of Textile Technology, Prilaz baruna Filipovic´a 30, HR-10000 Zagreb, Croatia. Tel: +385 1 37 12 521; Fax: +385 1 37 12 599; E-mail: [email protected] Principal investigator(s): Assoc. Prof. Emira Pezelj, PhD Research staff: Prof. Ruzˇica Cˇunko, PhD, Prof. Maja Andrassy, PhD, Assist. Prof. Edita Vujasinovic, Prof. Vili Bukosˇek, PhD, Antoneta Tomljenovic´, PhD, Sanja Ercegovic´, MSc, Maja Somogyi, BSc, Dubravka Gordosˇ, MSc

Research register

69

IJCST 22,6

MULIFUNCTIONAL HUMAN PROTECTIVE TEXTILE MATERIALS Academic partners

70

Industrial partners

Project start date: 1 January 2007 Project end date: 31 December 2011 Project budget: N/A Source of support: Croatian Ministry of Science, Education and Sport, Republic of Croatia Keywords: Protective textiles, Multifunctionality, Smart textiles, Ceramic coatings, Sol-gel process The investigations proposed have been motivated by the fact that people are more and more exposed to various influences from the environment, which can be harmful to their health. Such harmful influences are, for example, UV irradiation, electromagnetic smog, high temperature, fire, etc. Contemporary textile materials for personal protection are required to offer high efficiency, in most cases multifunctionality, as well as a necessary level of comfort. The fabrics used are high-performance ones and interdisciplinary approach is necessary in research dealing with their development and manufacture. The thesis we propose is that the application of contemporary research results in the field of materials can be used to offer a new contribution to the development of multifunctional protective textile materials. The accent will be given to a purposeful surface modification of fabrics, using environmentally friendly agents and processes, which is in accordance with contemporary European trends of research in the field of materials. Special attention will be paid to investigating modifications using the new sol/gel process, combined with preceding ultrasound, laser and plasma treatment of textile surfaces. New possibilities of manufacturing efficient protective layers will be investigated, using various inorganic substances, including functional layers of nano-dimension made of hybrid inorganicorganic polymers. The aim is to optimise modification parameters of achieving efficient protection from UV and EM irradiation, as well as to increase resistance to abrasion, cutting and heat in particular materials, establishing antimicrobial properties at the same time. Adequate testing procedures will be established to evaluate the newly created materials. New levels of knowledge is expected to be achieved regarding correlation of protective properties and textile fabric composition, as well as the development of practical processes of obtaining aimed fabric modifications and the development of the methods of new material evaluation. New knowledge will contribute to the quality of education in the field of textile materials. Transfer of knowledge into actual industrial production is also expected. The results will be presented on international conferences and will be published in relevant international publications. The obtained results to be obtained could be used to stimulate manufacture of new high-performance textile materials for special purposes in Croatia.

Project aims and objectives The purpose of the investigations is to obtain new knowledge in the field of material development, especially regarding the new composites with textiles as a basic component. The knowledge should be directly applicable in practice, and simultaneously used to improve the quality of education, of both students, young researchers and experts from the industry. The new knowledge is expected to further the

development of the Department of textile materials, where the investigations are organised. Based on the knowledge of high-performance materials, that has resulted in the development of the composites, and the role of textile component in them, the possibility will be investigated of obtaining high-performance composites for protection, in which textiles are the basic component. These are new textile materials to be used as protection from harmful influences of the general and working environment in high-risk industrial processes and other activities where people are exposed to risks of mechanical, thermal or chemical injuries, of infection by micro-organisms and even fatal risks from the causes. This is why protective materials are expected to offer high efficiency under various conditions, while the best solutions are aimed at obtaining multi-functional protection by a single material. The purpose of the research is to investigate the solutions that could be applied in textile industry, which could stimulate the introduction of knowledge-based and new-technology-based production in the industry, through adapting the industry to manufacture high-performance composite materials for special purposes. The aim of the investigation is to determine the procedures of obtaining multi-functional textiles for personal protection, simple to manufacture and use. The protective properties will be obtained by modifying the surfaces of the fabrics of various constructions, with the aim to establish optimal modification procedures and processing parameters which could offer efficient protection from individual influences, or, otherwise, protection from more influences. The investigations are supposed to result in solutions for objective evaluation of the effect achieved and the durability of protection as well, but also in the evaluation of the adequacy of the materials for a particular purpose. Adequate testing methods and procedures will be developed, appropriate indicators defined and the correlation of the modification parameters and properties achieved established.

Research deliverables (academic and industrial) The purpose of the investigations is to obtain new knowledge in the field of material development, especially regarding the new composites with textiles as a basic component. The knowledge should be directly applicable in practice, and simultaneously used to improve the quality of education, of both students, young researchers and experts from the industry. The new knowledge is expected to further the development of the Department of textile materials, where the investigations are organised. Based on the knowledge of high-performance materials, that has resulted in the development of the composites, and the role of textile component in them, the possibility will be investigated of obtaining high-performance composites for protection, in which textiles are the basic component. These are new textile materials to be used as protection from harmful influences of the general and working environment in high-risk industrial processes and other activities where people are exposed to risks of mechanical, thermal or chemical injuries, of infection by micro-organisms and even fatal risks from the causes. This is why protective materials are expected to offer high efficiency under various conditions, while the best solutions are aimed at obtaining multi-functional protection by a single material. The purpose of the research is to investigate the solutions that could be applied in textile industry, which could stimulate the introduction of knowledge-based and new-technology-based production in the industry, through adapting the industry to manufacture high-performance

Research register

71

IJCST 22,6

72

composite materials for special purposes. The aim of the investigation is to determine the procedures of obtaining multi-functional textiles for personal protection, simple to manufacture and use. The protective properties will be obtained by modifying the surfaces of the fabrics of various constructions, with the aim to establish optimal modification procedures and processing parameters which could offer efficient protection from individual influences, or, otherwise, protection from more influences. The investigations are supposed to result in solutions for objective evaluation of the effect achieved and the durability of protection as well, but also in the evaluation of the adequacy of the materials for a particular purpose. Adequate testing methods and procedures will be developed, appropriate indicators defined and the correlation of the modification parameters and properties achieved established.

Publications and outputs R., Cˇunko, S., Ercegovic´, D. Gordosˇ and E. Pezelj, “Influence of ultrasound on physical properties of wool fibres”, Tekstil, Vol. 55, pp. 1-9 (2006). A. Tomljenovic´, E. Pezelj and F. Sluga, “Application of TiO2 nanoparticles for UV protective shade textile materials”, Proceedings of 38th symposium of textile novelity, Ljubljana 21 June 2007, Slovenija. E. Vujasinovic, Z. Jankovic, Z. Dragcevic, I. Petrunic and D. Rogale, “Investigation of the strength of ultrasonically welded sails”, International Journal of Clothing Science and Technology, Vol. 19 Nos 3/4, pp. 204-14 (2007). E. Vujasinovic, Z. Dragcevic and Z. Bezic: “Descriptors for the objective evaluation of Sailcloth Weather Resistance”, Proceedings of 7th Autex Conference 2007, Tampere 26-28 June 2007, Finland. Ruzˇica Sˇurina i Maja Somogyi: “Biodegradable polymers for biomedical purpose”, Tekstil, Vol. 55 No. 12, pp. 642-5 (2006). Ruzˇica Sˇurina i Maja Andrassy: Resistance of lignocellulosic fibers to microorganisms, XX. hrvatski skup kemicˇara i kemijskih inzˇenjera, knjiga sazˇetaka, posvec´en Lavoslavu Ruzˇicˇki i Vladimiru Prelogu, hrvatskim nobelovcima u kemiji, Zagreb, 26. veljacˇa – 01. ozˇujka 2007., p. 286. Sˇurina Ruzˇica i Andrassy Maja: Quality of Modified Flax Fibers, The 18th International DAAAM Symposium, “Intelligent Manufacturing and Automation: Focus on Creativity, Responsibility and Ethics of Engineers”, 24-27 October 2007.

Zagreb, Croatia Faculty of Textile Technology, University of Zagreb, Prilaz baruna Filipivic´a 30, HR-10 000 Zagreb, Croatia. Tel: +38514877351; Fax: +38514877357; E-mail: [email protected] Principal investigator(s): Prof. emeritus, Ivo Soljacˇic´, PhD Research staff: Asoc. Prof. Tanja Pusˇic´, PhD, Prof. Ljerka Bokic´, PhD, Asst. Prof. Branka Vojnovic´, PhD, Iva Rezic´, PhD, Prof. Jelena Macan, PhD, Asoc. Prof. Barbara Simoncˇic´, PhD, Prof. Sonja Sˇostar-Turk, PhD., Asist. Prof. Sabina Fijan, PhD, Mila Nuber, MSc, Ivan Sˇimic´, MSc, Dinko Pezelj, PhD, Versˇec Josip, MSc

Ethics and ecology in textile finishing and care Academic partners

Industrial partners

University of Maribor and University Labud, d.d. Zagreb and Vodovod, Zagreb of Ljubljana, Slovenia Project start date: 1 January 2007 Project end date: 31 December 2011 Project budget: Source of support: Ministry of Science, Education and Sports, Republic of Croatia Keywords: Wellness finishing of textiles, Determination of harmful substances on textiles, Toxicological and alergenic properties, Environmental protection, Hygiene and effects of textile care, Textile material sample preparation Modern textile finishing processes have to fulfill high demands due to the expectations of new textile materials properties and their persistence during care. Especially interesting in this respect are the new production processes of socks which include implementation of microcapsules that can release active materials for skin moisturizing. Their primal role is prevention of dryness, dandruff and allergenic reactions of the skin. The most suitable analytical methods for determination of durability to washing, friction and sweat will be tested. Durability to washing of products with special properties will be tested with different amounts of anionic and cationic surfactants in liquid detergents. The mechanism of adsorption and desorption, their influence on primary effect of the treatment, and the influence of the pH value and the mechanical way of treatment will be tested. On the ground of the obtained results, analytical methods for determination of micro components in the macro components of textile materials should be proposed, without regards to the specifications of the materials or the method of the treatment. The testing will involve a review of the analytical method of each individual analytical procedure as well as its impact on the obtained information. The parameters of the analytical procedure will be worked out with the purpose of restoration of historical textile by destructive and non-destructive methods for the preservation of national heritage. European controlling methods of new materials have ethical demands involving the human population health which demands an environmental friendly process. For this purpose the processes of textile finishing and care will be optimized. The possibility of obtaining new preventive properties, which were not previously present on the textile material or improvement of present protection, will be tested. The impact of washing cycles with detergent and UV absorber on pastel colored textile materials made of cotton, polyester and their mixtures on UPF and the shade change will be investigated. The quality control of water and effluents will be based on the determination of micro quantities of potential allergens, heavy metals, pesticides, dyes, and surfactants. The traces of solvents will be controlled on the clothing material and in the air during the chemical cleaning and further treatment processing.

Project aims and objectives The main goal of this investigation is to stimulate ethic ecological demands on the production processes, care processes, and thereby on the utilization properties of the textile materials wherewith it would be possible to get the optimal properties of

Research register

73

IJCST 22,6

74

materials regarding their functional properties by avoiding all possible harmful allergenic and toxicological influences of textile materials to consumers. Elaboration of production and textile finishing processes for the optimal effects (wellness finishing, protection from unwanted changes of utilization properties in texcare, elaboration of pastel dyed textiles laundering in detergent with UV absorber, additional laundering quality – UV protection), formulation of new compositions for laundering for the purpose of avoiding secondary harmful effects in modern conditions with maximal saving of water and energy, more safe treatment with solvents during dry cleaning. By monitoring of harmful inorganic and organic substances that are present in micro quantities on the textile materials, textile accessories, textile wastewaters and finished textile products, new analytical methods would be determined. Sampling procedures, sampling preparation steps, selection of appropriate analytical method and the processing of the obtained result will be optimized. In this investigation the mathematical modes for guiding of analytical procedure will be applied, what is economically justified because the time spend for investigation is much shorter, and the consumption of chemical reagents, energy and emission of harmful substances to the environment reduced. Special contribution will be in development of analytical methods for determination of components present on the historical textile, for the purpose of avoiding the damaging of the textile material during restoration conservation treatments.

Research deliverables (academic and industrial) The project is scheduled over three years. Eco problems and human ecology, especially presence of heavy metal traces in textile processes and fibres will be investigated and some results will be published. Analytical methods for qualitative and quantitative determination will be developed. The influence of sweat on the heavy metal emission will be tested from colored textile materials. Possibility and durability of wellness finishing effects particularly on PA pantyhose’s as well as methods will be established. UPF and change in shade of white and pastel colored textiles made from cotton, PET, PA and their blend with cotton during laundering with addition fluorescent compounds in detergent will be researched, too. Hygienic laundering with chemothermic and chemical treatments in order to destroy micro-organisms in compliance with existing recommendations will be done. Potentially irritations of the skin caused by textiles, finishing agents and inadequate rinsing during laundering will be studied. Investigation of anionic, cationic and nonionic surfactant adsorption and desorption influenced by different composition of textile fibres, pH and temperature will be performed. The adsorption and desorption will be studied in order to establish a correlation between zeta potential and swelling capacity of textile fibres. Publications and outputs Fijan, S., Pusˇic´, T., Sˇostar-Turk, S. and Neral, B. “The influence of industrial laundering of hospital textiles on the properties of cotton fabrics”, Textile Research Journal (2007) (in publishing). Pusˇic´, T. and Soljacˇic´, I. “Changes in shade of cotton fabrics during laundering with detergents containing Fluorescent Brightening Agent and UV absorber”, AATCC Review (2007) (in publishing). Pusˇic´, T. Jelicˇic´, J. Nuber, M. and Soljacˇic´, I. “Istrazˇivanje sredstava za kemijsko bijeljenje u pranju”. Tekstil (2007) (in publishing).

Rezic´, I. and Steffan, I. “ICP-OES determination of metals present in textile materials”, Microchemical Journal, Vol. 85 No. 1, pp. 46-51 (scientific paper) (2007). Vojnovic´, B., Bokic´, L., Kozina, M. and Kozina, A. “Optimization of analytical procedure for phosphate determination in detergent powders and in laundry wastewater”, Tekstil (2007) (in publishing).

75

Zagreb, Croatia Faculty of Textile technology, Prilaz baruna Filipovica 28a, 10000 Zagreb, Croatia. Tel: +385 1 3712 540; Fax: +385 1 37 12 599; E-mail: [email protected] Principal investigator(s): Dubravko Rogale Research staff: Zvonko Dragcˇevic´, Gojko Nikolic´, Maja Vinkovic´, Snjezˇana Firsˇt Rogale, Slavenka Petrak, Goran Cˇubric´

Intelligent garment and environment Academic partners

Research register

Industrial partners

Project start date: 2007 Project end date: 2012 Project budget: e100 000 Source of support: Ministry of Science, Education and Sports Keywords: Intelligent garment, Thermal protection, Environment Investigations, construction and development of intelligent article of clothing related to its direct environment by developing an adaptable bed, adaptable ironing machine and measuring instrument for multiaxial testing physical-mechanical properties of technical textile and joined parts. The purpose of the project is that a research team makes researches resulting in a construction and realization of the first intelligent garment whose basic function is active thermal protection. It contains a sensor system for monitoring the values of air temperature inside and outside of the garment, data bus for data transfer, microcomputer and micro controller, and execution devices for the automatic regulation of thermal protection value. Controlling conduction and convection of the heat of the human body regulates thermal protection in such a way that based on anthropometric measurements several types of various air thermo insulation elastic chambers are constructed which are integrated into the construction of the garment between the outer shell and lining. Thermoinsulation chambers consist of several segments and have a twofold function so that by inflating sealing properties are assumed, and the heat loss of the human body by convection can be regulated and the thickness of the air chambers can be changed by program, whereby the heat loss of the human body by conduction can to be regulated. Micropneumatic elements and the chambers would be equipped with sensors of air pressure integrated into them, because depending on air pressure values in the chambers there will be defined chamber forms, their sealing properties and thickness on which thermal resistance depends. Investigations would prove that the integration and

IJCST 22,6

76

efficient joint operation of the integrated sensors, microcomputers with associated algorithms of intelligent behavior and actuators so that an independent action of the garment is realized with the aim of thermal protection whereby the garment would have the attribute of active, adaptable and intelligent behavior in variable temperature conditions. Communication possibilities of intelligent garment with the environment would be examined and an intelligent sick bed, adaptable ironing-machine and an instrument for testing load would be developed. They would practically use the same or very similar sensor, computer and micropneumatic actuator systems, connection techniques, constructions and design as well as intelligent garment.

Project aims and objectives The basic aim of the proposed research project is to investigate the possible construction and practical realization of an intelligent article of clothing with thermal protection, adaptable bed, ironing machine for the technological manufacturing process and necessary measuring instruments. The purpose of the investigation is to investigate characteristics of all elements of the system and behavior of the system as a whole and the communication between intelligent garment and environment. In addition to the basic aim of all investigations it is necessary to point out other aims too emerging as the result of the said investigation. Establishment of the leading European and world scientific role in investigations and development of intelligent garment. Writing 4 doctoral dissertations in the mentioned field (two dissertations in the mentioned field have been registered and approved by the Senate of the University of Zagreb. To prove that the Croatian clothing industry possesses a strong scientific research basis that guaranties it a technological excellence on demanding foreign markets.

Research deliverables (academic and industrial) The final assumed investigation results are practically applicable immediately upon completion of the project. It may be expected that the industrial production of intelligent garment with active thermal protection could be commenced immediately upon completion of the project. The examined and realized intelligent article of clothing would be a world unique item and might become an original Croatian product, especially from the point of view that original production principles have been protected by patent so that the future production rights are unquestionable. The developed intelligent article of clothing is very interesting for all the people being in extreme climatic conditions (soldiers, policemen, security services agents, construction workers, sailors, maintenance of roads, buildings and industrial facilities, drivers of trucks and construction machinery, athletes, recreationists, and other persons who wish to have such an article of clothing). In addition to the clothing industry, other industry branches such as mechanical engineers, electronics engineers, and programmers, participate in the production of intelligent garment, so that benefits can be expected for them too.

Publications and outputs Rogale, D., Firsˇt Rogale, S., Dragcˇevic´, Z., Nikolic´, G., “Intelligent Article of Clothing With an Active Thermal Protection”, European Patent Office, Munich, Germany, No. PCT/HR2004/000026. Firsˇt Rogale, S., Nikolic´, G., Dragcˇevic´, Z., Rogale, D., Bartosˇ, M., “Arhitecture of Clothing with an Active Thermal Protection. Proceedings of the 16th DAAAM International Symposium: Intelligent Manufacturing and Automation: Focus on Young Researchers and Scientists”, Katalinic´, Branko (ed.). Vienna, DAAAM International Vienna, 2005. 121-2. Rogale, D., Firsˇt Rogale, S., Dragcˇevic´, Z., Nikolic´, G., Bartosˇ, M.: “Development Of Intelligent Clothing With An Active Thermal Protection”, 6th World Textile Conference, 11-14 June 2006, North Carolina, 106-12. Petrak, S. and Rogale, D. “Methods of automatic computerised cutting pattern construction”. International Journal of Clothing Science and Technology, Vol. 13 Nos 3/4, pp. 228-39 (2001). Firsˇt Rogale, S., Dragcˇevic´, Z., Rogale D., “Determining Reaction Abilities of Sewing Machine Operators in Joining Curved Seams”, International Journal of Clothing Science and Technology, Vol. 15 Nos 3/4, pp. 179-88 (2003). Rogale D., Petrunic´ I., Dragcˇevic´ Z., Firsˇt Rogale S., “Equipment and methods used to investigate energy processing parameters of sewing technology operations”, International Journal of Clothing Science and Technology, Vol. 17 Nos 3/4, str., pp. 179-87 (2005). Petrak, S. and Rogale, D., “Systematic Representation and Application of a 3D computer-Aided Garment Construction Method, Part I 3D garment basic cut construction on a virtual body model”, International Journal of Clothing Science and Technology, Vol. 18 No. 3, pp. 179-87 (2006). Petrak, S., Rogale, D., Mandekic´-Botteri, V., “Systematic Representation and Application of a 3D computer-Aided Garment Construction Method, Part II Spatial transformation of 3D garment cut segments”, International Journal of Clothing Science and Technology, Vol. 18 No. 3, pp. 188-99 (2006). Firsˇt Rogale, Snjezˇana; Rogale, Dubravko; Dragcˇevic´, Zvonko; Nikolic´, Gojko; Bartosˇ, Milivoj, “Technical Systems in Intelligent Clothing with Active Thermal Protection”, International Journal of Clothing Science and Technology, Vol. 19 Nos 3/4, pp. 222-33 (2007). Firsˇt Rogale, Snjezˇana; Rogale, Dubravko; Dragcˇevic´, Zvonko; Nikolic´, Gojko; Runkas, Martin, Intelligent Clothing whit Programmabile Insulation. DAAAM International Scientific Book 2008, Branko Katalinic´ (ed.). Vienna: DAAAM International, 2008. Str. 273-86. Firsˇt Rogale, S., Rogale, D., Dragcˇevic´, Z., Nikolic´, G., Bartosˇ, M. “Technical Systems in Intelligent Clothing with Active Thermal Protection. Annual 2007 of Croatian Academy of Engineering”, Zlatko Kniewland (ed.). Zagreb: Croatian Academy of Engineering, 2007, pp. 301-17. Firsˇt Rogale S., Rogale, D., Dragcˇevic´, Z., Nikolic´, G., “Realization of the Prototype of Intelligent Article of Clothing with Active Thermal Protection”, Tekstil, Vol. 56 No. 10, pp. 610-26 (2007). Firsˇt Rogale, S., Rogale, D., Nikolic´, G., Dragcˇevic´, Z., Bartosˇ, M., “Chambers in the Intelligent Clothing with Active Thermal Protection”, Proceedings of 5th International Conference IMCEP 2007, Jelka Gersˇak (ed.). Maribor: University of Maribor Faculty of Mechanical Engineering, 2007, pp. 23-33. Nikolic´, G., Firsˇt Rogale, S., Rogale, D., Dragcˇevic´, Z., Bartosˇ, M., “Pneumatic System of the Intelligent Article of Clothing with Active Thermal Protection”, Ventil, Vol. 14 No. 6, pp. 552-6 (2008). Firsˇt Rogale, Snjezˇana; Rogale, Dubravko; Dragcˇevic´, Zvonko; Nikolic´, Gojko; Runkas, Martin, “Intelligent Clothing whit Programmabile Insulation”, DAAAM International Scientific Book 2008, Branko Katalinic´ (ed.). Vienna: DAAAM International, 2008, pp. 273-86. Firsˇt Rogale, S., Rogale, D., Nikolic´, G. and Dragcˇevic´, Z., “Controllable Ribbed Thermoinsulative Chamber of Continually Adjustable Thickness and its Application”, European Patent Office, Munich, Germany, No. PCT/HR2009/000008; 2009.

Research register

77

IJCST 22,6

78

ZAGREB, CROATIA University of Zagreb, Faculty of Textile Technology, Prilaz bazuna Filipovic´a 30, HR-10000 Zagreb, Croatia. Tel: +385 1 48 77 359; Fax: +385 1 48 77 355; E-mail: [email protected] ˚ urdica Parac-Osterman, PhD Principal investigator(s): Prof. A Research staff: Martinia Ira Glogar, PhD, Assist. Prof. Darko Golob, PhD, Assoc. Prof. Marija Gorensˇek, PhD, Assoc. Prof. Darko Grundler, PhD, Assoc. Prof. Nina Knesˇaurek, PhD, Prof. Nina Rezˇek-Wilson; Assist. Prof. Tomislav Rolich, ˚ urasˇevic´ BSc PhD, Ana Sutlovic´, MSc; Vedran A

COLOUR AND DYESTUFF IN PROCESSES OF ECOLOGICALLY ACCEPTABLE SUSTAINABLE DEVELOPMENT Academic partners

Industrial partners

JADRAN STOCKINGS FACTORY UNIVERSITY OF LJUBLJANA and MARIBOR, SLOVENIA Project start date: 1 January 2007 Project end date: 31 December 2011 Project budget: Source of support: Ministry of Science, Education and Sports, Republic of Croatia Keywords: Dyestuff selection, Nano-technology, Optimizing dyeing process, Purifying and decolouring wastewaters, Colour management, Fuzzy logic Scientific contribution to sustainable development relies on unlimited support of basic, developing and employable research. Therefore, selection of multi-functional dyes (UV protection, antibacterial protection, micro capsules of multi-functional performance), applying nano-technology in the dyeing processes with the aim of preventing water contamination, development of new methods as well as purifying dyed wastewaters contribute to sustainable development. Both input and output parameters of water will be controlled throughout the entire dyeing process: amount of residual dye in dye-bath using Lamber-Beer absorption model; X, Y, Z standard spectral characteristics of colour defined by specific absorption coefficient SAC and water quality defined by BOD5, COD, TOC, AOX, electrical conductivity and other defining values. System of control comprising advance models of control such as fuzzy logic (model based on rules) and model based on physical and chemical processes will be developed and applied. Capital area of research will involve models of dyeing processes, colour control and its correlation to dye as well as the interactive system of dye control. Models should describe and predict kinetics, reactivity, affinity, exhaustion, fixation and interaction of solutions containing various dyestuffs. Prediction of output process result as well as definition of both physical and chemical parameters crucial for controlling the process will be conducted based on afore-mentioned models. These models encompass kinetic models (according to Nernst and Langmuir) modified for interactions between dyestuff on fibre and in the solution.

Interdisciplinarity of dye within the system of sustainable development is based on spectral characteristics of colour as the fundament model dependent of the employment conditions. Instrumental measurement of colour is involved in all industrial production processes: textile technology, design, graphic industry and elsewhere which enables implementing control and colour harmonization. Application of evolutionary algorithms for modeling computer aided design of textiles based on principals of examinee’s subjective evaluation. Methods of descriptive statistics as well as methods of statistic reasoning will be applied within the frame of statistic analysis. Scientific affirmation of research results will be computer simulation as well as in vivo confirmation.

Project aims and objectives Contribution to sustainable development relies on unlimited support of basic, developing and employable research. Aim of the project is to contribute to humane ecology (regarding UV, antibacterial and other protective properties), through use of multifunctional dyes and selection of appropriate waste water discoloration and purifying methods, all in order of obtaining biological quality of water (free of toxic, aromatic components which may form in the process of dye degradation). Cognition of structure and use of thermo sensible nano sized dyes will enable their use on fibres for special purposes. Advance dyeing technologies, with the overview on pretreatment of textile substrates (enzymatic, plasmatic and other) in order of preserving environment and saving energy, will be applied. Base of the project is application of dyeing process control, including advance models of control; fuzzy logic and model based on physicochemical processes. Colour used as constant value will be applied for formation of fuzzy logic model, used for complex colour designing, automatic dye selection, direct transfer of colour coordinates data into the dyeing recipe setup system, advance recipe correction, as well as control and colour matching. Project research will enable use of new instrumental methods and development of researcher’s creativity, while graduate students and potential PhD students will be given a chance to get acquainted with scientific methodology, development of experimental skills and writing scientific papers.

Research deliverables (academic and industrial) Influence of dye’s chemical constitution and mode of dye-fibre bond onto antibacterial (e.g. Staphiloccoci, Escherichia), UVA and UVB protection properties. Influence of additives (electrolytes and surfactants) on dyeing process and degree of water pollution. Control of, in dye-bath and wastewaters, present electrolytes – elaboration of mathematical model. Further results considering influence of dye onto protection properties are expected. Application of thermo sensible dyes on children clothing. Influence of textile substrate’s pre-treatment (enzymatic, plasmatic pre-treatment, etc.) on dyeing kinetics and energy saving. Results of wastewater purifying and discoloration methods, with the emphasis on salt removal using physico-chemical methods, nano filtration and reverse osmosis. Colour as constant value of monitoring process, dye properties and colour matching in design applying evolutionary algorithms. Application of nano size particles. Influence of surfactants onto reactivity, affinity, exhaustion and fixation degree of reactive dyes. Advantages and disadvantages of

Research register

79

IJCST 22,6

80

physico-chemical decolouring methods. Dye degradation products and their toxicity (considering aromatic components) in wastewaters. Selection of dyestuff and its interaction with in the dye-bath present additives. Mathematical model based on measured values will be elaborated, while control system including a model based on physico-chemical processes will be applied. Fuzzy logic model, based on colour as constant value within control system will be worked out. Application of capsulated dyestuff and nano particles of zink and silver for special use (medical textiles). Pre-treatment of hydrophobic, synthetic fibres in the aim of increasing hydrophility and applicability of, in water soluble, dyes. From the economical and ecological aspect, a more acceptable system of purifying and decolouring wastewaters ii expected. Mathematical model based on the analyses of input and output measured values, considering coloured waters, will be elaborated, while a control system using advance models, such as fuzzy logic (model based on rules) and based on physicochemical processes model. Evolutionary algorithms for modelling computer design of fabrics based on principals of subjective asessment. Model must comply with standards, flexible, stabile, precise, and easily applicable. It includes complete process modelling: dyeing, colour control and its relation to dye. These models involve kinetics models (Nernst, Langmuir) modified for dye – dissolved dye. Applying CCM (computer colour matching) methods based on Kubelka-Munk theory, spectral characteristics and colour parameters according to CIEL*a*b* system, a model of fuzzy logic for complex design by colour, automatic dye selection, direct transfer of colour coordinates data into the dyeing recipe setup system, advance recipe correction, as well as control and colour matching, will be elaborated. Evolutionary algorithms for modelling computer design of fabrics based on principals of subjective asessment. Publications and outputs ˚ urdica Parac-Ostreman, Ana Sutlovic´, Vedran A ˚ urasˇevic´ and Tjasa Griessler Bulc, “Use of wetland for A dye-house waste waters purifying purposes”, Asian Journal of Water, Environment and Pollution, Vol. 4 No. 1, pp. 101-6. ˚ urdica Parac-Osterman, Vedran A ˚ urasˇevic´, Ana Sutlovic´: Comparison of Some Chemical and A Physical-Chemical Waste Water Discoloring Methods, Chemistry in Industry, in publication. ˚ urdica Parac-Osterman, Tomislav Rolich, “Fuzzy logic based Martinia Ira Glogar, Darko Grundler, A approach to textile surface structure influence in colour matching”, AATCC Review, in publication. ˚ urasˇevic´: Dyeing Properties of New Vesna Tralic´-Kulenovic´, Livio Racane, Ana Sutlovic´, Vedran A Benzothiazol Disperse Dyestuff, XX. Croatian Meeting of Chemists and Chemical Engineers, February 26-March 1, 2007, Zagreb, Croatia. ˚ urdica Parac-Ostreman, Nevenka Tkalec Makovec, Ana Sutlovic´, Ljerka Dugan: Staphylococcus A Aureus and Escherichia Coli Behavior on Undyed and Dyed Wool, XX. Croatian Meeting of Chemists and Chemical Engineers, February 26-March 1, 2007, Zagreb, Croatia. ˚ urdica Parac-Osterman, Ana Sutlovic´, Vedran A ˚ urasˇevic´: Application of Wetland System, Textile A Dyes Zagreb 2007, March 9, 2007, Zagreb, Croatia. ˚ urdica Parac-Osterman, Ana Sutlovic´, Martinia Ira Glogar: Dyeing Wool With Natural Dyes in Light A of the Technological Heritage, 7th annual Textile Conference by Autex: “From Emerging Innovations to Global Business”, 26-28 June 2007, Tampere, Finland. ˚ urdica Parac-Osterman, Ana Sutlovic´, Vedran A ˚ urasˇevic´: Application of Wool, CA and PP Fibers as A Filters in Wetland Pretreatment Media Formation, University of Zagreb, Faculty of Textile technology International Conference on Multi Functions of Wetland Systems, 26-29 June, Legnaro (Padova), Italy.

Zagreb, Croatia Faculty of Textile Technology, University of Zagreb, Prilaz baruna Filipovic´a 30, HR-10 000 Zagreb, Croatia. Tel: +38513712500; Fax: +38513712599; E-mail: [email protected] Principal investigator(s): Prof. Darko Ujevic´, PhD Research staff: Jadranka Akalovic´, BSc, Prof. Jadranka Bacˇic´; Prof. Zoltan Baracˇkai, PhD, Vinko Barisˇic´, BSc, Ing. Iva Berket; Bajro Bolic´, BSc, Blazˇenka Brlobasˇic´ Sˇajatovic´, BSc, Ksenija Dolezˇal, BSc, Mirko Drenovac, PhD, Prof. Milan Galovic´, PhD, Marijan Hrastinski, BSc, Renata Hrzˇenjak, BSc, M.D. Natasˇa Kaleboti; Prof. Isak Karabegovic´, PhD, Ivan Klanac, BSc, M.D. Irena KosTopic´; Prof. Tonc´i Lazibat, PhD, Nikol Margetic´, BSc, Prof. Zlatka Mencl-Bajs; M.D. Zˇeljko Mimica; Prof. Gojko Nikolic´, PhD, Alem Orlic´, BSc, PhD M.D. Vedrana Petrovecˇki; BSc M.E. Zˇeljko Petrovic´; Prof. Dubravko Rogale, PhD, Prof. Andrea Russo; Igor Sutlovic´, PhD, Prof. Vlasta Szirovicza, PhD, Irena Sˇabaric´, BSc, MSc. M.D. Nadica Sˇkreb-Rakijasˇic´, Marija Sˇutina, BSc, Prof. Larry C. Wadsworth, PhD

Anthropometric measurements and adaptation of garment size system Academic partners

Industrial partners

Project start date: 1 January 2007 Project end date: 31 December 2011 Project budget: Source of support: Ministry of Science, Education and Sports, Republic of Croatia Keywords: Anthropometric measurements, Garment size system Systematic anthropometric surveys have been conducted since 1901 with the aim of developing and improving systems for clothing and footwear sizes. The measurement results show how a national population changes over a period of several decades in physical build and size due to a series of factors (food habits, sports development, genetic predispositions, population migrations, climatic conditions, etc.). Based on the results of anthropometric measurements in the Republic of Croatia (2004/05) on the sample of 30,866 test persons aged between 1 and 82 a statistical analysis of body measurements was performed, a database including 5 basic studies of sex and age as well as a new standard for clothing and footwear was built. These results enable a significant and stimulating continuation of scientific research and a comparison to other national standards and their contributions to the creation of systems for clothing and footwear sizes. Elements common for national standards of garment sizing by an exact approach will be investigated and analyzed, in particular because the presumptions of national systems and starting elements, respectively, are not universally founded like intersize

Research register

81

IJCST 22,6

82

intervals which differ in sizes since the conformity of individual starting places is missing. Data will be provided for a common base with methods of body measuring and size designation of clothes according to the recommendations of the Technical Committee TC133 within ISO and EN standards as well as the design and development of a sophisticated computer system (DOV-KO) for unifying all body measurements and basic garment construction based on one or all other sizes. Within the scope of this project and based on experience, a very important cycle of anthropometric measurements of the sporting population in football, water polo, rowing, basketball and handball will be performed. 4,000 test persons from Zagreb, Osijek, Rijeka, Split and Dubrovnik will be measured, whereby specific body differences and deformations of muscles caused by longstanding training will be analyzed. A comparative analysis of the representative sample of the anthropometric measurements of sportsmen and other population as well as the investigation of other trends of body measurements will be performed. This will enable an exceptional insight into the anthropometric dimensions which reflect body shape, proportionality, composition and elements of success in sports, respectively. Stadiometar or a new measuring instrument for continuous measuring body height, foot length and width will be designed too.

Project aims and objectives Problems of garment sizing and fit affect the market globally, and a consequence of bad predictions of the quantity of necessary stocks for manufacturers and dealers poses a risk of high costs. In the case of domestic manufacturers samples may be additionally divided. Particular solutions may be considered more efficiently by interpreting the data from the anthropometric database connecting 5 studies according to sex and age and the system of sizing. Therefore, using the results of the anthropometric measurements taken and the basic projection of the new standard for clothing and footwear, one of the directives of the project is to investigate other national standards of Europe and the world and to create size intervals and a new Croatian standard. The study of body differences and specific deformities of the body muscles during the longstanding practice of athletes such as leg circumference, chest circumference, torso, shoulder width, arm and leg length, body height, palm length is an additional aim of this project which will show the body elements affecting success level in sport. Besides a greater adaptation of clothing and footwear to the home market, it would be advisable to ensure the continuity of investigating the national size standard by creating a sustainable Croatian system of clothing and footwear sizes in conformity with anthropometric surveys that are conducted periodically and systematically in developed countries in which a change in the morphology of the human body occurred over the last decades. By way of proof, systematic anthropometric measurements and sizing in France showed a tendency of average height growth. In Great Britain in female population a growth of bust circumference was recognized, whereas in USA studies point to the tendency of an increase in obesity (besides an aesthetic also a health problem). The interest of the scientific and professional public, manufacturers, trades people and consumers in sizing will continue to grow, since a faster change in established body proportions may be expected thanks to changes in living and food habits of the

population, an unavoidable mingling of ethnic groups, increase in the number of older consumers of clothing and footwear which will be doubled in the next two decades, etc. Thus, it is additionally stressed how much it essential at the moment to ensure a valid starting point or a Croatian standard to pursue next movements of body dimensions in order to avoid a discrepancy and imposition of the specificities of domicile consumers to home and foreign manufacturers.

Research register

83 Research deliverables (academic and industrial) Anthropometry is the study of the measurement of the human body, but Pheasant has expanded it as “applied anthropometry” including quantitative data of size, forms and other physical characteristics of people that can be used in garment design. Since the form of the human body changed through time, the problem of ageing proved to make a contribution to perceived changes in body shape and size more than any other individual factor, such as for example improved nutrition and prolonged life, in particular the knowledge that the number of older consumers will be doubled by 2003. Therefore, the systems of sizing shall be updated periodically to ensure a correct fit of ready-made clothing. On the other hand, the home industry of clothing, fashion wear and footwear disposes of modest and aged data based on the out-of-date anthropometric measurements from 1962. It was therefore necessary to conduct a new cycle of anthropometric measurements and to use the obtained results. World fashion industry shows a special interest in measuring anthropometric characteristics of the population so that it gathers such data permanently, motivated by the wish for designing articles of clothing for all population groups, including the persons with pronounced specificities (higher stature, higher body weight, etc.). The use of investigations will contribute to creating a new and modern Croatian standard for clothing and footwear harmonized with ISO and EN standards. Besides the clothing and footwear industry, pediatricians, specialists of occupational and sports medicine, experts in wood processing industry, automotive industry, in the army and police will benefit from the investigation results. Teachers and students in undergraduate and graduate studies as well as teachers and pupils at technical schools will benefit from the development results of the computer system based on the selection of garment sizes. By using the investigations of the sporting population, one can get an insight into tendencies of diversities of body measurements and changes in muscles as a result of longstanding practice. Various specialists of sports medicine, orthopedists, garment and footwear designers will benefit form the results of this investigation because based on previous experience it is evident that mass customization is necessary for athletes. Knowing dimensional characteristics, this method would be considerably promoted and improved. Publications and outputs Ujevic´, D., Rogale, D., Hrastinski, M., Drenovac, M., Szirovicza, L., Lazibat, T., Bacˇic´, J., Prebeg, Zˇ., Mencl-Bajs, Z., Mujkic´, A., Sˇutina, M., Klanac, I., Brlobasˇic´ Sˇajatovic´, B., Dolezˇal, K., Hrzˇenjak, R. (2006), “Normizacija, antropometrijski pregledi i Hrvatski antropometrijski sustav”, Tekstil, Vol. 55 No. 10, pp. 516-26. Ujevic´, D., Firsˇt-Rogale, S., Nikolic´, G., Rogale, D. (2006), “Pregled razvojnih dostignuc´a u tehnologiji sˇivanja – IMB 2006”, Tekstil, Vol. 55 No. 12, pp. 624-31.

IJCST 22,6

84

Ujevic´, D., Dolezˇal, K., Lesˇina, M. (2007), “Analiza antropometrijskih izmjera za obuc´arsku industriju”, Poslovna izvrsnost, Vol. 1 No. 1, pp. 171-83. Ujevic´, D., Hrzˇenjak, R., Dolezˇal, K., Brlobasˇic´ Sˇajatovic´, B. (2007), “Hrvatski antropometrijski sustav – jucˇer, danas, sutra”, HZN Glasilo, Vol. 3 No. 1, pp. 5-10. Hrzˇenjak, R., Ujevic´, D., Dolezˇal, K., Brlobasˇic´ Sˇajatovic´, B. (2007), “Investigation of anthropometric characteristics and body proportions in the Republic of Croatia”, Proceedings of 7th Annual Textile Conference by Autex, Tampere, Finland, 25-28 June, pp. 1191-8. Ujevic´, D., Brlobasˇic´ Sˇajatovic´, B., Dolezˇal, K., Hrzˇenjak, R., Mujkic´, A. (2007), Rezultati prvog antropometrijskog mjerenja stanovnisˇtva Republike Hrvatske, Drugi kongres hrvatskih znanstvenika iz domovine i inozemstva, Split, Croatia, 5-10 May. Nikolic´, G., Ujevic´, D. (2007), Protractor for Measuring Shoulder Slope, Patent. Ujevic´, D. (2007), One-arm and/or two-arm anthropometer, Patent.

Zagreb, Croatia Faculty of Textile Technology, University of Zagreb, Prilaz baruna Filipovic´a 30, HR-10 000 Zagreb, Croatia. Tel: +385 1 48 77 352; Fax: +385 1 48 77 352; E-mail: [email protected] Principal investigator(s): Prof. Drago Katovic´, PhD Research staff: Asoc. Prof. Sandra Bischof Vukusˇic´, PhD, Prof. emeritus Ivo Soljacˇic´, PhD, Dubravka Dosˇen Sˇver, PhD., Sandra Flincˇec Grgac, BSc, Asoc. Prof. Radovan Despot, PhD, Asist. Prof. Jelena Trajkovic´, PhD, Asist. Prof. Branka Lozo, PhD, Luka avara, MSc, Bozˇo Tomic´, MSc, Prof. Charles Yang, PhD, Prof. Christian Schram, PhD

Alternative eco-friendly processing and methods of cellulose chemical modification Academic partners

Industrial partners

Faculty of Forestry, Croatia; Faculty of Cˇateks, d.d., www.cateks.hr Graphic Art, Croatia; University of Georgia, USA; University of Innsbruck, Austria Project start date: 1 January 2007 Project end date: 31 December 2011. Project budget: Source of support: Ministry of Science, Education and Sports, Republic of Croatia Keywords: Multifunctional eco-friendly textile finishing, Polycarboxylic acids, Protective functionalities, Chemical modification of cellulose, Microwave treatment of cellulose materials One of the requests of European Union for higher competitiveness of European market is rebuilding and reconstruction of traditional industrial sectors, especially textile and wood industry. According to the strategical goals of the Republic of Croatia the project

emphasizes the use of highly sofisticated production processes and treatments of cellulose materials, i.e. obtaining additional and improved characteristics of wooden and paper materials which can be achieved by using high-tech processes and by introduction of nanomicro- and bio-technologies. One of the alternative methods for replacing the conventional reactants containing formaldehyde which were used in textile and wood treatments so far, would be the modification with eco-friendly agents such as polycarboxylic acids. Efficiency of these treatments will be determined quantitatively by ester crosslinking analytical methods or by means of isocratic HPLC and spectrophotometric FTIR method. Standard methods of textile, wood and paper material testing would be used for examining their protective performance and resistance to weathering conditions. Part of the proposed project will be development of optional multifunctional treatment that would provide better protection of cellulose materials against microorganisms, UV, electromagnetic rays, flame, oil or water. Therefore, a particular attention will be payed to development and application of the agents which will not only improve the characteristics of textile materials but also give it permanent freshness and provide additional care and protection, i.e. medical characteristics. Optimisation of alternative processing and methods will provide ecologically and economically favorable characteristics of treated materials. Further process optimisation in order to improve processing quality could be obtained with new alternative method using microwave energy. Improved characteristics obtained with this method in our previous research confirm its usability in textile finishing processes as well as in chemical modification of wood. Previous research in this field represent worldwide novelty which should be by all means continued.

Project aims and objectives The purpose and aim of the proposed project is to obtain highly valuable and multifunctional treated textile materials that will acquire analogous price on the demanding market. This is the basic condition for the survival of Croatian textile, wood and paper industry on EU market. In textile area experiments will be conducted to obtain multifunctional environmentally friendly textile material which will simultaneously offer dimensional stability, flame retardancy, crease and antimicrobial resistance and will have no effects on human health. Further goal is to obtain chemically modified wood that will have reduced shrinking and water absorption as well as to obtain flame retardancy on wood and paper products. One of the equally important goals is construction of a semi industrial microwave device for continuous planar treatment of cellulose materials. The results obtained would be presented in the world best known papers in the relevant field. The most important goal of the project is affirmation of Croatian science in Europe and rest of the World, by presenting the results in international papers so as on International Conferences. It is important to stress that established cooperation with EU and USA experts, so as with their scientific institutions will be continued and expanded. In this project, where will scientists from abroad have an active contribution with their work, further contribution to development of high quality products will be added. We certainly hope it will affect development of Croatian industry and economy.

Research register

85

IJCST 22,6

86

Publications and outputs Bischof Vukusˇic´, S., Flinecˇ Grgac, S. and Katovic´, D. (2007), “Antimicrobial Textile Treatment and Problems of Testing Methods”, Tekstil, Vol. 56, accepted for publication. Bischof Vukusˇic´, S., Flincˇec Grgac, S. and Katovic´, D. (2007), “Catalyst influence in low formaldehyde flame retardant finishing system”, 7th AUTEX Conference, Tampere, pp. 60-1. Flincˇec Grgac, S., Katovic´, D. and Bischof Vukusˇic´, S. (2007), “Combination of organophosphorus agent and Citric acid in Durable Press Finishing of Cellulose Fabrics”, XX. Croatian Society of Chemical Engineers, Zagreb, Croatia, pp. 281. Katovic´, D., Bischof Vukusˇic´, S. and Flincˇec Grgac, S. (2007), “Crosslinking cotton with citric acid and organophosphorus agent for the purpose of flame retardant finishing”, 85th Textile Institute Conference, Colombo, Sri Lanka, pp. 820-4.

Zagreb, Croatia Faculty of Textile Technology, University of Zagreb, Prilaz baruna Filipovic´a 30, HR-10 000 Zagreb, Croatia. Tel: +385 1 4877 360; Fax: +385 1 4877 355; E-mail: [email protected] Principal investigator(s): Prof. Ana Marija Grancaric´, PhD Research staff: Assoc. Prof. Tanja Pusˇic´, PhD, Assist. Prof. Zˇeljko Penava, PhD, Anita Tarbuk, MSc, Lea Markovic´, BSc, Assist. Prof. Jasenka Bisˇc´an, PhD, Sonja Besˇenski, MSc, Ivancˇica Kovacˇek, PhD, D. Med., Prof. Djamal Akbarov, PhD, Prof. Emil Chibowski, PhD, Prof. Rybicki Edward, PhD, Prof. Eckhard Schollmeyer, PhD, Prof. M.M.C.G. Warmoeskerken, PhD

Interface phenomena of active multifunctional textile materials Academic partners

Industrial partners

Pamucˇna industrija Duga Resa, Duga Resa Croatian National Institute of Public Health, Zagreb; Tashkent Institute of Textile and Light Industry, Uzbekistan; Maria Curie-Skłodowska University, Lublin, Poland; Technical University of Lodz, Poland; Deutsches Textilforschungsinstitut Nord-West eV; Institut der Universitat Duisburg Essen; University of Twente, Netherlands Project start date: 1 January 2007 Project end date: 31 December 2011 Project budget: Source of support: Ministry of Science, Education and Sports, Republic of Croatia

Keywords: Textile material, Interface phenomena, Surface modification and finishing, Multifunctionality The goal of the project is synergistic effects of some compounds on modified textile surfaces for achieving multifunctionality of textiles. Interface phenomena of textile surfaces with special accent on surface free energy, zeta potential, electroconductivity, adsorption and desorption of surfactants and other compounds usually used in textile finishing will give a great contribution to multifunctionality of textile. The mechanism of adsorption and desorption of surfactants and other finishing agents on modified textile surfaces is expected to be clarified in the present project. Different surface modifications, pretreatment and finishing of textile, especially cotton and polyester, will be performed according to European Technology Platform for the future of textile and clothing. For such purpose advance processes like mercerization, cationization, alkali, EDTA, other compounds and enzymes for surface hydrolysis of PET fabric, optical bleaching, implementation of nano antimicrobial active silver ions and mineral delivery mechanism, zeolite and others will be performed. Aminofunctional and other compounds will be added to azalides for the synergistic high antimicrobial effects. In cotton pretreatment enzymatic scouring will be applied using enzymes pectinase and the newest cutinase, for removal of pectins and bioplymers from cotton impurities with lipophylic character, instead of ecologically unfavorable alkali scouring. The goal of the project is synergistic effects of some compounds on modified textile surface. Interface phenomena of the new textile materials produced from electroconductive, low electro resistance fibers will be investigated for the purpose of static electricity and electromagnetic protection and for its implementation as sensors or other electronic devices in intelligent textiles. Traditional protection and aesthetic role of textile will be spread in active textile multifunctionality. Project will deal with elektrokinetic phenomena (zeta potential, isoelectric point, IEP, point of zero charge, PZC, surface electrical charge, surface free energy), hydrophility and hydrophobilicity, whiteness, fluorescence and phosphorescence, friction, fabric cover factor, elasticity, air and water vapor permeability of textile materials and their protection on UV radiation, microbes and fungi, coldness, heat and flame, static electricity and electromagnetic field.

Project aims and objectives Project will continue researching on assignments from previous project (0117012). Purpose of these investigations is based on lightening of interface phenomena on textile which effect directly to its adsorption and interaction intensity between textile fibers and chemical compounds. Almost all possibilities in modification during manufacturing high performance synthetic fibers are used, therefore nowadays attention and research is on textile surface modification. Procedures and compounds for that modification varies, as their effect varies, but the purpose and aim are directed to synergism of two or more components for accomplishing hydrophob or hydrophil textile, textile highly resistant to atmospheric condition, bacteria, microbe and fungi, UV radiation and open flame. Furthermore, important aim of the project is cotton high level of purity by unconventional agents and material pretreatment procedures for mercerization and

Research register

87

IJCST 22,6

88

cationization. Pectinase in previous project investigation showed good elimination of pectine from primary cotton layer, but hydrophility was not so high like alkali scoured cotton. Chemical composition of cotton cuticula has lypophilic polymers, biopolyesters, which can be degraded by cutinase, new enzymes for degradation of waxes for better hydrophility. Cotton cationization during mercerization is the most important innovation of previous project and the patent for it was asked. Electronegative cotton surface charge, of which anionic substances adsorption depends, is lower after cationization in harsh mercerization conditions. The aim of this project is antibacterial, UV and flame protection by nanoparticle implementation (Ag) using mineral delivery compound (zeolite and others) as well. Electroconductive fibers implementation in yarns of textile materials should result in static electricity removal, and hopefully other effects. The aim of polyester surface modification, optical bleaching, other compounds treatments is well-known aesthetic, as well as high UV protection, high material elasticity as a result of changes in fiber microstructure. Interface phenomena research on wide range possible fabric knitted and woven construction will give the solution of problems of fabric construction influence to high effect in this project.

Research deliverables (academic and industrial) Interface phenomena of textile materials surface in wet medium results in textile electric surface charge cognition and surface free energy as well on which adsorption depends. Important application of this project results is in ecological enzymatic scouring with pectinases. Enzymatic scouring with new enzymes, cutinase, will remove biopolyester cuticula and improve cotton hydrophility, and therefore replace harsh conventional alkali scouring entirely. Important application will have, patent requested cotton cationization during mercerization. By this pretreatment electropositive cotton is achieved, with great anion adsorption on its surface in all textile finishing processes. These anions enclose all low and high molecular compounds for textile finishing and all pricondensates. Implementation of nanoparticles (Ag and others) is predicted during mercerization and cationization processes, therefore it is important to emphasize rational component of these procedures which gives cotton multifunctionality in all textile usage. The next important application is antibacterial textile accomplished with azalide treatment especially in synergism with aminofunctional and other compounds and systems. It is well-known that fluorescence of optically bleached increases whitening of textiles. Optical brighteners and other compounds researching will be of great importance in UV protection with textile material. Heavy metals are toxic and their research is of great importance in human health protection. Furthermore, in nowadays growing demands on life safety from external influences especially UV radiation, research of differently structured textile material interface phenomena will find application in textile for summer clothing.

Publications and outputs Ana Marija, G., Anita, T., and Ivancˇica K. Micro and nanoparticles of Zeolite for the protective Textiles. Book of Proceedings of 7th Annual AUTEX Conference, AUTEX 2007, P-1123, Tampere, Finland. Ana Marija, G., Lea M., Anita, T. and Eckhard, S. Properties of Multifunctional Cotton in Accordance with International Standards. Conference of Textile days, Zagreb 2007.

Anita, T., Ana Marija, G., and Mirela, L. Surface Free Energy of Pretreated and Modified Cotton Woven Fabric. Book of Proceedings of 7th Annual AUTEX Conference, AUTEX 2007, P-1104, Tampere, Finland. Anita, T., Ana Marija, G. and Volker, R, “Electrokinetic Phenomena of Textile Fibers. Book of abstracts XX”, Croatian Meeting of Chemists and Chemical engineers 2007, 301. Grancaric´, A., Pusˇic´, T. and Tarbuk, A. “Enzymatic Scouring for Better Textile Properties of Knitted Cotton Fabrics”. Biotechnology in Textile Processing, Guebitz, Georg; Cavaco-Paulo, Artur; Kozlowski, Rysard (ED.). New York: The Haworth Press, Inc., 2006. Grancaric´ Ana, M., Tarbuk, A., Dumitrescu, I. and Bisˇc´an, J. “UV Protection of Pretreated Cotton – Influence of FWA’s Fluorescence”, AATCC Review, Vol. 6 No. 4, pp. 2-6 (2006).

Research register

89

Research index by institution

IJCST 22,6 Institution

90

Page

College of Sliven

60

Cornell university

42

Dokuz Eylul University

44, 45, 46, 47, 48

Heriot Watt university

8, 9, 10, 11, 13, 15, 16

Jiangnan University

63

Konkuk University

59

Loughborough University

49

Sci-Tex Tainan University o Technology The Hong Kong Polytechnic University University of Delaware University of Gent

54, 55 62 38, 39, 41 53 17, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 31, 32, 33, 35, 36, 37

University of Maribor

51

University of Pisa

56

University of Ulster

6

University of Zagreb

64, 66, 68, 72, 74, 77, 80, 83, 85

Research index by country Country

Page

Belgium

17, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 31, 32, 33, 35, 36, 37

Bulgaria

60

China

63

Croatia

64, 66, 68, 72, 74, 77, 80, 83, 85

England, UK Hong Kong (ROC) Japan

49 38,39,41 54, 55

Ireland, UK

6

Italy

56

Korea

59

Scotland, UK

8, 9, 10, 11, 13, 15,16

Slovenia

51

Taiwan (ROC)

62

Turkey USA

44, 45, 46, 47, 48 42, 53

Index by country

91

Research index by subject

IJCST 22,6

92

Carpets Wear, Evaluation, Image processing Colour Dyestuff selection, Fuzzy logic, ISO colour, Dyes, pH sensitive dyes Environmental textiles Toxicology, Allergenic properties, Ecology, Medical textiles, Multifunctional textiles, Protective textiles, Microwave, Cellulosic, Protective clothing, Smart textiles

28 9, 29

6, 40, 71, 73, 82, 55, 20

Fabric composites 62, 63, 3D woven fabric, 65, 47, Binder yarn, 25, 6 Clothing technology, Reinforcement, Nonwovens, Impact fabrics, Knitted fabric, Renewable natural fibre composites Fibres 31, 21, Hollow fibre, Cotton, Morphology, Thermal analysis, Spectroscopy, 25, 22, 25 Microscopy, Natural fibre composite Garment sizing Anthropometric measurement, Made to measure systems, Body shape, 3D body scanning Ink jet printing pH sensitive dyes

79, 41

9, 29

Medical textiles 6, 71, Biotechnology, 32, 19, Enzymes, Bio textiles, 21, 25, Insect repellent textiles, 23, 26 Release system, Bioengineering of cotton fibre, End uses, Bio-functional material, Self-assembling peptides, Wound dressing, Electrotherapy

Nanotechnology Nanofibres, Nanoclays, Biopolymer nanostructure, Nanoparticles, Surface properties, Filtration, Electrospinning, end uses)

24, 30, 32, 33, 55, 76, 35, 14, 26

Networking

29

Plasma

44

Protective clothing Intelligent garments

73

Simulation 10, 13, 3D, Cloth, Particle based method, 50, 58 Collision detection, Fabric mechanics, Fabric drape, Comfort, Yarn, Fabric, Garment, Modelling high performance, Filtration, CFD, CAE Smart Textiles Apparel sol-gel, Photovoltaics, Multifunctional, Protective textiles, Ceramic coating, Sol-gel process, Surface modification, Finishing, Physiological load, Sensors, Cooling systems, Functionalisation, Grafting Technical textiles Flame retardant textiles, Artificial turf

8, 15, 17, 19, 29, 61, 65, 68, 84, 40, 49, 55, 32, 36, 23, 28

53, 30, 31, 20

Technology transfer

32

Synthetic ropes Dynamic behaviour, Extreme conditions

34

Wearable electronics Smart fabrics, Garments, Fabric sensors, 3D pressure sensors, POF,

15, 17, 37, 38, 44, 45,

FBG, Deformation textile sensor, 49, 55, 36, 28 E-textiles, Textile antennas, Stretchable electronics, Wearable garments, Conductive textiles, Fibre conductivity Weaving Air jet, Weft preparation system, Multi rapier technology, 3D weaving

35, 20

Yarns Friction, Force, Friction coefficient, Blended, Hairiness, Unevenness, Tencel, Modal, Sirospun, Cotton yarns

43, 47, 59

Index by subject

93

Research index by principal investigator

IJCST 22,6

94

Principal investigator

Page

Arif Kurbak

47

Liu Jihong

63

Ashdown P. Suzan

42

Mather, R.R.

8

Ayse Okur

48

Miao-Tzu Lin

62

Parac-Osterman Aurdica

77

Aysun Cireli Aksit

45, 46

Christie, R.H.

10

Rahnev Ivelin

60

Danila De Rossi

56

Rita Chang

53

De Clerck Karen

22, 23, 24, 25, 31, 32, 37

Samodi Zeljko

64

Skenderi Zenum

66

Dubravko Rogale

74

Soden Julie

8

Emira Pezelj

68

Soljacic Ivo

72

Gersak Jelka

51

Stylios GK

11,13,15,16

Grancaric, A.M.

85

Tao Xiaoming

38, 39, 41

Havenith George

49

Tatsuki Matsuo

Huantian Cao

53

Van Langenhove Lieva

Hwa Kyung Song

42

InHwan Sul

59

Vildan Sular

Jennifer McCord

53

Wardman, R.H.

83

Westbroek Philippe

24

Wilson, J.I.B.

8

Katonic Drago Kiekens, P.

29,31,33,35

54, 55 17, 20, 21, 27, 28, 29, 35, 36, 37 44 9, 10