Warp knitting of metal fibre cloths for use as separation material in automotive glass production

Transcription

1 Warp knitting of metal fibre cloths for use as separation material in automotive glass production Daan De Keyzer Promotors: prof. dr. ir. Lieva Van Langenhove, dr. ir. Filip Lanckmans Thesis submitted to obtain the degree of Master of Science in Textile Engineering Department of Textiles Chairman: prof. dr. Paul Kiekens Faculty of Engineering Academic Year

2 Acknowledgement Hereby I would like to thank everyone who contributed to the accomplishment of my thesis. First I would like to thank my external promotor at Bekaert, dr. ir. Filip Lanckmans, for the extensive guidance during my research and the time spent on reviewing my writings. Thank you also, Frank De Ridder, for your valuable input and advice. I would like to thank my internal promotor, prof. dr. ir. Lieva Van Langenhove, for providing me the opportunity to choose this topic, and for the useful advice throughout the research. Furthermore, I would like to thank prof. dr. Yordan Kyosev, for providing me the license to his program TexMind. It proved to be very useful for pattern selection of warp knits. I would like to express my appreciation to all the people at Bekintex that helped me during my practical tests. Thanks to the people from the R&D team, for all the amusing distractions and the pleasant atmosphere. Thanks to all my E-team classmates for two years of unforgettable moments. It was truly an incredible experience. Special thanks go to Sam, for reviewing this work. Thank you Charlotte, my girlfriend, for being at my side in Istanbul and motivating me whenever needed. Also thank you Linde, my sister, for your practical and moral support. Finally I would like to thank my parents, for making it possible to follow E-team and supporting me during all the years: nen grote merci. i

3 Copyright notice In so far as allowed by the non-disclosure agreement of 18 January 2012 between Bekintex NV, Universiteit Gent and Daan De Keyzer; the author gives permission to make this master dissertation available for consultation and to copy parts of this master dissertation for personal use. In the case of any other use, the limitations of the copyright have to be respected, in particular with regard to the obligation to state explicitly the source when quoting results from this master dissertation. Daan De Keyzer May 31, 2012 ii

4 Warp knitting of metal fibre cloths for use as separation material in automotive glass production By Daan DE KEYZER Promoters: prof. dr. ir. Lieva Van Langenhove, dr. ir. Filip Lanckmans Thesis submitted to obtain the academic degree of Master of Science in Textile Engineering Summary Department of Textiles Chairman: prof. dr. Paul Kiekens Faculty of Engineering Academic Year In this thesis, warp knitted metal fibre cloths for use as separation material in automotive glass production are investigated. The separation materials currently used are made by circular weft knitting technology. Warp knitted fabrics are structurally very different and this results in a completely different deformability, air permeability and thickness, all important for the application as separation material. The aim of this thesis is to investigate the effect of structural warp knitting parameters on the fabric properties. Various types of warp knitted structures are selected and knitted. Two structural parameters are found to influence the fabric deformability, the machine gauge and take-off speed, which determine the loop density in respectively wale and course direction of the fabric. Another parameter is the movement variation of the guide bars, which influences the length of the underlap and adds stability to the fabric when increased. Next, the presence of inlay threads is found to influence the deformability and air permeability. Finally, double needle bed structures are found to be the least promising method to tailor the fabric properties. Two warp knitted structures show promising properties when compared to the reference weft knitted sample, and resulted in a patent application. These two fabrics can be further tested, and improved by altering the investigated parameters. Keywords: warp knitting, metal fibre, deformability, air permeability iii

5 Warp Knitting of Metal Fibre Cloths for use as Separation Material in Automotive Glass Production Daan De Keyzer Supervisors: prof. dr. ir. Lieva Van Langenhove, dr. ir. Filip Lanckmans Abstract - This research describes the properties of warp knitted metal fibre fabrics that can be used as heat resistant separation material in automotive glass production. Various types of warp knitted structures are selected and knitted. The fabrics are analysed to assess the potential of warp knitted structures for automotive glass production. Keywords - warp knitting, metal fibre, deformability, air permeability F I. INTRODUCTION abrics made out of stainless steel (SS) fibres are used in the production process of press bended automotive glass. The process consists of pressing flat glass around a mould at high temperatures ( C). Direct contact between the glass and the mould would lead to optical distortions and defects such as scratches. To avoid this, knitted stainless steel fibre fabrics are used as a heat resistant separation material (HRSM) to cover the mould and protect the glass. Due to the direct contact between the fabric and the glass, the structural fabric characteristics strongly influence the quality of the formed glass. The increasing complexity of the glass shapes and higher optical requirements for windows have led to a demand for a new generation of HRSM fabrics. The currently used technology, circular weft knitting, does not provide enough patterning possibilities to meet the demand for improved fabrics. An alternative technology with more patterning possibilities is warp knitting, but it also has an inherently different structure, as shown in Figure 1. Figure 1: Comparison of a) warp and b) weft knitted structure [1]. This structural difference results in a completely different deformability, air permeability and thickness, all important for the application as HRSM. The aim of this thesis is to investigate the effect of structural warp-knitting parameters on the fabric properties. II. IMPORTANT HRSM FABRIC PROPERTIES A. Fabric deformability The deformability is determined by the tensile force at break and elongation of the fabric. Knitted fabrics have anisotropic properties. Generally the fabrics are measured in course and wale direction. The leading car glass manufacturers have set a standard of 49 N for the tensile force at break in each direction. The elongation at 49 N is ideally between 20 to 40 % in both fabric directions. Higher elongations can possibly lead to overstretching the fabric when draping it on the mould. This deforms the structure and results in a lower optical quality. The ratio of wale over course elongation at break (w/c) is preferably close to one, this enables good drape ability without structural deformations of the fabric over the mould. B. Air permeability The air permeability of the fabric determines the necessary vacuum pressure in the process to form the bended glass. When the air permeability is too low, the process complexity increases and the window may not be shaped correctly. The preferred value is considered to be 800 l/(dm 2.min). C. Fabric thickness The fabric thickness influences the necessary mould and glass temperature to form the bended glass. Thicker fabrics will slow down the heat transfer from the mould to the glass. III. EFFECT OF WARP KNITTING PARAMETERS A. Sample specifications All samples were made on a Karl Mayer Raschel warp knitting machine with gauge twelve. The yarn used is a stainless steel (AISI 316L) fibre with Nm 15/2 yarn count. B. Effect of the underlap length A warp knitted loop consists of an overlap and underlap. The underlap length is determined by the sideways shogging movement of the guide bar. As can be found in the literature, single guide bar (GB) structures are dimensionally to unstable and split easily when damaged [2]. Warp knitted structures knitted with two guide bars are more stable and already provide many patterning possibilities. Different patterns are made by changing the sideways movement of both guide bars. The movement is defined by the number of needles lapped during the shog. The set of samples consists of patterns with a different value for both guide bars. Two statistical design of experiments (DOE) were performed on the samples to determine the effect of varying GB 1 and/or 2. The results show that the underlap length has a statistical significant influence on the elongation at 49 N in course direction. The course elongation at 49 N for satin-pillar was significantly lower compared to cord-pillar, respectively 88% and 112%. In wale direction there is a slight decrease when changing from pillar to tricot on GB 2, while keeping a tricot stitch on GB 1.

6 The tensile force at break in wale direction is mainly influenced by the machine gauge: it will increase for a higher gauge. In course direction it increases for a longer underlap. When comparing tricot-pillar with satin-pillar the course tensile force at break is respectively 19 N versus 170 N. Longer underlaps result in a closer and tighter structure, which decrease the air permeability. However, all structures are well above the preferred value of 800 l/(dm 2.min). The presence of an underlap between more than two wales adds an extra thread on the fabric surface, which increases the thickness. An even longer underlap however, will not further increase the thickness. C. Effect of inlay threads The insertion of inlay threads is the second structural parameter used to obtain a dimensionally stable fabric. The properties of the satin-pillar structure were compared with a pattern consisting of a pillar and inlay thread over four needles (Figure 2). tricot and cord, were knitted and compared to the promising single needle bed samples (satin-pillar and pillar-inlay). The results show that these structures do not give any improved properties compared to the single needle bed samples. The thickness and elongations at break are almost twice as high, which makes them not so interesting for HRSM fabrics. IV. COMPARISON WEFT-WARP KNITTED HRSM The two most promising patterns for HRSM fabrics are the satin-pillar and pillar-inlay patterns. Although the pillar-inlay pattern has better deformation properties, the satin-pillar is interesting for its relative high air permeability for a thick fabric, which could result in good wear properties. The course elongation at 49 N could be lowered even more with a longer underlap e.g. velvet stitch instead of satin. When both samples are compared to the weft knitted ½ pattern (Table 1), it is clear that the pillar-inlay has comparable properties for air permeability and thickness, but a significantly improved elongation at 49 N. Table 1: Comparison between data weft and warp knitted HRSM Figure 2: Pillar stitch combined with (a) knitted satin loop and (b) inlay threads over four needles [3] a The results show a significant difference in fabric properties between both structures. The air permeability is significantly higher for the inlay pattern because the inlay threads are not knit into loops like the satin stitch. The thickness is significantly lower, with 1,18 mm versus 1,95 mm for the knitted loop pattern. Furthermore, the inlay pattern has improved elongations at 49 N for both wale and course directions, with 20 w% and 45 c% compared to 27 w% and 88 c% for the satin-pillar sample. This results in an increased ratio w/c, as shown in Figure 3. b Pattern Weft Knit 1/2 Satin-pillar Pillar-(4)inlay AP (l/(dm 2.min)) Thickness (mm) 1,24 1,95 1,18 WaleBF (N) CourseBF (N) WaleE49 (%) CourseE49 (%) Ratio w/c 1,78 0,30 0,44 Legend: BF = tensile force at break / E49 = elongation at 49 N V. HIGH TEMPERATURE CHARACTERISATION Additional investigation was done on the behaviour of SS fibre fabrics at high temperature. The first test consisted of testing the effect of high temperature oxidation on the tensile properties at fibre, yarn and fabric level. At 780 C, SS will be oxidised at the surface. The loss of weight results in a decrease of strength and elongation at all three levels. The second test, named sagging test, is meant to test the cyclic loading on the HRSM fabrics in a high temperature environment. The result is expressed in an amount of mm that the fabric sags after cyclic loading. The results showed that warp knitted fabrics had slightly higher values: 25 mm versus 21 mm for weft knitted fabrics. VI. CONCLUSION Warp knitting is a promising alternative to weft knitting for HRSM fabrics. Two types of warp knitted structures have been found to have interesting properties. Specifically the pillar-inlay type has significant improvements over the weftknitted sample in terms of elongation at break. Figure 3: Difference in ratio w/c between satin-pillar and pillar- (4)inlay with 95% confidence levels. D. Effect of single versus double needle bed Warp knitted fabrics can be knitted on either a single needle bed or double needle bed, both methods result in a different structure with different properties. Different types of double needle bed structures with two guide bars, such as double REFERENCES [1] ROZANT, O., P.E. BOURBAN, and J.-A.E. MANSON, Drapability of dry textile fabrics for stampable thermoplastic preforms. Composites: Part A, : p [2] RAZ, S., Warp knitting production.1987, Heidelberg: Melliand Textilberichte. ISBN [3] KYOSEV, Y. and W. RENKENS. TexMind 2011; Available from:

7 Contents ACKNOWLEDGEMENT... i COPYRIGHT NOTICE... ii SUMMARY... iii EXTENDED ABSTRACT... iv CONTENTS... vi LIST OF FIGURES... viii LIST OF TABLES... x UTILISED ABBREVIATIONS... xi CHAPTER 1 INTRODUCTION... 1 CHAPTER 2 LITERATURE REVIEW AUTOMOTIVE GLASS PRODUCTION TECHNOLOGY Tempered glass Laminated glass Production technology MOULD COVERING FABRICS Warp knitting versus weft knitting Warp knitting machines Warp knitted structures MODELLING OF WARP KNITTED STRUCTURES Overview of existing models TexMind model PROPERTIES OF WARP-KNITTED FABRICS CHAPTER 3 METHODOLOGY INTRODUCTION TEXTILE PARAMETERS Yarn parameters Structural fabric parameters MATERIALS DESIGN OF EXPERIMENT CHARACTERISATION OF THE TEXTILE STRUCTURE At room temperature At process temperature CHAPTER 4 TEST RESULTS EFFECT OF STRUCTURAL TEXTILE PARAMETERS ON FABRIC PROPERTIES Effect of gauge and underlap movement GB Effect of gauge and underlap movement GB vi

8 4.1.3 Effect of inlay threads and take-off speed Effect of the amount of needle beds HIGH TEMPERATURE CHARACTERISATION Effect on the fibre properties Effect on the yarn properties Effect on the fabric deformability COMPARISON OF WEFT AND WARP KNITTED SAMPLES SUMMARY CHAPTER 5 CONCLUSION APPENDIX A: WARP KNITTED STRUCTURES APPENDIX B: FABRIC TEST DATA LITERATURE LIST vii

9 List of Figures FIGURE 2.1: HRSM PRODUCTS USED IN AUTOMOTIVE GLASS PRODUCTION [2]... 2 FIGURE 2.2: STRESS DISPERSION IN GLASS THICKNESS [3]... 3 FIGURE 2.3: GLASS TEMPERING PROCESS [3]... 4 FIGURE 2.4: EFFECT OF UNEVEN HEATING ON GLASS SHAPE [3]... 5 FIGURE 2.5: TEMPERATURE DIFFERENTIAL DURING QUENCHING [3]... 5 FIGURE 2.6: LAMINATED GLASS [2]... 6 FIGURE 2.7: GRAVITY SAGGING FOR LAMINATED GLASS [2]... 6 FIGURE 2.8: OVERVIEW OF BENDED GLASS TECHNOLOGIES... 7 FIGURE 2.9: IN-FURNACE GRAVITY SAG BENDING [2]... 8 FIGURE 2.10: IN-FURNACE PRESS BENDING [2]... 8 FIGURE 2.11: OUT-OF-FURNACE PRESS BENDING [2]... 9 FIGURE 2.12: COMPARISON BETWEEN PLAIN WOVEN (A), WARP-KNITTED (B) AND WEFT-KNITTED (C) STRUCTURE [9] FIGURE 2.13: DIFFERENCE BETWEEN WEFT (LEFT) AND WARP (RIGHT) KNITTING [10] FIGURE 2.14: TECHNICAL FACE OF PLAIN KNITTED STRUCTURE [10] FIGURE 2.15: FLOAT (LEFT) AND TUCK (RIGHT) STITCHES [10] FIGURE 2.16: DIFFERENCE BETWEEN TRICOT (LEFT) AND RASCHEL (RIGHT) WARP KNITTING FIGURE 2.17: GUIDE BAR LAPPING MOVEMENT FIGURE 2.18: BASIC OVERLAP/UNDERLAP VARIATIONS [11] FIGURE 2.19: DIFFERENCE BETWEEN OPEN (A) AND CLOSED (B) LOOP [11] FIGURE 2.20: PROPERTIES OF OPEN LOOPS VERSUS CLOSED LOOPS [12] FIGURE 2.21: LOOP FORMATION ON SINGLE NEEDLE BED RASCHEL MACHINE [10] FIGURE 2.22: LOOP FORMATION ON DOUBLE NEEDLE BED RASCHEL MACHINE [10] FIGURE 2.23: PLAITING OF THREADS [10] FIGURE 2.24: PLAITING PRINCIPLE DURING FRONT GB OVERLAP [10] FIGURE 2.25: TECHNICAL BACK OF SINGLE GUIDE BAR WARP KNITTED FABRIC [10] FIGURE 2.26: TECHNICAL FACE OF BALANCED DOUBLE TRICOT STRUCTURE FIGURE 2.27: COMMON PATTERNS WITH TWO GUIDE BARS [11] FIGURE 2.28: PRINCIPLE OF INLAY [11] FIGURE 2.29: DOUBLE NEEDLE OPEN PILLAR STITCH [11] FIGURE 2.30: PRODUCTION OF DOUBLE FACED FABRIC [11] FIGURE 2.31: THE LOOP MODEL BY G.L. ALLISON [20] FIGURE 2.32: THE MACHINE STATE LOOP MODEL [11] FIGURE 2.33: MODELLING HIERARCHY OF KNITTED STRUCTURES [31] FIGURE 2.34: INTRA AND INTER-LOOP INTERACTIONS [33] FIGURE 2.35: 2D LOOP TOPOLOGY WITH (A) MAIN DIMENSIONS AND (B) ANCHOR POINTS [31] FIGURE 2.36: KEY POINTS IN 3D LOOP TOPOLOGY FIGURE 2.37: GENERATED DOUBLE NEEDLE BAR STRUCTURE [32] FIGURE 2.38: TENSILE DIAGRAM OF A KNITTED STRUCTURE IN WALE DIRECTION [36] FIGURE 2.39: EFFECT OF FRICTION ON THE TENSILE PROPERTIES IN WALE DIRECTION [36] FIGURE 2.40: JAMMING MECHANISMS DURING BENDING OF A TWO-BAR WARP KNITTED FABRIC [40] FIGURE 2.41: INFLUENCE OF UNDERLAP LENGTH ON THE BREAKING STRESS IN COURSE DIRECTION [44] FIGURE 3.1: PROJECT FLOW FIGURE 3.2: EXAMPLE OF PATTERN CARD FIGURE 3.3: PREPARATION OF FABRIC SAMPLE FOR OXIDATION TEST viii

10 FIGURE 3.4: SAGGING SAMPLE PREPARATION FIGURE 3.5: SCHEMATIC PROCEDURE OF THE SAGGING TEST FIGURE 3.6: TEST SET-UP IN OVEN WITH SAMPLE FIGURE 3.7: TYPICAL PROCESS DIAGRAM OF SAGGING TEST FIGURE 4.1: EFFECT OF GB 1 AND GAUGE ON THE BREAKING STRENGTH FOR WALE (TOP) AND COURSE (BOTTOM) DIRECTION IN A MEANS PLOT WITH 95 % CONFIDENCE LEVELS (LEFT) AND A PARETO CHART OF STANDARDISED EFFECTS (RIGHT) FIGURE 4.2: EFFECT OF GB 1 AND GAUGE ON THE ELONGATION AT 49 N FOR WALE (TOP) AND COURSE (BOTTOM) DIRECTION IN A MEANS PLOT WITH 95 % CONFIDENCE LEVELS (LEFT) AND A PARETO CHART OF STANDARDISED EFFECTS (RIGHT) FIGURE 4.3: EFFECT OF GB 1 AND GAUGE ON THE AP IN A MEANS PLOT WITH 95 % CONFIDENCE LEVELS (LEFT) AND A PARETO CHART OF STANDARDISED EFFECTS (RIGHT) FIGURE 4.4: EFFECT OF GB 1 AND GAUGE ON THE THICKNESS IN A MEANS PLOT WITH 95 % CONFIDENCE LEVELS (LEFT) AND A PARETO CHART OF STANDARDISED EFFECTS (RIGHT) FIGURE 4.5: EFFECT OF GB 1 ON TENSILE FORCE AT BREAK AND ELONGATION AT 49 N IN A MEANS PLOT WITH 95 % CONFIDENCE LEVELS FIGURE 4.6: EFFECT OF GB 2 AND GAUGE ON THE BREAKING STRENGTH FOR WALE (TOP) AND COURSE (BOTTOM) DIRECTION IN A MEANS PLOT WITH 95 % CONFIDENCE LEVELS (LEFT) AND A PARETO CHART OF STANDARDISED EFFECTS (RIGHT) FIGURE 4.7: EFFECT OF GB 2 AND GAUGE ON THE ELONGATION AT 49 N IN WALE DIRECTION IN A MEANS PLOT WITH 95 % CONFIDENCE LEVELS (LEFT) AND A PARETO CHART OF STANDARDISED EFFECTS (RIGHT) FIGURE 4.8: EFFECT OF GB 2 ON THE ELONGATION AT 49 N IN COURSE DIRECTION (LEFT) AND ON THE RATIO W/C (RIGHT) IN A MEANS PLOT WITH 95 % CONFIDENCE LEVELS FIGURE 4.9: EFFECT OF GB 2 ON THE AIR PERMEABILITY IN A MEANS PLOT WITH 95 % CONFIDENCE LEVELS (LEFT) AND A PARETO CHART OF STANDARDISED EFFECTS (RIGHT) FIGURE 4.10: EFFECT OF GB 2 ON THE FABRIC THICKNESS IN A MEANS PLOT WITH 95 % CONFIDENCE LEVELS (LEFT) AND A PARETO CHART OF STANDARDISED EFFECTS (RIGHT) FIGURE 4.11: PILLAR STITCH COMBINED WITH KNITTED LOOP (LEFT) AND INLAY (RIGHT) OVER FOUR NEEDLES FIGURE 4.12: COMPARISON OF COURSE ELONGATION AT BREAK (LEFT) AND RATIO W/C (RIGHT) BETWEEN SATIN-PILLAR AND PILLAR-INLAY STRUCTURE IN A MEANS PLOT WITH 95 % CONFIDENCE LEVELS FIGURE 4.13: COMPARISON OF COURSE ELONGATION AT 49 N (LEFT) AND RATIO W/C (RIGHT) BETWEEN TWO PILLAR-INLAY SAMPLES WITH DIFFERENT COURSE DENSITIES IN A MEANS PLOT WITH 95 % CONFIDENCE LEVELS FIGURE 4.14: TENSILE CURVES OF OXIDISED (- -) AND NON-OXIDISED ( ) 12 µm STAINLESS STEEL FIBRES FIGURE 4.15: EFFECT OF HTO ON FIBRE TENSILE FORCE (LEFT) AND ELONGATION (RIGHT) AT BREAK IN A MEANS PLOT WITH 95 % CONFIDENCE LEVELS FIGURE 4.16: SEM IMAGES OF OXIDISED 12 µm FIBRES AT 1000X (LEFT) AND 5000X (RIGHT) FIGURE 4.17: TENSILE PROPERTIES OF OXIDISED (- -) AND NON-OXIDISED ( ) NM 11/2 YARN FIGURE 4.18: TENSILE PROPERTIES OF OXIDISED (- -) AND NON-OXIDISED ( ) 15/2 NM YARN FIGURE 4.19: EFFECT OF HTO ON THE TENSILE FORCE AND ELONGATION AT BREAK OF NM 11/2 AND 15/2 YARNS IN A MEANS PLOT WITH 95 % CONFIDENCE LEVELS (LEFT) AND PARETO CHART OF STANDARDISED EFFECTS (RIGHT) FIGURE 4.20: RATIO OF TENSILE FORCE AT BREAK PRE- AND POST-OXIDATION FOR FIBRES, YARNS AND FABRICS FIGURE 4.21: RATIO ELONGATION AT BREAK PRE- AND POST-OXIDATION FOR FIBRES, YARNS AND FABRICS FIGURE 4.22: CORRELATION BETWEEN FABRIC WEIGHT AND SAGGING WITH R² = 0, FIGURE 4.23: CORRELATION BETWEEN ELONGATION AT BREAK AND SAGGING FOR SAMPLES 4A, 5A_1 AND 5A_ ix

11 List of Tables TABLE 3.1: YARN PROPERTIES TABLE 3.2: THEORETICAL COMPOSITION RANGE OF EN ALLOY IN % MASS [45] TABLE 3.3: SAMPLES IN THE MAIN DOE TABLE 3.4: SAMPLES FOR THE INVESTIGATION OF OTHER STRUCTURAL PARAMETERS TABLE 3.5: OVERVIEW OF UTILISED ISO NORMS FOR TESTS AT ROOM TEMPERATURE TABLE 3.6: TENSILE TEST SPECIFICATIONS FOR FIBRES AND YARNS TABLE 3.7: SAGGING TEST SETTINGS TABLE 4.1: SUB-DOE A TABLE 4.2: SUMMARY OF AVERAGE TEST RESULTS OF SUB-DOE A TABLE 4.3: SUB-DOE B TABLE 4.4: SUMMARY OF AVERAGE TEST RESULTS SUB-DOE B TABLE 4.5: AVERAGE TEST DATA OF SATIN-PILLAR AND PILLAR-INLAY STRUCTURE TABLE 4.6: PATTERN DETAILS OF DOUBLE NEEDLE BED SAMPLES TABLE 4.7: AVERAGE TEST DATA OF DOUBLE NEEDLE BED SAMPLES TABLE 4.8: FIBRE TEST DATA TABLE 4.9: YARN TEST DATA TABLE 4.10: SAGGING TESTING VALUES TABLE 4.11: SUMMARY OF TEST RESULTS FOR COMPARISON WITH WEFT KNITTED REFERENCE SAMPLE TABLE 4.12: SAGGING COMPARISON BETWEEN WEFT AND WARP KNITTING TABLE 4.13: IMPORTANT DRIVERS FOR WARP KNITTED HRSM x

12 Utilised abbreviations AP BF BE DOE E49 GB GF HRSM HTO Nm SS Air permeability Tensile force at break Elongation at break Design of experiment Elongation at 49 N Guide bar Glass fibre Heat resistant separation material High temperature oxidation Number metric, unit for yarn count, expressed in m/g Stainless steel xi

13 Chapter 1 Introduction Fabrics made out of stainless steel (SS) fibres are used in the production process of press bended automotive glass. The process consists of pressing flat glass around a mould at high temperature (up to 700 C). Direct contact between the glass and the mould would lead to optical distortions and defects such as scratches. To avoid this, knitted SS fibre fabrics are used as a heat resistant separation material (HRSM) to cover the mould and protect the glass. Due to the direct contact between the fabric and the glass, the structural fabric characteristics strongly influence the quality of the formed glass. Mould covering fabrics made out of SS fibres have a high maximum working temperature and wear resistance. The increasing complexity of the glass shapes (with a higher radius of curvature) and the recently developed quantitative method by ISRA Vision [1] to measure optical distortion have led to a demand for improved HRSM fabrics. Furthermore, a new method to produce windshields, which have higher optical requirements than side and back windows, by press bending instead of gravity sagging, puts even more pressure on the development of a new generation of HRSM fabrics. The currently used technology, circular weft knitting, does not provide enough patterning possibilities to meet the demand for improved fabrics. They also have certain disadvantages such as non-isotropic deformation characteristics and sagging of the fabric at high temperature (700 C). An alternative technology is warp knitting, which has more structural parameters that can be adjusted and therefore more patterning possibilities. It has an inherently different structure and a higher flexibility regarding the orientation, length and shape of the loop. However, this structural difference results in a completely different deformability, air permeability and thickness, all important for the application as HRSM. The aim of this thesis is to investigate the effect of structural warp knitting parameters on the fabric properties, which is the first step in developing a new generation of HRSM fabrics. The goal is to obtain a fabric with improved deformability, sagging and draping behaviour, in the end to improve the overall optical quality of the glass. The focus of this work is on the properties of warp knitted fabrics at room temperature, followed by an indication of the behaviour at process temperature. In the next chapter a literature review is given, from the production of automotive glass to the characteristics of warp knitted fabrics. The third chapter will describe the material specifications and test methods used. In the next chapter, the test results of the different samples will be given and compared. In the last chapter a conclusion is made, indicating which warp knitted structures are promising to use as a mould covering fabric. 1

14 Chapter 2 Literature review This chapter will give a literature review of all topics regarding the production of automotive glass and warp knitted SS fibre fabrics. In the first part an overview of the glass production technologies will be given and in which parts of the process the fabrics are being used. In the second section the technology of warp knitting and the differences with weft knitting will be discussed. Furthermore, several warp knitted structures will be explained in detail. In the third section an overview of the existing models and visualization methods of warp knitted structures will be presented. In the future these models could allow an accurate prediction of the fabric properties and simulate its deformation behaviour. The fourth section gives a review of all the research regarding the properties of warp knitted fabrics. 2.1 Automotive glass production technology Figure 2.1 gives an overview of the heat resistant separation material (HRSM) products used in automotive glass production [2]. Three types of applications for HRSM exist and only the mould covering fabrics are discussed in this paragraph. The other two HRSM applications, for example the roller covering sleeves in the furnace, are not included the scope of this study. Figure 2.1: HRSM products used in automotive glass production [2] 2

15 Automotive glass can be subdivided in the front (windshield), back, side, quarter and sunroof windows. The industrial term given to these windows is lite, for example sidelites. Quarterlites are the smaller shaped windows in the side flanks of a car. Two types of safety glass are used in automobile windows: tempered and laminated glass. Front windows or windshields are made from laminated glass and the other windows are made of tempered glass to increase the strength of the window Tempered glass Tempering is a process which makes glass stronger by creating a protective compressive stress on the glass surface with a thermal toughening process [3]. The definition of fully tempered glass by the American Society for Testing and Materials (ASTM-C 1048) is thermally treated glass having a final surface compression of 10,000 psi (69 MPa) or more, or an edge compression of 9,700 psi (67 MPa) or more [4]. For tempered glass to qualify as safety glass it must also meet the requirements by ANSI Z97.1, which limits the size of the broken glass particles in a crash [5]. The stress dispersion throughout the thickness of tempered glass is shown in Figure 2.2. The tempering process creates a differential stress between the outside and inner glass surface. The outer layers are under a compressive stress, which gives the tempered glass its high strength and protects the weaker inner surface. In order for tempered glass to be broken, the magnitude of the impact (bending) force must overcome the built-in compressive surface stresses. Figure 2.2: Stress dispersion in glass thickness [3] When the surface of glass is penetrated by a deep scratch or impact, the stress gradients will make the crack propagate in a curved shape. This leads to the formation of small glass pieces called cullet. Cullet is safer than the sharp edged pieces formed by broken annealed glass. The higher the compressive stress level, the more cracks that will form, the faster they will propagate and the smaller the cullet particles will be. 3

16 The tempering process is possible due to the viscoelastic properties of glass at high temperature and the thermal expansion characteristics. Glass will become less viscous and expand with an increasing temperature. For glass tempering two conditions are required: a uniform heating of the glass followed by a rapid uniform cooling of the whole surface. The process of tempering glass is subdivided into three phases: heating, quenching and cooling. The temperature change during the three phases is shown in Figure 2.3. T E M P E R A T U R E D E G R E E S C C 540 C STRESS RELIEF TEMP SURFACE MIDPLANE 690 C FURNACE TEMPERATURE 580 C 410 C MIDPLANE SURFACE TIME HEATING TIME QUENCH TIME Figure 2.3: Glass tempering process [3] COOLING TIME The first phase is a uniform heating of the glass in the furnace. The heating time depends on the glass thickness; the general rule is 40 seconds per mm glass. As the glass is heated up, it expands. During the heating in the furnace the surface of the glass gets hot faster than the inner section. When the glass reaches a temperature above 540 C (annealing point), all residual stresses in the glass from previous processing are removed. The glass is heated up further to a temperature between C. Below 620 C, glass is too cold to develop the correct amount of compressive stress required for tempered glass. Above 640 C, problems will arise with optical distortion and shape stability. At the end of the heating phase, the temperature should be the same at the top and bottom surface of the glass. The uniform heating is very important, since a temperature differential between both sides leads to a non-uniform bending, as shown in Figure 2.4. This nonuniform bending is caused by a larger contraction of the hotter surface compared to the colder surface during cooling. 4

17 610 C Exit temperature of hot glass = 20 C Differential 615 C 630 C Glass shape after quenching and cooling Figure 2.4: Effect of uneven heating on glass shape [3] The second phase is quenching, this is a rapid (± 5 s) and uniform cooling of the glass surface. Due to quenching, the outside surface of the glass rapidly loses its heat and starts to shrink. The best quenching is obtained with quench settings that lead to a temperature difference of 170 C between the centre and the surface of the glass, as shown in Figure 2.5. The goal is to hold this temperature difference until the centre is colder than 540 C. By cooling the glass further, the inside zone starts to shrink thermally, but the already solid outer surfaces cannot shrink any further. This stress differential puts the surface into compression and the centre in tension, which is the goal of the tempering process. The surface compression strengthens the glass and protects the more vulnerable central tension layer. Thinner glass needs a more rapid quenching to obtain the necessary temperature differential. The third phase consists of cooling the glass to reach a comfortable temperature for further processing. Q u e n c h A i r 410 o C 170 o C 580 o 170 o C Q u e n c h A i r Figure 2.5: Temperature differential during quenching [3] 5

18 2.1.2 Laminated glass Laminated glass consists of two layers of glass bonded together under heat and pressure with a tough plastic polyvinylbutyral (PVB) layer. This glass/pvb sandwich behaves as a single unit and looks like normal glass. Laminated glass breaks rather easily on impact, but the plastic interlayer prevents the fragments from falling out of the window and causing injuries. Furthermore the PVB layer blocks UV-radiation. In Figure 2.6 the principle of laminated glass is shown. Figure 2.6: Laminated glass [2] Laminated automobile glass is used mainly for windshields, but also for sidelites with a darker or customized colour. It is traditionally produced by a gravity sagging technique, where the glass is transported on a carousel skeleton through a heating zone (Figure 2.7). In the heating zone, two glass layers are simultaneously bent by their own weight in order to guarantee exactly the same shape. The shape of the windshield is determined by the skeleton ring. After gravity sagging the two glass pieces are separated and an intermediate PVB layer is inserted. Since the glass is shaped purely by its own weight and not by a pressing mould, HRSM fabrics are not used in this carousel technology. However, due to the increasing shape complexity of automotive glass, there is a trend to use press-bending for the complex shapes, in which HRSM fabrics are used. Specifically for windshields, which have higher optical requirements than side or backlites, there is a need for improved HRSM fabrics to fulfil these requirements. Figure 2.7: Gravity sagging for laminated glass [2] 6

19 2.1.3 Production technology Depending on the type of safety glass and application area there are different technologies to produce automobile glass, as shown in Figure 2.8. The two main technologies differ in the way the glass is bent: for in-furnace technology the glass is bent inside the hot furnace, in out-of-furnace technology this happens directly after the furnace. The two technologies can be further subdivided into gravity sagging and press bending techniques. Figure 2.8: Overview of bended glass technologies In-furnace The in-furnace bending technology was mainly developed by the companies Glasstech and Sekurit Saint-Gobain. Two methods for shaping the glass in-furnace exist: press bending and gravity sag bending. The complexity of the glass is the main parameter for determining which of these methods is used. Windows with a low complexity will be made by gravity sagging, while for more complex shapes a press bending process is used. The first step of the process is heating the flat glass part, which is heated from room temperature until approximately 620 C in a furnace with temperature of 690 C. The furnace is normally heated by a combination of hot air convection and infrared radiation. In the second step either a gravity sagging or press bending technique is used to shape the window at the end of the furnace. In the gravity sagging process the heated glass is bent by its own weight, the shape depends on the shape of the shuttle ring. In Figure 2.9 the gravity sag bending system is shown. The pick-up is done by a 7

20 mould under vacuum that is covered with HRSM material. The mould can be flat or slightly bent to pre-bend the shape before sagging. The mould drops the hot window on the shuttle ring which results in the sagging of the window. Finally, the window is rapidly quenched in order to get the required temperature differential. Figure 2.9: In-furnace gravity sag bending [2] In the press bending process, shown in Figure 2.10, the window is bent by a hot mould with extra vacuum that presses the glass in the shuttle ring. Again, the mould is covered with the knitted HRSM material. There is only a small difference between the two technologies: the action of the mould in the pick-up step. In the gravity sagging process the mould only has a vacuum suction function, while in the press bending process the mould has both a vacuum and pressing function. Figure 2.10: In-furnace press bending [2] In some cases a separate bend and quench ring are used in the press bending process. This enables the production of more complex shapes, for example the highly curved backlites in modern cars. 8

21 When only one ring is used for both bending and quenching, it is covered with a quenching fabric strip [6] [7], as previously shown in Figure 2.1. The function of this warp-knitted HRSM fabric strip is completely different compared to the mould covering fabrics. The fabric not only has to prevent the glass from breaking when the window falls from the mould on the ring, but it also has to have a very open structure to permit air flow during quenching. However, this limits the pressing force that can be used, since a high pressing force would leave an imprint of the knit structure on the edges of the glass window. When a separate bending and quench ring is used, the bending ring is not covered with HRSM fabric and is also narrower, which allows a higher pressing force and thus more complex shapes. Additionally, highly complex shapes require special bend rings with side wings to bend the sides of the window Out-of-furnace Out-of-furnace bending is the process of bending glass outside of the furnace. Due to the fact that the window is shaped outside of the hot furnace, the temperature loss is the reason that only tempered windows with a lower complexity, such as small side and quarter windows, can be made using out-of-furnace bending. The advantage of this technology is that it can be used to produce both tempered and laminated glass, depending on the quench settings. In case of tempered glass, quenching will cool the window rapidly, for laminated glass it will be slowly annealed. Traditionally, laminated windshields are produced by a carousel gravity sagging method, which does not use knitted HRSM since there is no mould. However, the increasing complexity of windshields requires the use of press bending techniques. Windshields also have higher optical requirements, which demands for improved knitted HRSM fabrics. Again different variations of the press bending technology exist. In Figure 2.11 the technology with the segmented bending ring is shown. The window is heated in the hot furnace. Upon its exit, it is pressed by the mould on the segmented bend ring that has moved up from under the rollers. After pressing, the bending ring moves down and the rollers transport the window to the quenching zone. Press bending of laminated windows presents new challenges, for instance to get two windows with exactly the same shape for lamination since the two windows are bent separately. In the traditional skeleton method this was no issue since the two windows were bent simultaneously. Figure 2.11: Out-of-furnace press bending [2] 9

22 2.2 Mould covering fabrics Mould covering fabrics are all fabrics used to cover the moulds in the automotive glass production, as discussed in the previous paragraph. The fabrics currently used are made by a circular weft knitting technology of 100% SS or mixed GF/SS yarns. The requirements, based on experiences with manufacturers, for a mould covering fabric are: High temperature resistance (> 680 C). Fabrics need to be able to withstand the high temperature developed in the furnace. Although metal oxides are formed on the surface of the fibres, SS fibres are able to withstand these conditions. Sufficient air permeability (> 800 l/(dm 2.min)). The fabric needs to allow sufficient air flow to enable the vacuum between the mould and glass. A low permeability would result in an incomplete bent glass. Thickness (< 2 mm). A higher thickness will require the furnace to be set at a higher temperature for sufficient heat transfer, which increases the energy consumption. High softness. Breaking strength in all directions should be higher than 49 N. This is a standard set by the automobile glass manufacturers. It is assumed to be the average force subjected to the fabric by the stretching process over the mould, the vacuum pressure and the releasing force when a glass is pressed. Sufficient drape-ability, which is determined by the tensile elongation (%) of the fabric. Not only the absolute value of the elongation in wale and course direction is important, but also the ratio between both influences the drape-ability. Ideally the ratio wale over course elongation is approximately one. Woven fabrics cannot be applied as mould covering materials due to their very limited drapeability. They are characterised by a low tensile elongation of the fabric, due to the straight orientation of the threads. Also the air permeability is in general too low to permit sufficient air flow. Figure 2.12 shows a comparison of plain woven, warp-knitted and weft-knitted structures. Clear differences in thread orientation and permeability of the structure can be noted Warp knitting versus weft knitting Knitted fabrics are defined as fabrics in which at least one system of threads is formed into knitted loops, that are intermeshed into stitches [8]. The knitted stitch is formed when the knitting needle receives a new loop and knocks over the old loop from the previous knitting cycle. Knitted fabrics can be subdivided into two main groups: weft and warp knits, depending on the way the stitches are formed. In weft knitted fabrics the stitches are made from the weft thread across the width of the fabric. In warp knitted fabrics the stitches are made from each warp thread along the length of the fabric. Similar to the warp and weft terms for weaving, the terms wale and 10

23 course exist for knitting. The difference with woven structures can be seen in Figure The rows of knitted loops across the width of the fabric are called courses and the columns along the length of the fabric are called wales. Figure 2.12: Comparison between plain woven (a), warp-knitted (b) and weft-knitted (c) structure [9] The methods used to make these fabrics are substantially different and are shown in Figure The needles (A-B-C-D) move up and down to intermesh the formed loops into stitches. In a weft knitting machine, the yarn feeding and loop formation occur at successive needles across the needle bed in one knitting cycle. One thread at a time is fed to form a course of loops from this single thread. In a warp knitting machine there is a simultaneous yarn feeding and loop forming action, which occurs at every needle across the needle bar during the same knitting cycle. The warp yarns are fed simultaneously from a warp beam to the needles by the guides (E-F-G-H) to form one loop per needle per cycle. The wales are interconnected by the sideway movements of the guides. Figure 2.13: Difference between weft (left) and warp (right) knitting [10] 11

24 The possible stitch types that can be produced with weft knitting machines are more limited compared to warp knitting. In weft knitting there are four basic structures: plain, rib, interlock and purl. Rib and interlock are two types of double faced structures produced on double needle bed knitting machines. Purl is a special single face structure that shows the same left loops on both sides of the fabric. Figure 2.14: Technical face of plain knitted structure [10] A plain single needle bed structure is also called single jersey. Such weft knitted fabric types are currently used to cover the moulds in automotive glass bending. Single jersey fabrics are made on circular weft knitting machines, which are more productive and efficient than flat weft knitting machines. Besides the knitted loop stitch, shown in Figure 2.14, there are two more stitches commonly used in single jersey knits: the float stitch and tuck stitch. Both stitches, shown in Figure 2.15, are made with a held loop, which is an old loop that the needle retained in the previous knitting cycle. Tuck stitches are used to increase the stability of the fabrics in the course direction. Other stitch variations include elongated loops and transferred loops. However, these stitches are not so interesting to tailor the mechanical properties of the HRSM fabrics and are more used for aesthetical purposes. Figure 2.15: Float (left) and tuck (right) stitches [10] 12

25 2.2.2 Warp knitting machines Two types of warp knitting machines exist: tricot and raschel machines, both shown in Figure The machines differ only in the design and function of the sinkers during loop formation. In tricot machines the sinker has two functions: the first one is to prevent the knitted loops from coming upwards with the needles and the second one is to clear the formed loops from the needle head and knock-over the previous loop. The sinker in raschel machines is only used to keep the fabric down, the loop clearing and knock-over is done by the trick-plate and a high take-up tension. Latch needles can be used on both tricot and raschel machines, but bearded needles can only be used on tricot machines. Raschel machines are more flexible in what types of fabric that can be made. Open net structures for example, cannot be held easily by the sinkers on tricot machines. This is not a problem on raschel machines since the sinker does not have a fabric supporting function. Figure 2.16: Difference between tricot (left) and raschel (right) warp knitting For both types the movement of the guide bars for the loop formation is the same, as shown in Figure The loop formation is characterised by an overlap and underlap, which is realised by a shog and swing movement of the guide bar. Figure 2.17: Guide bar lapping movement 13

26 Every knitted loop is made by one of the five variations of overlap and underlap, which are shown in Figure Every black dot represents a needle. The letters O and U stand for overlap and underlap. A row of numbers represents the movement of the warp guide bar. Every number represents a position between two needles. In (a) the guide bar moves from position one to zero (overlap) and then from position zero to one (underlap) to form one closed knitted loop. In the chain link notation only the overlap movements are written out, the underlap is represented by a slash. (a) Closed lap: overlap followed by an underlap in the opposite direction (b) Open lap: overlap followed by an underlap in the same direction (c) Only overlaps (d) Laying-in: only underlaps and no overlaps (e) Miss-lapping: no overlap or underlap Figure 2.18: Basic overlap/underlap variations [11] The difference between an open and closed loop can be seen in Figure An open loop is formed when the overlap and next underlap are made in the same direction. When done in opposite direction, a closed loop is formed. 14

27 Figure 2.19: Difference between open (a) and closed (b) loop [11] Regarding the choice of open loops versus closed loops for the application as HRSM material for mould covering, the following preferred properties need to be looked at: High elastic strain recovery: when the fabric is pulled over the mould the elastic strain on the loops keeps the fabric close to the mould and could have an influence on sagging Low and equal extensibility in course and wale direction: this enables the fabric to be applied similarly each time, regardless of the operator. Reduced edge curling: edge curling decreases the handling ability of the fabric when applying it to the mould High lustre: a smooth, shiny fabric will improve the optical quality of the glass The properties of the knitted fabric are different for open and closed loops, as shown in Figure As can be seen in the figure, open loops have all the properties desired in the fabric, except for the elastic strain recovery. As the fabric extensibility is considered to be an important factor for the draping procedure of the fabric over the mould, open loops are preferred. However, open loop movements are more difficult to form, except for the open pillar stitch. The successful formation of open loops strongly depends on the tension of the warp yarns. If the applied tension is incorrect, the loop can slip off the needle during the next knitting cycle. Fabric property Open loops Closed loops Desired in fabric Elastic strain recovery Lower Higher High Extensibility Lower Higher Low Edge curling Lower Higher Low Lustre Higher Lower High Wear on knitting elements Low High Low Figure 2.20: Properties of open loops versus closed loops [12] 15

28 Single needle bed raschel machines The loop formation on a single needle bed raschel machine is shown in Figure 2.21, which can be divided into six steps: (a) The sinker holds down the fabric while the guide bars move into position for the next loop (underlap). (b) The needles rise, the loop in the needle head opens the latch and clears the needle (c) The two guide bars swing from the front to the back of the needle bar, every thread of every guide forms an overlap on the corresponding needle on the needle bar. (d) Both guide bars make the return swing from the back to the front to finish the overlap (e) The needle bar moves down, the previous loop closes the latch which holds the new loop in the needle head (f) The needle bar moves further down and the previous loop is knocked-over Figure 2.21: Loop formation on single needle bed raschel machine [10] 16

29 Double needle bed machines Double needle bed raschel machines have a second needle bed that is opposite to the one in single needle bed machines. The loop formation, shown in Figure 2.22, is similar, but the loops are formed alternating on the front and back needle bed. (a) The front needle bar rises to clear the previous course from the needle heads and latches (b) During the overlap the guide bar swings around the needles (c) The needle bar descends to knock-over the stitches and the guide bar does the underlap shog (d) The back needle bar starts its knitting cycle Figure 2.22: Loop formation on double needle bed raschel machine [10] Warp knitted structures Two main structures can be distinguished in warp knitting: single needle and double needle bed structures. Single needle bed structures always have two structurally different sides called the technical face and back. The technical back, the visible top side during knitting, is characterised by the underlaps. The technical front side is characterised by the knitted loops. Double needle bed structures have either two different or two identical sides and are called double face structures. Besides the amount of needle beds, the amount of guide bars is another important parameter that greatly determines the structure. This amount can vary from minimum one guide bar to six or more. More complex structures will require more guide bars. Every guide bar is programmed with a certain pattern and the combination of these patterns leads to the warp knitted structure. Two guide bars are commonly used and already provide many patterning possibilities. 17

30 When two guide bars with loop-forming function are used, there is a plaiting phenomenon. Only the threads of a certain guide bar will be visible on the top or bottom side of the knit, as shown in Figure Plaiting is inherent to the set-up of the machine. When the two guide bars make the overlap movement, shown in Figure 2.24, the threads will cross and the threads of the front guide bar will be on top of the technical back. When the guide bars shift and form the underlap, the threads of the front guide bar will be on top and visible on the technical back. For double needle bed structures the phenomenon is similar and allows the production of an identical double face structure with two different threads on each face. For HRSM fabrics this can be advantageous, since a fabric with on the glass side 100% glass fibre (GF) and on the mould side 100% SS fibres is possible. Research has shown that mixed GF/SS fabrics result in better optical quality of the window compared to 100% SS fibre fabrics [13]. Figure 2.23: Plaiting of threads [10] Figure 2.24: Plaiting principle during front GB overlap [10] 18

31 Single needle bed structures The most basic structures that can be made with a single needle bar machine are those with one guide bar. These structures are dimensionally unstable and split easily when damaged [11]. An example of a single guide bar fabric with an unbalanced loop structure is shown in Figure The non-linear configuration of the wale loops is also called loop inclination. Figure 2.25: Technical back of single guide bar warp knitted fabric [10] Double guide bar structures are more stable due to the opposite loop orientation of the two guide bars. When the yarn tension in both guide bars is balanced the loops will be erect, as can be seen in Figure Figure 2.26: Technical face of balanced double tricot structure In Figure 2.27 the common double guide bar patterns are shown. The left pattern represents the pattern of the front guide bar, the right pattern the back guide bar. When the patterns are switched, as in b-e / c-f / d-g, this will give a different structure with different appearance and handle due to the plaiting property of the machine. For example, the difference between locknit (b) and a reverse locknit (e) is that the locknit will give a softer touch and higher elasticity due to the free-floating underlaps. In the reverse locknit the longer back guide bar underlaps are locked under the shorter front guide bar underlaps, which restrict the movement of the structure. There 19

32 are similar effects between satin and sharkskin structures, with the satin having a smooth technical back and the sharkskin a rough one. Another important parameter is the length of the underlap. Longer reciprocating lapping movements are used to increase course wise stability, weight and density of the fabric. The longer the floating underlaps on the technical back are, the brighter and smoother the fabric will be [11] [14-16]. Figure 2.27: Common patterns with two guide bars [11] Another technique to increase the width wise stability is by laying-in course threads. These threads are not knitted into loops, but are laid in the loops of the pillar stitches. Figure 2.28 shows the principle of inlay with one thread. The front guide bar is fully threaded and is responsible for the formation of the pillar stitches, while the back guide bar only has one thread to lay into the pillar loops. Different laid-in patterns can be made depending on the threading of the guide bar (partial or full) and the inlay length. Yarns that are technically difficult to knit can be used and inserted as laid-in threads in the knitted structure, which is the main advantage of this technique. An example 20

33 of such materials are yarns with a bending stiffness that is too high to form loops but low enough to form curves when inserted in the structure. Figure 2.28: Principle of inlay [11] Two-needle overlap stitches, shown in Figure 2.29, are a special type of stitches used to add body and stability to a single guide bar fabric. Each yarn is wrapped around two needles during the overlap cycle and both these needles draw the loops simultaneously. However, the fact that two loops are drawn from the same relatively small amount of yarn causes a large amount of stress on the yarn and needles. Since the metal fibre yarn is difficult to knit and requires a large amount of oil to remove the friction, this additional stress can have a potential negative influence on the knit ability. Figure 2.29: Double needle open pillar stitch [11] Double needle bed structures Double needle bed machines can be used to make a large variety of structures. The possibilities of these structures in HRSM materials is that thicker structures can be made, which could have a 21

34 softer and more dampening effect on the windows. Some disadvantages for using these structures as HRSM fabrics is their lower air permeability and higher thickness, which restricts the heat transfer from the hot mould to the glass. Designing this type of structures is also more complex since an additional parameter, the amount of needle beds, must be taken into account. Pattern drawings for a double needle bed machine are made by a different procedure compared to single needle bed structures, as can be seen in Figure Two rows of dots now represent one cycle of the machine: the front needle bar cycle is always represented by the first row, while the second row represents the back needle bar cycle. Figure 2.30: Production of double faced fabric [11] Three basic types of double needle bed structures can be distinguished. The first type are the double faced fabrics, in which both guide bars overlap on both needle bars. Each lapping movement is doubled on both needle beds before an underlap is done. An example of this type is shown in Figure Double needle bed machines can also be used to produce two separate single faced fabrics, this is the second type. If the front guide bar only overlaps on the front needle bed and miss laps on the back bed, and the back guide bar only overlaps the back bed and miss laps on the front bed, the two fabric sides will be separate. A third type is when the two fabrics are connected by the underlaps to form one fabric. This occurs when the back bar only overlaps the front needle bed and the front bar only overlaps the back needle bed. Warp knitting makes it possible to combine two or more different sets of loops in the same structure. This provides warp knitting with more possibilities, compared to weft knitting, to tailor a fabric with the desired properties. However, it also becomes more complex to predict the outcome and behaviour of a certain structure. This is why the use of a modelling program can be very useful to speed up the design process and reduce the amount of time spent in practical trials. 22

35 2.3 Modelling of warp knitted structures It is of interest in the development of new knitted structures that one has the ability to model certain structures prior to experimentally producing them. Modelling structures allow a preselection and optimization in order to reduce the (often costly) development time. In this paragraph an overview of the existing models regarding warp knitted structures is given, together with a description of the model specifically used to visualize the warp knitted mould covering fabrics Overview of existing models In the past decades numerous studies have been done on the geometry of knitted structures, most of which have been done of weft knits. Not only was the industrial significance of warp knitted fabrics lower, the warp-knitted structure is also more complex due to its dependency on the threading and the lapping movement. Early attempts consisted of experimental studies on the dimensional properties of warp knitted structures [17, 18] [19]. The first attempt to relate stitch length to fabric geometry was the geometrical model of Allison, in which the warp knitted cell was split into four sections: the loop s head as a semi-circle, two straight lines for the loop s legs and a third straight line for the underlap. The model is shown in Figure Figure 2.31: The loop model by G.L. Allison [20] Although this model was more based on geometrical shapes than on the physical reality, it provided accurate results regarding the amount of yarn run-in. A few years later, Grosberg developed the first model based on the physical yarn configuration in the knitted cell [21]. The model was based on the assumption that the yarn is an elastic unit and that the shape of the loop is created by forces at the base of the loop. The shape created under these conditions is called an elastica, which exhibits a constant relationship between loop height and loop length. Experiments showed that the underlaps are straight in the fabric on the machine, but part of a circle in the relaxed fabric. The loops however, do not change noticeably after relaxation of the 23

36 fabric. This led to a simplification of the model by assuming that the loop and underlap are isolated from each other by friction at the cross-over point, so the unit loop can be considered as two separate parts. Later, Grosberg experimentally proved this assumption by experimentally comparing the properties of relaxed tricot fabrics with his model. The resulting formula is very complex, but the Grosberg model made it possible to calculate very accurate run-in values and to check fabric analyses [22]. Another approach was by Raz with the machine state loop model, in which he states that the loop shape of the fabric in the machine is more likely to be determined by the physical pull of the takeup mechanism, than by the bending forces suggested in Grosberg s model [11]. The geometry of this model is shown in Figure Figure 2.32: The machine state loop model [11] The geometrical models previously discussed have one thing in common: they do not take the real three-dimensional geometry of the knitted loop into account. Since these two-dimensional models do not provide a complete definition of the actual structure, they cannot be used in modern computer-modelling to predict the mechanics of the warp-knitted structure [23]. An overview of the different modelling theories and methods, applied on textiles, has been given by Sherburn [24]. Different methods have been reported to develop a real three-dimensional model, most of which use complex mathematics to define the topology of the loop structure [25],[26],[27, 28],[29, 30]. These models can only be applied to a limited amount of knitted structures and cannot be used at an industrial level. A step-by-step approach to model and predict the properties of a general warp knitted fabric, as shown in Figure 2.33, was given by Kyosev and Renkens [31]. There are two different approaches: on a structural level or on a material level. Defining an accurate and simple topology is the first step in making a general model for warp knitted structures on a structural level. From this topology, a geometrical model can be defined. The third and most complex step is to take the yarn 24

37 mechanics into account to predict the fabric properties. On a material level the different effects occurring in and between the yarns are considered. In Figure 2.34 the intra and inter-yarn interactions in a plain weft knitted loop are given, similar effects will occur in the warp knitted structure. The model of Kyosev and Renkens has been successfully developed into an industrial tool under the brand name of TexMind [32]. The program, although still under development, provides a useful tool in the daily visualization of all warp knitted structures. Figure 2.33: Modelling hierarchy of knitted structures [31] Figure 2.34: Intra and inter-loop interactions [33] 25

38 2.3.2 TexMind model The TexMind model distinguishes itself from the other models by its simple topology definition and the ability to automatically generate the mathematical model for warp knitted structures. The topology is similar to the one defined for weft-knitted fabrics by Moesen et al [34]. The control parameters of warp knitting machines, such as lapping movement, threading, take-up speed and machine gauge, are used as input data of the model. The modelling process is divided into three steps: pre-processing, solution and post-processing. The pre-processing step is basically checking the knitability of the input data, for example to check the diameter of the yarn with the space between the needles (gauge). In the solution step the basic structural elements are created (topology), from which the yarn path in every element is calculated (loop form calculation). The post-processing of the data comprises of visualizing the structure and exporting it to other programs, for example to finite element modelling programs. In this section the solution step of the model will be discussed to give a clear view on how the model is built up and is based on the information found in the following articles [31, 35, 36] Topology The first step in a structural model is defining the topology elements of which the knitted structure exist. The 2D topology can be defined using the contact points between the loops in the X-Y plane. The yarn thread of each loop is described by a curve through six contact points, as shown in Figure The position of the contact points are defined by the loop height B, loop width L, the distance between two wales A, the distance between the feet K and the height of the feet yb. Figure 2.35: 2D loop topology with (a) main dimensions and (b) anchor points [31] The coordinates of these points are defined as (for 0 < i < n): 26

39 The loop height B, which is the distance between two courses, is defined by the take-up speed of the warp knitting machine. The distance between the wales A can be defined as with E the machine gauge in needle per inch. The distance may not be bigger than the distance between the needles, but in reality the fabric relaxes which decreases the distance A. The parameters K and yb are determined directly by the yarn radius r: The points 1, 2, 2, 1 define the position of a loop head, all coordinates of other loops in the knitted structure can be calculated using a simple translation of these points in X and Y directions. In order to make the structure more visual, smoothing curves such as splines can be drawn through the key points. The 3D topological representation can be derived from the two-dimensional one by considering the z-axis, as shown in Figure Figure 2.36: Key points in 3D loop topology 27

40 Every point 1 i,j and 2 i,j is now associated with two points +z and z, with Δz > R (yarn radius): The key-points selected for the visual representation depend on the warp knitted structure. For instance the loops in a double tricot warp knit consist of two yarns, which means that four z- positions are required for every point (-z, -2z, z, 2z). The key points are located around the local X-Y plane for a single needle bed machine. For a double needle bed machine two such local planes will be required to visualize the structure Loop form calculation Once the key points of single loops are known, the next step is building the geometrical and mechanical model. The geometric modelling consists of adjusting the positions of the key points according to the yarn geometrical parameters and the calculation of the yarn axis form. Basically, all the distances between the key points need to be checked and adjusted according to crosssection of the used yarns. An example of an image generated by TexMind is shown in Figure It shows a double needle bar fabric with double tricot pattern. Figure 2.37: Generated double needle bar structure [32] The geometrical model can generate an accurate image of the warp knitted structure, but it represents an idealized image. This image does not take into account the internal and external forces in the geometry which occur for example during fabric relaxation. These forces are considered in the mechanical models, which can be calculated by continuum models (force and energy), or with a discrete model, which reduces the yarns to mass-spring systems. A different approach is the use of FEM software, which can be used to calculate the deformations (small or big) of the knitted structure. In the TexMind model, three different FEM tools are used and implemented to consider the mechanical influences. The first one is based on truss, which represents the loop as a frame of trusses to calculate uni- and biaxial deformations. At low 28

41 deformations, the knitted structure can be seen as a truss framework in order to model the deformation behaviour [37]. The second tool is based on beam elements, since they transmit bending moments, which can be used to calculate the compressibility. The third one is an explicit FEM to simulate the knitting process for the contact calculations between yarns (friction effects). These calculations are important for the mechanical modelling, but the main difficulty is to obtain a stable algorithm with an acceptable calculation time. The effect of fabric relaxation or tensile forces on the structure is described in mechanical models. Kyosev and Renkens have described a theoretical model for the tensile properties of warp knitted structures [36]. Different effects occur within the structure when stretched. Depending on the tensile direction (wale or course) the structural deformation will be different. In wale direction the tensile diagram is divided into two phases. In each phase a different parameter determines the shape of the curve. The first phase in the tensile diagram is determined by the bending stiffness of the yarn. In Figure 2.38 it is shown how the wale loops are bent and increase in length when strained. Stiffer yarns will require a higher force to be strained. The second phase in the diagram starts when the loop is completely stretched (structural elongation is maximum) and the yarns themselves are being stretched. This phase is mainly determined by the yarn tensile properties. Figure 2.38: Tensile diagram of a knitted structure in wale direction [36] For a more accurate and realistic prediction, a third parameter, can be taken into account. This parameter describes the effect of friction and sliding between the loops and underlaps. As discussed in the Grosberg model in the previous paragraph, the friction between the loops and the underlaps plays an important role in the deformation of the structure under tensile stress. The yarn sliding and friction effects concerning single guide bar structures under tensile stress was previously investigated by Stumpf et al. [38]. The effect of yarn slippage on the friction force between the yarns was determined experimentally. The results showed that from a certain force 29

42 level, the yarn slippage stops and further fabric deformation is based on elastic elongation of the yarns. Furthermore, it was determined that the friction is independent of the sliding velocity. It was also proposed to model the deformation of the structure during yarn slippage by a truss framework, since there is relatively little change in the loop structure in this phase. The effect of friction and sliding on the tensile diagram is shown in Figure The loops will start sliding when put under strain. The effect of friction and sliding becomes greater at higher elongation values. It is only when the loops are completely stretched that the material will be strained, which is in the second phase in the diagram. However, the effect of this parameter is difficult to define and is therefore sometimes neglected to simplify the model. Similar effects will occur during the deformation in course direction, but the tensile diagram will be completely different. Straightening of the loops through sliding will contribute greatly to the, in general, higher elongation in course direction. It can be expected that structures with a longer underlap will stabilize the structure in course direction, due to the relative smaller length of yarn in the loop compared to the underlap length. Therefor smaller loops and longer underlaps will decrease the tensile elongation in course direction. Smaller loops can be achieved by decreasing the machine take-off to increase the amount of courses per centimetre. Figure 2.39: Effect of friction on the tensile properties in wale direction [36] Mechanical models could be very useful to predict the properties of warp knitted fabrics. For HRSM fabrics, it would allow a prediction of the tensile and draping properties of the fabric when stretched over the mould. Up until now however, little research has been conducted on the deformability of warp knitted fabrics. The main reason is that the deformability properties are very structure and material specific, and cannot be generalized easily for all warp knitted structures, especially for SS yarns. 30

43 2.4 Properties of warp-knitted fabrics In this paragraph, a review of the past and more recent research on the properties of warp knitted fabrics is given. Although little research can be directly applied to the HRSM fabrics made out of SS, the research gives an indication of the effects occurring within the structure that can possibly influence the fabric properties. The review is focused on the two-guide bar structures, since single guide bar structures have a too open and unstable structure to be applied as HRSM fabric. In the past some research has been done on the influence of two-guide bar warp knitted structures on the different fabric properties. Yanagawa and Kawabata (1972) determined the biaxial tensile properties of two-bar warp-knit fabrics [39]. The bending properties of warp-knitted outerwear fabrics were experimentally investigated by Gibson et al [40] for a wide range of materials and structures. Different two-bar fabrics with a different underlap length were tested and the influence of the underlaps on the bending rigidity was described. Certain jamming mechanisms in the knitted structure were proposed as cause for the different bending rigidity in course or wale direction, but also for positive (technical face on outside of curve) or negative (technical face on inside of curve) curvature. In Figure 2.40 two different jamming mechanisms are described, in (a) the length jamming is shown, when the bending moment is applied parallel to the courses for a positive curvature, in (b) the width jamming is shown, when the bending moment is applied parallel to the wales for a negative curvature. Figure 2.40: Jamming mechanisms during bending of a two-bar warp knitted fabric [40] The results show that the bending rigidity has the strongest correlation with the fabric weight per unit area and too lesser extent with the fabric thickness. Since the weight per unit area is greatly influenced by the length of the underlap, it can be concluded that a longer underlap will provide a greater resistance to bending. This is due to the fact that for a relatively long underlap, more yarns are located between each course at any needle space, which provides a greater resistance to the bending of these underlaps. Similarly, the shear properties were also investigated [41] to determine important properties such as the drape and handle of warp-knitted outerwear fabrics. This led to the conclusion that fabrics 31

44 with longer underlaps have a higher resistance to shear deformation. Again, this can be explained by the fact that longer underlaps result in a higher amount of yarns between each course at any needle space. More recent studies on the bending rigidity and shear friction of warp knitted structures have been done by Jeddi et al [42, 43], which have given similar results regarding the influence of the underlap length as stated by Gibson et al. Moreover, it was also shown that the density of the structure influences the bending rigidity: tighter knits with higher density will have a higher bending rigidity. For the surface friction, Jeddi et al reported a decrease in surface friction for an increased fabric density. This can be easily explained by the fact that more densely knitted structures will have a tighter and smoother surface. In another recent study [44], the influence of the knitted structure on the tensile properties and fatigue behaviour was determined. Two structural parameters were found to have an important effect on the fabric elasticity: the space available for yarn movement and the length of the underlap. The space available for yarn movement is determined by the space between the overlaps and the front guide bar underlap. For example a locknit structure will have more space than a reverse locknit structure. Tests were carried out on tricot, locknit, reverse locknit, threeneedle satin, four-needle satin, three-needle sharkskin and four-needle sharkskin structures with different knit densities. The results show that for an increasing underlap length, the breaking strain will decrease and the breaking stress will increase, which is shown in Figure The breaking stress is given in cn/course, to eliminate the difference in fabric course density of the tested samples. Figure 2.41: Influence of underlap length on the breaking stress in course direction [44] 32

45 Chapter 3 Methodology 3.1 Introduction In this chapter the methodology of the thesis is discussed. The materials used, the investigated textile parameters and test methods are described in detail. In Figure 3.1 the project methodology is schematically shown in a flow chart. The aim of this thesis is to investigate the effect of structural warp knitting parameters on the properties of HRSM fabrics. The first step in this investigation is to determine the deformability characteristics at room temperature, and to determine which structural fabric parameters influence it. Based on the literature review, a design of experiment is made to investigate the effect of these parameters. The TexMind program is used to make 3D images of the structures, which helps visualization and selection of the most interesting structures for this study. In the next step, the samples are produced. One part of the samples is made in-house at Bekaert on a single needle bed crochet warp knitting machine. Another part of the samples were outsourced to a partner and were made on a single needle bed raschel warp knitting machine. After the samples are knitted, the mechanical properties of both the gauge eight and twelve samples are characterised. In the next chapter the results are statistically analysed to determine the effect and interactions of the structural parameters, both material and machine wise. Additionally, high temperature tests are performed to give an indication of the fabric properties at the process temperature. Parallel to the evaluation of the warp knit samples, a currently used weft knitted HRSM is tested. The results of the weft and warp knitted HRSM are compared to determine if warp knitted fabrics are a viable alternative. From the parametrical analysis of warp knitted fabrics and the comparison between weft and warp knitted fabrics, a conclusion will be made regarding the possibilities of warp knitting for HRSM fabrics. 33

46 Figure 3.1: Project flow 3.2 Textile parameters The yarn and structural fabric parameters need to be taken into account when designing warp knitted textile structures based on metal fibres Yarn parameters The following yarn parameters influence the properties and handle of the warp knitted fabric structure: Type: Spun yarn or continuous filament yarn Material: type of alloy (AISI or EN) Fibre diameter 34

47 Fibre length Twist [twist/meter] Tensile force at break [N] Elongation at break [%] Structural fabric parameters In the warp knitting process SS yarns are knitted into a HRSM fabric, as discussed in Chapter 2. The type of machine and the parameters set determine the structure, density and handle of the fabric. The following structural parameters determine the knitting pattern: Gauge [needles per inch]: fixed value, depends on the machine Amount of needle beds: single or double Amount of guide bars Amount of warp threads: has an influence on the fabric width Guide bar threading: full or partial threading influences the pattern Underlap length Open or closed loops Presence of inlay threads 3.3 Materials In this paragraph the specifications of the materials used in the knitting tests are given. Two types of spun yarns were available for the knitting tests. For the gauge eight tests, a SS Nm 11/2 yarn is used. For the gauge twelve tests, a SS Nm 15/2 yarn is used, the Nm 11/2 type is too coarse for the knitting machine. In Table 3.1 the details of the used yarns are given. As can be seen, the yarns have practically no tensile elongation which makes the warp knitting more difficult. Both yarns have been oiled to improve the knit ability. In Table 3.2 the theoretical composition range of the used EN alloy is given, this alloy is also known as AISI 316L according to the AISI standards. Table 3.1: Yarn properties Yarn Count [Nm] Alloy Twist [tpm] Tensile strength [N] Elongation at break [%] 11/2 EN ,37 1,15 15/2 EN ,98 1,10 Table 3.2: Theoretical composition range of EN alloy in % mass [45] C Si Mn P max. S N Cr Cu Mo Nb Ni 0,030 1,00 2,00 0,045 0,015 0,11 16,5-18,5-2,00-2,50-10,0-13,0 35

48 3.4 Design of experiment The goal of this thesis is to explore the use of warp knitted SS fibre fabrics as an HRSM for mould covering, by investigating the effect of structural warp-knitting parameters on the fabric properties. The first step in this investigation is to determine the deformability characteristics at room temperature, and to determine which structural parameters influence it. In the literature review it is found that single guide bar (GB) warp knitted structures are dimensionally too unstable and split easily when damaged [11]. Warp knitted structures knitted with two guide bars are more stable and already provide numerous patterning possibilities. Different patterns can be made by changing the sideways movement of both guide bars. The movement is defined by the number of needles lapped during the sideways shog. However, these two guide bar fabrics are generally more stable in warp direction, and more stretchable in course direction. The underlap length, which is determined by the guide bar movements, is claimed to have a significant influence on the fabric stability and deformability in the course direction. The main design of experiment (DOE) is made to determine the effect of gauge and the movement variation of GB 1 and/or 2 on the deformability of warp knitted stainless steel fibre fabrics. The set of samples consists of six different structures at two machine gauges, based on six different combinations of guide bar movements. The DOE contains a total of twelve samples, as shown in Table 3.3. The second (GB1) and third (GB2) column represent the movements of the guide bars, expressed in an amount of needles lapped. This amount expresses between how many needles the guide bar shifts during the underlap. For example, when two needles are lapped the guide bar shifts between two needles, which results in a tricot stitch. The minimum lap is over one needle, which results in a pillar stitch. Pattern GB 1 [# needles lapped] Table 3.3: Samples in the main DOE GB 2 [# needles lapped] Gauge 12 [with 15/2 Nm] Gauge 8 [with 11/2 Nm] Tricot-pillar 2 1 X X Cord-pillar 3 1 X O Satin-pillar 4 1 X X Tricot-tricot 2 2 X X Cord-tricot 3 2 O O Satin-tricot 4 2?? Legend: X = sample made; O = limited material availability;? = limited by machine 36

49 The choice is made to utilize closed loops as much as possible, except for the pillar stitch. Although open loops could have certain advantages for the deformability of HRSM fabrics (Chapter 1), it is technically more difficult to form open loops with SS yarn (other than the pillar stitch) compared to closed loops. First, the gauge eight samples were made to test the knit-ability of SS yarn, when this proved successful the gauge twelve samples were made. Two samples were not made in gauge eight, with the main reason being the limited amount of Nm 11/2 yarn available. Keeping in mind however, that the currently used weft knitted fabrics are produced on machines with gauge twelve or higher, the decision is made to focus more on the gauge twelve samples. The production of the satin-tricot sample was not successful on both gauges, due to the technical difficulties in knitting this structure faultless with SS fibre yarn. For each structure a pattern card with chain notation and lapping diagram is made, together with 3D images of the structure to aid visualization. An example of such a pattern card can be seen in Figure 3.2. The other pattern cards can be found in Appendix A. Besides the machine gauge and guide bar movements, three other structural parameters are expected to influence the deformability of warp knitted structures. The first one is the use of inlay threads, which can reduce the elongation at break in course direction. The second one is the use of double needle overlaps, in which the overlap is done over two needles. As previously discussed in Chapter 1, double needle overlaps put a large amount of stress on both needles and threads, which is why it is technically too difficult to knit this structure with SS fibre yarns. However, single needle bed structures with inlay threads can be produced. Two variations of one structure with inlay threads are produced, differing in their course density. Additionally, the influence of the third structural parameter, the amount of needle beds, on the overall fabric properties is investigated. Four variations of double needle bed structures are produced. It is not possible to do a full scale DOE for these parameters, only a limited amount of variants are be made. The structures are summarized in Table 3.4. Similar to the main DOE samples, all detailed patterns can be found in Appendix A. Table 3.4: Samples for the investigation of other structural parameters Name Gauge Needle bed Yarn [Nm] Structure [# needles lapped] GB 1 GB 2 Pillar-inlay low dens. 12 Single 15/2 1 Inlay over 4 Pillar-inlay high dens. 12 Single 15/2 1 Inlay over 4 Double face 12 Double 15/ Double tricot 12 Double 15/2 2 2 Double cord 12 Double 15/ Double pillar with inlay 12 Double 15/2 1 Inlay over 4 37

50 Pillar-cord Lapping diagram and chain link notation 3D Images Technical face Guide bar threading Selvedge Technical back Figure 3.2: Example of pattern card 38

51 3.5 Characterisation of the textile structure The warp knitted fabrics are characterised by tests at room and process temperature. The characterisation of the fabrics at room temperature is the core of the thesis, but the high temperature tests are done to give a first indication of the high temperature behaviour. As discussed in Chapter 1, the mould has a maximum process temperature of 680 C At room temperature The following parameters are measured at room temperature: Wale density [number of wales per 10 cm] Course density [number of courses per 10 cm] Fabric weight [g/m 2 ] Fabric thickness [mm] Air permeability [l/(dm 2.min)] Tensile force at break in wale and course direction [N] Elongation at break in wale and course direction [%] Elongation at 49 N in wale and course direction [%] These parameters are measured by a variety of tests. Some of the tests are done according to ISO norms, others according to own specification. The utilised ISO norms are shown in Table 3.5. Test parameter [unit] Table 3.5: Overview of utilised ISO norms for tests at room temperature Norm Air permeability [l/dm 2 /min] ISO 9237:1995 Fabric weight [g/m 2 ] In a conditioned atmosphere (ISO 139) Fabric thickness [mm] ISO 5084:1996 Tensile force at break [N] ISO :1999 Elongation at break [%] ISO :1999 The apparatus used for the tensile tests is a Zwicki 1120 by Zwick Roell Instruments. The wale and course densities are measured by visually counting the amount of respectively wales and courses in 10 cm of fabric. The fabric thickness [mm] is measured with an Interapid thickness gauge with deep throat, reading to 0,001 mm. The presser-feet have a diameter of 30 mm and the pressure is 2 N. All samples are measured at least five times in the fabric areas specified by ISO 5084:1996. The fabric is measured in a relaxed state. 39

52 3.5.2 At process temperature In the press bending process, HRSM fabric is subjected to a maximum surface temperature of 680 C. This affects the fabric properties at both yarn and fibre level due to high temperature oxidation (HTO) effects. Two tests are performed to measure the influence of HTO on the deformability: an oxidation test and a sagging test. These tests only give an indication of the fabric tensile properties at process temperature and are not used for a detailed analysis. Additionally, scanning electron microscopy (SEM) is used to visualise the effects occurring at fibre surface by HTO Oxidation test The goal of an oxidation test is to simulate the high temperature effects within the furnace. The test consists of heating test material (fabric, yarn or fibres) in an oven at 680 C for 1 hour. After this, the samples are subjected to the same mechanical tests at room temperature. The results are then analysed to determine the effect of high temperature on the material properties. The test is performed at fabric, yarn and fibre level. The samples are prepared by clipping the material without tension on a metal frame, which goes in the oven. The preparation of a fabric sample is shown in Figure 3.3. Figure 3.3: Preparation of fabric sample for oxidation test The following parameters are then measured to indicate the effects of this degradation: Fabric tensile strength [N] in wale and course direction after 1 h at 680 C Fabric elongation at break [%] in wale and course direction after 1 h at 680 C Yarn tensile strength [N] Yarn elongation at break [%] Fibre tensile strength [N] Fibre elongation at break [%] The fabric tensile strength and elongation at break are measured according to ISO :1999 on the same apparatus as at room temperature. The tensile properties for yarns and fibres are 40

53 measured according to own specification. For these tests, a DMA Q800 of TA Instruments is used in the DMA Strain Rate mode. The test specifications are shown in Table 3.6. Table 3.6: Tensile test specifications for fibres and yarns Material Gauge [mm] Preload [N] Strain rate [%/min] Fibre 10 0,02 0,2 Yarn 20 0,5 0, Sagging test HRSM fabrics are subjected to mechanical loads during their application as mould covering fabrics in automotive glass production. The goal of the sagging test is to simulate the forces on the fabric generated by the vacuum suction and sticking force on the fabric after bending. It measures the resistance of the fabric to cyclic loading at a temperature of 680 C. The sample preparation is done by stretching the fabric in a metal ring with a pre-load of 10 kg. This load stands for the amount of stretching done on the fabric to pull it over the mould. The preparation procedure for the samples is shown in Figure 3.4. Figure 3.4: Sagging sample preparation One cycle of the test consists of loading the fabric with the plunger to 4 N and then returning to the initial position. The pressing distance necessary to reach the pre-set load of 4 N is measured. A summary of the settings is shown in Table

54 Table 3.7: Sagging test settings Number of cycles 100 Force [N] 4 Speed of plunger [mm/s] 5 Temperature furnace [ C] 680 The test procedure, schematically shown in Figure 3.5, is done as followed: 1) One cycle at room temperature 2) 100 cycles at 680 C. The distance measurement of the last cycle is used to determine the value of sagging, defined by Y-X [mm]. 3) Cooling of the sample to room temperature Figure 3.5: Schematic procedure of the sagging test Figure 3.6 shows the test set-up in the oven. After the test is performed, the sagging of the fabric can be seen (marked by the red arrow). Figure 3.6: Test set-up in oven with sample The online process diagram of the sagging test is shown in Figure 3.7. The red curve represents the vertical position of the plunger and the blue curve the force of the load cell. A higher value of 42

55 sagging means that the fabric is less resistant to cyclic loading at high temperature. This means that the plunger needs to press deeper (higher value for vertical position) to obtain the same load of 4 N. Similarly, the plunger will only start registering at a deeper distance since the fabric is not stretched anymore in the ring. This will be seen in the diagram, with the red curve shifting up towards the end of the test cycles. Figure 3.7: Typical process diagram of sagging test Scanning electron microscopy (SEM) SEM analysis is used to investigate the changes at the surface of the fibres due to high temperatures. A protective oxide layer will be formed at the surface of the fibres, which will influence the friction properties at both fibre and yarn level. With SEM the change in surface friction of the fibres and yarns due to high temperature oxidation can be visualized. The images were made at 15 kv and at different magnifications. However, a detailed analysis of the inter-yarn and fibre friction at process temperature is not investigated, this falls outside of the scope of this study. 43

56 Chapter 4 Test results In this chapter the test results of the samples described in Chapter 3 are given. From these results the effect of structural fabric parameters on the properties, important for an HRSM fabric, are determined. As discussed in Chapter 2 those properties are: Fabric deformability, determined by the tensile force at break (> 49 N), the elongation at 49 N and a ratio w/c approaching one. Preferably the elongation at 49 N is minimal to have a fabric with sufficient stiffness. The ratio w/c is defined by the ratio of elongation at 49N in wale and course direction. Sufficient air permeability, preferably above 800 l/(dm 2.min). Thickness, preferably lower than 2 mm to enable adequate heat transfer between mould and glass. The deformability properties are the most critical parameters for the drape ability of the fabric over the mould, and determine the quality of the HRSM fabric. Therefore these parameters are investigated and compared for the different types of fabrics, in order to relate the structural fabric parameters to the fabric properties. 4.1 Effect of structural textile parameters on fabric properties In this paragraph the influence of machine gauge, underlap movement GB 1 and 2, inlay threads and the effect of single or double needle bed on the fabric properties are investigated Effect of gauge and underlap movement GB 1 The fact that due to technical complications not all samples in the main DOE can be made brings along difficulties to do a full factorial analysis. This is why the main DOE is split up into two smaller sub-doe s to analyse the effects of both guide bars and the machine gauge. Sub-DOE A investigates the effect of two independent parameters: the machine gauge and the underlap movement of GB1. In this set of samples the underlap movement of GB 2 is kept constant at one needle lap (pillar stitch). Two variants of two factors result in a DOE with four samples. A summary of the samples with their structural details is shown in Table 4.1. The cord-pillar sample 3A technically does not fit in the DOE, but it is added to analyse the gauge twelve samples further in a one-way ANOVA. 44

57 Name GB 1 (# needles lapped) Table 4.1: Sub-DOE A GB 2 (# needles lapped) Gauge 12 (sample nr.) Gauge 8 (sample nr.) Tricot-pillar 2 1 2A 2B (Cord-pillar) 3 1 3A O Satin-pillar 4 1 4A 3B Legend: O = not made due to material availability Five measurements were done of each sample. In Table 4.2 the average values for the dependent variables of each sample are shown. The detailed test results, which are used in the statistical analysis, can be found in Appendix B. In theory, the wale and course density should be equal for all samples. The wale density is mainly determined by the gauge, which is fixed at gauge eight for the B samples and at twelve for the A samples. Even so, there is a difference between the gauge eight or twelve samples. This is due to the structural contraction of the fabric after knitting and the washing process to remove the knitting oil. The contraction is different for the different knitting patterns. The patterns with the highest underlap length will have the highest contraction and therefor the highest wale density. The course density is less influenced by the knitting pattern and is mainly determined by the machine take-off setting. However, due to technical reasons it is not always possible to obtain a similar value for each knitting pattern. Table 4.2: Summary of average test results of sub-doe A Wales/ Courses/ Wale_BF Course_BF Wale_E49 Course_E49 Ratio AP Thickness Nr. 10 cm 10 cm (N) (N) (%) (%) w/c (l/(dm 2.min)) (mm) 2B , ,54 3B , ,69 2A (239) ,53 3A , ,77 4A , ,95 Legend: AP = air permeability ; BF = tensile force at break ; E49 = elongation at 49 N; ( ) = elongation at break Figure 4.1 shows the effect on the tensile force at break in both wale and course directions. It can be expected that the underlap movement of GB 1 will not have a significant influence on the breaking force in the wale direction (top chart), since the underlap is oriented in course direction. This is confirmed in the pareto chart of standardised effects (top-right) which is lower than p = 0,05 for GB 1. However, the machine gauge does have a significant influence in wale direction. Although the yarn used in the gauge 12 samples is less strong (Nm 15/2 versus Nm 11/2) than in the gauge 8 samples, the tensile force at break is higher for the gauge 12 samples. This indicates that the gauge itself has a significant effect on the tensile force at break. The higher wale density, resulting from the higher gauge causes the increased tensile force at break. The effect can be clearly seen in the top-left means plot. 45

58 In course direction, both the machine gauge and underlap movement of GB 1 have an effect on the tensile force at break. When the movement of GB 1 increases, then the tensile force at break will increase because the underlaps are longer and straighter in the course direction. This effect is illustrated in the pareto chart of standardised effects in Figure 4.1. From this chart it can also be seen that a higher gauge results in a lower tensile force at break. However, this can attributed to the weaker yarn used in the gauge 12 tests. In theory, the machine gauge should not have a significant effect on the tensile force in course direction. The small interaction seen in the 1 by 2 bar is due to the higher course density of sample 3B. Figure 4.1: Effect of GB 1 and gauge on the breaking strength for wale (top) and course (bottom) direction in a means plot with 95 % confidence levels (left) and a pareto chart of standardised effects (right) When considering the preferred value for tensile force at break, all samples but 2A (2B barely) have values higher than 49 N. Because sample 2A has a tensile force at break lower than 49 N in course direction, the elongation at 49 N can only be determined in the wale direction, as shown in Figure 4.2. The value for elongation in course direction in Table 4.2 is substituted by the elongation at break. The results suggest that the wale elongation at 49 N decreases for a higher underlap length and gauge. From a theoretical point of view, this effect is difficult to explain, especially since samples 2A and 4A have the same course density. In wale direction the elongation is mainly determined by the elongation of the open pillar stitch. An increased density of pillar stitches would 46

59 in theory not affect this elongation, unless there is a significant interaction between the pillars. In these samples however, the connection by the tricot or satin loops is the interaction between the pillars. It can be expected that this interaction is higher for a longer underlap, for example a satin loop, since more than two adjacent pillars (as in the tricot loop) are connected. In course direction, no conclusions can be drawn between both machine gauges since for sample 2A break was reached at 19 N. However, when comparing the values in Table 4.2 for samples 2B and 3B, respectively 155 % versus 49 %, it can be concluded that an increased movement of GB 1 significantly reduces the elongation at 49 N in course direction. A lower elongation at 49 N means that the fabric is more stiff and provides a better response to the applied force. The reduction in course elongation also results in an increase from 0,25 to 0,67 for the ratio w/c, which means that the properties in both directions become more equal. Figure 4.2: Effect of GB 1 and gauge on the elongation at 49 N for wale (top) and course (bottom) direction in a means plot with 95 % confidence levels (left) and a pareto chart of standardised effects (right) In Figure 4.3, the variation of the air permeability is shown. A higher underlap length decreases the air permeability because the structure becomes more closed and has less voids. Increasing the gauge from eight to twelve increases the wale density, which also makes the fabric less air permeable. However, even for the highest underlap length (over four needles) the air permeability is still well above the preferred value of 800 l/(dm 2.min) for HRSM fabrics. In the pareto chart (right) correlation can be found between gauge and movement GB 1 in the 1 by 2 bar. However, this can be considered as distortion since sample 3B (Gauge 8; GB 1 = 4) has an increased number of courses per cm compared to sample 2B. This increased course density results in less air permeability. Figure 4.4 shows the variation in fabric thickness. The underlap length of GB 1 clearly affects the thickness, as can be seen in the pareto chart. The difference between a tricot-pillar and a satinpillar structure is that the underlap in a tricot-pillar is between two adjacent wales, while in a satinpillar it laps across two wales. When the underlap is laid upon a pillar lap this adds thickness to the fabric. The machine gauge should in theory not have a significant influence, though the plot shows 47

60 correlation between machine gauge and fabric thickness when GB 1 laps over four needles. This could be explained by the fact that a higher gauge results in a denser structure and therefor the longer underlaps will lie closer, and perhaps even cross-over. Figure 4.3: Effect of GB 1 and gauge on the AP in a means plot with 95 % confidence levels (left) and a pareto chart of standardised effects (right) Figure 4.4: Effect of GB 1 and gauge on the thickness in a means plot with 95 % confidence levels (left) and a pareto chart of standardised effects (right) The effects of the movement of GB 1 on the dependent variables can be checked with a one-way ANOVA, in which sample 3A (movement GB 1 = 3) is included. The requirements for ANOVA are confirmed with two tests. First, the Levene test is performed to test the homogeneity of variance between the different groups of data. Second, the Shapiro-Wilk test is done to test the normality of the data in one group. The data passed both tests and therefor ANOVA is applicable. The average test values of sample 3A for air permeability and thickness are situated as expected between sample 2A and 4A, which can be seen in Table 4.2. A higher underlap length results in a fabric with lower air permeability and a slightly higher thickness. The effect of GB 1 on the absolute 48

61 values for the tensile force at break and elongation at 49 N is shown in Figure 4.5. For the tensile force at break (top plots) the effects found in the DOE analysis are confirmed. The underlap length has no significant effect on the wale tensile force at break, but a longer underlap increases the tensile force at break in course direction. It should be mentioned that the tensile force at break of the tricot-pillar (2A) is insufficient for use as an HRSM fabric (> 49 N). For the wale elongation at 49 N (bottom plot), the trend found in the DOE analysis cannot be confirmed since the value for the cord-pillar sample is significantly higher than the tricot-pillar sample. The interaction between the different wales, as previously suggested, does not seem to have an influence other than in the satin-pillar sample 4A. In course direction the elongation trend, previously found in the gauge 8 samples, is confirmed for gauge 12. A higher underlap length will decrease the course breaking elongation significantly. The data suggest that the satin-pillar sample is the best option for application as HRSM fabric, due to its lowest course elongation at 49 N of 88%. When looking at the w/c ratio, this sample does not have the best ratio (0,3), because the wale elongation at 49 N does not decrease accordingly. However, the first step is to obtain low elongations at 49 N before finding the sample with the best ratio w/c. Fabrics with ratio w/c equal to one but elongations above 200 % are not interesting for an HRSM. Figure 4.5: Effect of GB 1 on tensile force at break and elongation at 49 N in a means plot with 95 % confidence levels 49

62 4.1.2 Effect of gauge and underlap movement GB 2 Sub-DOE B investigates the effect of the machine gauge and the underlap movement of GB 2 on the fabric properties. Similar to sub-doe A, this also results in a DOE with four samples. The underlap movement of GB 1 is kept constant at a two needle lap (tricot stitch). It is investigated whether the substitution of the pillar by a tricot on GB 2 improves the overall fabric properties for use as HRSM. A summary of the samples with their structural details is shown in Table 4.3. Similar to sub-doe A, five measurements were done of each sample. In Table 4.4 the average values for the dependent variables of each sample are shown. The detailed test results that were used for the statistical analysis can be found in Appendix B. The tensile force at break in course direction of sample 1A could not be measured since these samples have an elongation at break that exceeds the maximum extension of the testing apparatus (250 %). The test was stopped before break was reached. The obtained tensile force is lower than 49 N, which means that the elongation at 49 N cannot be determined. Table 4.3: Sub-DOE B Name GB 1 (# needles lapped) GB 2 (# needles lapped) Gauge 12 (sample nr.) Gauge 8 (sample nr.) Tricot-pillar 2 1 2A 2B Tricot-tricot 2 2 1A 1B Table 4.4: Summary of average test results sub-doe B Wales/ Courses/ Wale_BF Course_BF Wale_E49 Course_E49 Ratio AP Thickness Nr. 10 cm 10 cm (N) (N) (%) (%) w/c (l/(dm 2.min)) (mm) 1B , ,54 2B , ,54 1A ((18)) 32 ((248)) ,62 2A (239) ,53 Legend: AP = air permeability ; BF = tensile force at break ; E49 = elongation at 49 N; (( )) = not measured to break ; ( ) = elongation at break First the deformability of the samples is investigated. Figure 4.6 shows a higher tensile force at break in wale direction for the tricot-tricot structure compared to the tricot-pillar structure. The additional connection between the wales strengthens the structure by stabilizing it and dispersing the tensile forces in an additional direction. For the same reason the tensile force at break in course direction should also be slightly higher for the tricot-tricot structure. When looking at the gauge eight samples, this difference is significantly higher due to the increased course density for the tricot-tricot sample. No difference can be seen between the gauge twelve samples (bottom-left plot) but this is because the maximum value for sample 1A could not be determined. The gauge 12 data for tensile force in course direction is thereby inconclusive. It can be concluded though from the gauge 8 data that the tensile force at break for small GB movements depends more on the 50

63 course and wale densities, other than on the presence of an additional small underlap like in the tricot-tricot pattern. Figure 4.6: Effect of GB 2 and gauge on the breaking strength for wale (top) and course (bottom) direction in a means plot with 95 % confidence levels (left) and a pareto chart of standardised effects (right) The wale elongation at 49 N in the top-left plot of Figure 4.7 appears to be lower for the tricottricot structure than for the pillar-tricot structure. The difference between the gauge eight samples is higher due to a higher course density of the tricot-tricot sample (GB 2 = 2). A higher course density decreases the wale breaking elongation due to shorter loop lengths. The gauge twelve samples have the same course density and therefor can be compared. The gauge 12 data show that the effect of pillar versus tricot on the wale breaking elongation is minimal, but that there might be a slight decrease in favour of tricot loops. The extra connection for the tricot-tricot between the wales could cause this effect. The data for course elongation at 49 N (and thus ratio w/c) is only available for the gauge 8 samples (Figure 4.8). The plot for course elongation (left) shows an increase for the tricot-tricot structure compared to the pillar-tricot structure, which is opposite to the effect found in wale direction. The effect is difficult to explain theoretically since the underlap length between the 51

64 wales stays constant in both patterns. A possible reason could be that the pillar loops limit the extensibility in course direction by keeping the tricot loops vertically together. The ratio w/c decreases when the underlap movement by GB 2 is increased, as shown in the plot (right). The extra connection between the wales in a tricot-tricot pattern decreases the wale elongation, but increases it in course direction. This leads to an overall decrease of the w/c ratio, which is less interesting for the application as HRSM fabric. It can therefore be concluded that the use of a tricot loop instead of pillar loop does not improve the deformation characteristics. Figure 4.7: Effect of GB 2 and gauge on the elongation at 49 N in wale direction in a means plot with 95 % confidence levels (left) and a pareto chart of standardised effects (right) Figure 4.8: Effect of GB 2 on the elongation at 49 N in course direction (left) and on the ratio w/c (right) in a means plot with 95 % confidence levels Figure 4.9 shows the variation of the air permeability for the different structures. A higher gauge and tricot-tricot structure results in a lower air permeability due to the higher loop density. The distortion seen in the plot and the high 1 by 2 interaction in the pareto chart can be attributed 52

65 again to the higher course density of sample 1B (Gauge 8; GB 2 = 2). However, all samples still have air permeability above the preferred value of 800 l/(dm 2.min). Theoretically it can be expected that there will be not a significant difference in fabric thickness between a tricot-pillar and tricot-tricot structure, since both have the same underlap. This is confirmed in Figure It can be concluded that increasing the movement of GB 2 does not improve the overall characteristics of the fabric for use as an HRSM. Figure 4.9: Effect of GB 2 on the air permeability in a means plot with 95 % confidence levels (left) and a pareto chart of standardised effects (right) Figure 4.10: Effect of GB 2 on the fabric thickness in a means plot with 95 % confidence levels (left) and a pareto chart of standardised effects (right) 53

66 4.1.3 Effect of inlay threads and take-off speed The third method to tailor the deformability of knitted fabrics is to insert inlay threads in the structure. The effect of an inlay thread in a warp knitted structure is investigated by comparing the satin-pillar and the pillar-inlay (over 4 needles) pattern. Figure 4.11 shows the structural differences between both patterns. In the pillar-inlay pattern the threads of the second guide bar are not knitted in the structure but laid into the pillar loops. The consequence is that the fabric weight and thickness is significantly lower, while the air permeability is almost twice as high, as can be seen in Table 4.5. This means that a pillar-inlay structure shows certain advantages as an HRSM fabric in comparison with the satin-pillar pattern. Examples of such advantages are the higher air permeability and lower thickness, which improves the suction force and heat transfer between mould and glass, while also reducing the energy consumption. Figure 4.11: Pillar stitch combined with knitted loop (left) and inlay (right) over four needles Nr. Wales/ 10 cm Courses/ 10 cm Table 4.5: Average test data of satin-pillar and pillar-inlay structure Wale_BF (N) Course_BF (N) Wale_E49 (%) Course_E49 (%) Ratio w/c AP (l/(dm 2.min)) Thickness (mm) 4A , ,95 5A , ,40 5A , ,19 Legend: AP = air permeability ; BF = tensile force at break ; E49 = elongation at 49 N Samples 4A and 5A_1 are compared to determine the effect of inlay threads versus knitted loops. Both samples are made at the same take-off speed, so their course density is near to equal. In Table 4.5 it can be seen that the elongation at 49 N in wale direction does not differ much. The values of 4A and 5A_1 are very close, with respectively 27 % versus 25 %. In course direction the 54

67 presence of inlay threads compared to knitted loops does have a significant effect, as shown in Figure The strong decrease in elongation at 49 N in course direction (left) also results in an increase of the ratio w/c (right) from 0,30 to 0,37. Figure 4.12: Comparison of course elongation at break (left) and ratio w/c (right) between satin-pillar and pillar-inlay structure in a means plot with 95 % confidence levels The fourth structural parameter that influences the fabric properties is the take-off speed. A lower take-off speed will result in an increased course density, characterised by shorter loops. To determine this effect, a second variant 5A_2 of the pillar-inlay sample is produced with higher course density, with respectively 51 courses per 10 cm versus 41. The fabric properties are also shown in Table 4.5. The most important parameters, when comparing these two structures, are their mechanical deformation properties. The elongation at 49 N in wale direction is lower due to the shorter loops, with 20 % versus 25 %. The change in elongation at 49 N in course direction is even higher, with 45 % versus 69 %. This increases the ratio w/c ratio significantly to a value closer to 1 (Figure 4.13). Figure 4.13: Comparison of course elongation at 49 N (left) and ratio w/c (right) between two pillar-inlay samples with different course densities in a means plot with 95 % confidence levels 55

68 4.1.4 Effect of the amount of needle beds The fifth structural parameter that influences the properties of warp knitted fabrics is the amount of needle beds. In order to determine this influence, some basic double bed structures were made to compare with the single bed ones. Due to high structural differences between single and double needle bed patterns the fabrics cannot easily be compared with one another. Four different types of double needle bed structures were produced and tested according to the same procedures as for the single needle bed samples. The pattern details are shown in Table 4.6 and the detailed pattern visualisations can be found in Appendix A. Table 4.6: Pattern details of double needle bed samples Name Gauge Number Yarn (Nm) Structure (# needles lapped) GB 1 GB 2 Double face 12 6A 15/ Double tricot 12 7A 15/2 2 2 Double cord 12 8A 15/ Double pillar with inlay 12 9A 15/2 1 Inlay over 4 The properties of these fabrics are compared with the two most promising warp-knitted structures from the previous paragraphs: samples 4A and 5A. The data of all samples are shown in Table 4.7. When comparing the tensile properties of the samples, it is found that sample 8A fulfils the minimum tensile force at break of 49 N in both directions. Samples 6A and 7A have an elongation that is too high to be measured and thus the force at break could not be determined. Sample 9A fails to fulfil the requirement in course direction. When the course elongation at 49 N is compared with the single needle bed structures it is seen that the value is more than twice as high, even for the double needle bed inlay variant. Additionally, the thickness is almost two times higher when compared to sample 4A, which would require a higher furnace temperature. From these results it can be concluded that double needle bed warp knitted fabrics are not promising to use as HRSM fabrics, mainly due to their higher elongations at 49 N in course direction. Nr. Wales/ 10 cm Courses/ 10 cm Table 4.7: Average test data of double needle bed samples Wale_BF (N) Course_BF (N) Wale_E49 (%) Course_E49 (%) Ratio w/c AP (l/(dm 2.min)) Thickness (mm) 4A , ,95 5A , ,40 5A , ,19 6A ((33)) 58 ((249)) ,24 7A ((24)) 66 ((250)) ,26 8A , ,51 9A (193) ,27 Legend: BF = tensile force at break ; E49 = elongation at 49 N; ( ) = elong. at break ; (( )) = not measured to break 56

69 Tensile force (N) 4.2 High temperature characterisation At high temperatures, the metal fibres will be subjected to high temperature oxidation (HTO), and will go through metallurgical transformations. This will affect the deformability and stability of the knitted fabric once it is stretched over the mould. The focus of these experiments is to indicate the change in fabric properties by investigating the effect of HTO on the fibre, yarn and fabric tensile properties. First, the effect on the fibres will be investigated, then on the yarns and finally in the fabrics to determine the correlation between the three levels. Additional measurements, for example of the inter yarn friction, are not examined in this thesis but are necessary to fully characterise the fabric behaviour Effect on the fibre properties The effect on the yarn properties is determined by comparing the tensile tests of oxidised fibres and non-oxidised fibres. The two types of yarns used, Nm 11/2 and 15/2, are both spun from 12 µm fibres. The tensile force of these fibres was determined before and after oxidation. The oxidation process consisted of heating the fibres for one hour at 780 C. Ten samples of each type were tested. Figure 4.14 shows the tensile curves of the tested fibres. The curves suggest that oxidised fibres have a significantly lower elongation at break and as well as a lower tensile force at break. This can be explained by the formation of metal oxides on the surface of the fibre that makes the fibres thinner and less strong Elongation (%) Figure 4.14: Tensile curves of oxidised (- -) and non-oxidised ( ) 12 µm stainless steel fibres 57

70 Table 4.8 shows the maximum values of the ten samples before and after oxidation. The data are used for an ANOVA analysis to determine if the difference pre and post oxidation is significant. Figure 4.15 clearly shows that HTO will lower the fibre tensile force and elongation at break. Test nr. Tensile force at break (N) Table 4.8: Fibre test data Non-oxidised SS Fibre ø 12 µm Elongation at break (%) Tensile force at break (N) Oxidised Elongation at break (%) 1 0,1695 1,790 0,131 1, ,1579 1,737 0,1254 1, ,1572 1,504 0,1166 0, ,1669 1,844 0,1185 1, ,1324 1,385 0,1333 1, ,1692 1,738 0,1276 1, ,1462 1,451 0,1351 1, ,1551 1,505 0,1272 1,076 Average 0,1568 1,619 0,1268 1,124 Stdev 0,0119 0,165 0,0062 0,124 Figure 4.15: Effect of HTO on fibre tensile force (left) and elongation (right) at break in a means plot with 95 % confidence levels The presence of metal oxide fragments on the fibre surface can be seen by optical analysis with a scanning electron microscope (SEM). Figure 4.16 shows the presence of the metal oxide fragments on the fibre surface. The metal oxide fragments are brittle and make the fibre surface less strong. 58

71 Tensile Force (N) Figure 4.16: SEM images of oxidised 12 µm fibres at 1000x (left) and 5000x (right) Effect on the yarn properties The effect on the yarn properties is determined by comparing the tensile test results of oxidised and non-oxidised yarns. Figure 4.17 and Figure 4.18 show the tensile curves for Nm 11/2 and 15/2 yarns. A clear effect of oxidation on the tensile properties of Nm 11/2 yarn can be seen in the plot. The oxidised yarns have a lower tensile force and elongation at break compared to the non-oxidised samples. With an average fibre length of 80 mm and a test gauge length of 15 mm, the fibre tensile properties can be related to these results. The lower tensile force and elongation of the oxidised fibres will contribute to the effect seen in the figure. The inter-fibre friction undoubtedly also has an effect, due to the presence of metal oxide particles, but has not been investigated in this thesis Elongation (%) Figure 4.17: Tensile properties of oxidised (- -) and non-oxidised ( ) Nm 11/2 yarn 59

72 Tensile force (N) The effects for the Nm 15/2 yarn are not as clear as for the Nm 11/2 yarn. While all samples have a lower maximum force and strain, the slope of the curves are different. This variation could be explained by an unequal oxidation in the different sections of the yarn, possibly due to a higher amount of spinning oil Elongation (%) Figure 4.18: Tensile properties of oxidised (- -) and non-oxidised ( ) 15/2 Nm yarn The effects on both yarns are statistically analysed in a DOE with two factors, oxidised/nonoxidised and yarn count. The test values are shown in Table 4.9. Test nr. Tensile force at break (N) Table 4.9: Yarn test data SS 11/2 Nm SS 15/2 Nm Non-oxidised Oxidised Non-oxidised Oxidised Elongation Tensile Elongation Tensile Elongation Tensile at break force at at break force at at break force at (%) break (N) (%) break (N) (%) break (N) Elongation at break (%) 1 15,12 1,142 11,73 0,901 12,22 1,081 11,34 1, ,16 1,137 12,74 0,983 13,01 1,152 9,35 0, ,54 1,163 12,19 0,882 12,68 1,147 8,99 0, ,75 1,165 12,91 0,982 11,92 1,230 9,07 1, ,97 1,181 13,15 1,003 12,18 1,081 7,74 0, ,36 1,170 12,66 0,942 12,38 1,064 11,00 1, ,53 1,133 12,69 0,874 11,03 1,064 10,35 0, ,20 1,136 13,81 1,039 12,01 1,115 9,81 0, ,94 1,032 12,43 0,956 10,08 1,031 10,90 0, ,15 1,203 13,35 0,919 12,24 1,079 11,29 0,966 Average 14,37 1,146 12,77 0,948 11,98 1,104 9,99 0,959 Stdev 1,07 0,044 0,56 0,052 0,80 0,055 1,13 0,071 60

73 The results of the analysis are shown in Figure The tensile force at break (top) is significantly lower for Nm 15/2 yarn than for Nm 11/2 yarn, due to the higher fineness. Oxidation clearly lowers the tensile force at break, for both the Nm 11/2 and 15/2 yarn. There is no 1 by 2 interaction effect between the yarn count and oxidation treatment for tensile force. The elongation at break is not influenced by the yarn count: a thinner yarn does not necessarily result in a lower strain. Oxidation does have a clear effect: for both yarn counts it lowers the elongation at break significantly. Figure 4.19: Effect of HTO on the tensile force and elongation at break of Nm 11/2 and 15/2 yarns in a means plot with 95 % confidence levels (left) and pareto chart of standardised effects (right) 61

74 Ratio tensile force at break pre/post ox. (%) Effect on the fabric deformability The effect on the deformability is determined by the sagging and oxidation test. The oxidation test measures the influence of oxidation on the fabric tensile properties, while the sagging test is used to determine the stretch stability of the fabric under a cyclic force at high temperature Oxidation test The oxidation test consists of the tensile testing of oxidised fabric strips. Figure 4.20 shows the effect of oxidation on the tensile force at break at fibre, yarn and fabric level. Samples 6A, 7A and 9A are not included in the graph because their elongation exceeds the maximum values of the testing apparatus (> 250 %). There is a clear loss in the tensile force at break at all levels. The loss in tensile force at fibre level does not correspond with the loss at yarn level. This is because the tensile force at break of a yarn is mainly determined by the inter-fibre friction due to the yarn twist. At fabric level there is also a significant loss, which can be related to the strength loss of the yarns. Once the knitted structure has completely deformed under tension, the breaking load will be determined by the yarn properties. For the different fabric samples, the loss is not equal in both directions, but no trend can found between the samples Fibre Ø12 µm 11/2 Yarn 15/2 Yarn REF 1B 2B 3B 3A 4A 5A_1 5A_2 8A Sample Fibre Yarn Fabric wale direction Fabric course direction Figure 4.20: Ratio of tensile force at break pre- and post-oxidation for fibres, yarns and fabrics Figure 4.21 shows the effect of oxidation on the elongation at break at fibre, yarn and fabric level. There is a loss in elongation at break for all samples. Similar to the tensile force loss, the loss in elongation at fibre level does not correspond with the loss at yarn level. It can be expected that the lowered elongation at break of the yarns will results in a lowered value at fabric level. 62

75 Ratio elongation at break pre/post ox. (%) Fibre Ø12 µm 11/2 Yarn 15/2 Yarn REF 1B 2B 3B 3A 4A 5A_1 5A_2 8A Sample Fibre Yarn Fabric wale direction Fabric course direction Figure 4.21: Ratio elongation at break pre- and post-oxidation for fibres, yarns and fabrics Sagging test The sagging test represents the cyclic loading of the HRSM fabric. It measures the resistance of the fabric at high temperature to the vacuum force between mould and glass, and the sticking force when the glass is released from the mould. Due to the large amount of fabric required for this test, not all samples could be tested. The two most promising single needle bed patterns were tested, as were also three double needle bed samples. The sagging values for each sample can be found in Table Table 4.10: Sagging testing values Sample Sagging (mm) Weight (g/m 2 WaleE49 CourseE49 ) Ratio w/c (%) (%) 4A 25, ,30 5A_1 27, ,37 5A_2 25, ,44 6A 32, (249) - 7A 41, (250) - 8A 38, ,36 Legend: AP = air permeability ; BF = tensile force at break ; E49 = elongation at 49 N; ( ) = maximum elongation measured The five parameters previously discussed, that influence the fabric pattern, can be expected to have an influence on fabric sagging. The properties determined by the pattern such as weight, elongation at 49 N are investigated to determine which parameters influence the sagging. 63

76 Sagging (mm) The sagging is plotted in function of fabric weight in Figure Double needle bed fabrics with a higher weight appear to have higher sagging values, but this can be mainly explained by the different fabric structures (single versus double) which also have different elongation values. The higher weight is a direct consequence of the different structures. When the double needle bed structures are compared, those with higher weight have higher sagging values. However when the single needle bed structures are compared, the lightest fabric has the highest value for sagging (sample 5A_1). From this it can be concluded that the fabric weight does not necessarily have an influence on sagging Fabric weight (g/m 2 ) Figure 4.22: Correlation between fabric weight and sagging with R² = 0,7402 Sample 5A_2, which has exactly the same pattern as 5A_1 except with a higher course density, has less sagging, with 25,07 versus 27,99 mm. This indicates that the course density has an effect on sagging, because it affects the elongation properties and ratio w/c. The influence of the elongation at 49 N on sagging is determined in Figure 4.23 for the single needle bed samples. Fabrics with a higher elongation at 49 N will be more easily strained than those with a lower elongation. This could result in a higher susceptibility to sagging. However, when looking at the curve, the sample (4A) with the highest elongations at 49 N does not have the highest value for sagging. Both sample 4A and 5A_2 have similar values for sagging but a completely different structure and elongation values. This indicates that sagging is not determined by one structural parameter, but by a combination of them. Further research is necessary to determine the exact influence of the structural parameters. 64

77 Sagging (mm) Elongation at 49 N (%) WaleE49 (%) CourseE49 (%) Figure 4.23: Correlation between elongation at break and sagging for samples 4A, 5A_1 and 5A_2 4.3 Comparison of weft and warp knitted samples In this paragraph the most promising warp-knitted samples, derived from the test results, are compared with the reference weft-knitted sample. The samples, both weft and warp knitted, are all made on a gauge twelve machine with a Nm 15/2 yarn. The two most promising warp knitted patterns are the satin-pillar and pillar-inlay types. A summary of the values is shown in Table Table 4.11: Summary of test results for comparison with weft knitted reference sample Wales Courses Weight AP Thickness WaleBF CourseBF WaleE49 CourseE49 Ratio Nr /10 cm /10 cm (g/m 2 ) (l/dm 2.min) (mm) (N) (N) (%) (%) w/c REF , ,56 4A , ,30 5A , ,44 Legend: AP = air permeability ; BF = tensile force at break ; E49 = elongation at 49 N When the wale and course densities are compared, it is found that the densities for the weft knitted sample are both higher in wale and course direction. Still, the weight of the warp knitted samples, especially for the pillar-inlay sample, is not so much different. This is because two sets of threads are knitted into one warp knitted structure. For the pillar-inlay sample there is one guide bar and one inlay bar, which results in a weight comparable to the weft knitted reference. For the satin-pillar sample, the weight is higher because of the two guide bar structure, which knits two sets of threads on one set of needles. 65

78 The reference sample has about the same air permeability as the pillar-inlay sample, but a higher value than the satin-pillar sample. The long underlap is responsible for the lower air permeability, but is still well above the requirement (> 800 l/(dm 2.min)). This underlap is also responsible for the higher thickness, which is still lower than the preferred value (2 mm). The higher thickness could possibly be an advantage for the application in automobile glass production. It could have a higher compressibility and result in a window with improved optical quality. However, mould tests needs to be done to investigate this effect. When the tensile forces at break are compared, the difference between weft and warp knitting can be clearly seen. Warp knitted fabrics are stronger in warp direction, due to the pillar stitch. Weft knitted fabrics are stronger in course direction. When comparing the elongations at 49 N, it can be seen that the wale elongation at 49 N is significantly lower for the warp knitted samples. The weft knitted sample has a wale elongation at 49 N of 120 % versus 20 % for sample 5A_2. The same warp knitted sample also has a lower course elongation at 49 N: 45 % versus 68 %. Sample 4A has a slightly higher value: 88 %. However, the lower values for the pillar-inlay sample do not result in a ratio w/c that is closer to 1 compared to the weft knitted sample: 0,44 versus 0,56 for the weft knitted sample. This means that the weft knitted fabric has more equal properties in both wale and course direction. Finally, the sagging, which is an important factor for the lifetime of the fabric, can also be compared. The values are shown in Table The values for the warp knitted fabrics are almost equal with a value around 25 mm, while the weft knitted sample has a lower value of 21 mm. This means that the produced warp knitted samples do not yet improve the sagging compared to weft knits. Table 4.12: Sagging comparison between weft and warp knitting Sample Structure Sagging (mm) Reference weft knit Weft knit 21,18 4A Satin-pillar 25,36 5A_2 Pillar-inlay 25,07 The overall conclusion is that elongation properties have been improved compared to weft knitted fabrics but other important properties such as sagging and ratio w/c are still better for weft-knitted fabrics. 66

79 4.4 Summary This investigation has identified the most important drivers influencing the properties of warp knitted heat resistant separation fabrics. These drivers can be adjusted to tailor the properties of a warp knitted HRSM. In Table 4.13 the drivers and their effects on the fabric properties are shown. The effect of gauge on tensile force at break and elongation at 49 N in course direction is inconclusive, because they could either not be measured or reached break before 49 N as previously mentioned. The effect on the ratio w/c is not displayed since this depends on the exact increase or decrease of the elongation. Table 4.13: Important drivers for warp knitted HRSM Drivers AP Thickness WaleBF CourseBF WaleE49 CourseE49 Gauge =/? =/? Course density =/ =/ Underlap length =/ = = Knitted -> Inlay thread =/ = Single -> double bed =/ =/ Legend: AP = air permeability / E49 = elongation at 49 N / BF = tensile force at break =/ : small influence? = inconclusive By adjusting the wale and course densities in structures with a high underlap length or with inlay threads, a stable warp knitted fabric can be formed that would fulfil the requirements for HRSM fabrics. 67

80 Chapter 5 Conclusion Warp knitted fabrics have been investigated as an alternative to weft knitted heat resistant separation materials (HRSM). The applicability of warp knitted fabrics as HRSM is determined by five fabric properties: tensile force at break, elongation at 49 N, ratio wale/course elongation, air permeability and the fabric thickness. The influence of five structural parameters, machine gauge, course density, guide bar (GB) underlap movement, presence of inlay threads and single or double needle bed, on these fabric properties is determined in this study. A series of two guide bar warp knitted structures are selected with the aid of a visual 3D modelling program to investigate these parameters. Not all samples could be successfully made due to technical difficulties when knitting SS fibre yarn. The samples were subjected to a sequence of tests to determine the fabric properties. First, the parameters that influence the fabric thickness are looked into. It is found that the fabric thickness is little influenced by the structural parameters, except for single or double bed structures. The GB movement has little influence. There is a difference between pillar-tricot and pillar-cord fabrics but for an even longer underlap there is not much difference. Inlay threads do not influence the fabric thickness since they do not lie on top of the pillar stitches but are inserted inside the pillar loops. Single needle bed or double needle bed patterns have a strong influence on the thickness, with the double needle bed fabrics almost being twice as thick. Second, the parameters that influence the air permeability are determined. All parameters have a significant influence on the air permeability. A higher machine gauge, course density and a longer underlap result in lower air permeability. Structures with inlay threads are generally more open, which results in higher air permeability. Double needle bed fabrics are less air permeable due to their higher thickness. Third, the parameters that influence the tensile force at break are investigated. The tensile force at break is subdivided in the force in wale and course direction. In wale direction, the tensile force at break will increase for a higher gauge and for double needle bed structures instead of single bed. The other parameters do not have a significant influence. In course direction, the force will increase with an increased GB movement and increased course density. Switching from a knitted loop to an inlay thread, and from single to double bed will decrease the tensile force at break. 68

81 Fourth, the parameters that influence the elongation at 49 N are characterised. Again, the elongation is subdivided in wale and course direction. In wale direction, the elongation will increase with the course density and when switching from single to double bed. In course direction the elongation is strongly determined by the underlap length and the presence of inlay threads. Increasing the gauge and switching from single to double bed will increase the course elongation. Fifth and last, the parameters that influence the ratio wale/course elongation are looked into. Since this property is fully dependent on the value for wale and course elongation, the same effects found for the elongation at 49 N will affect the ratio w/c. Next, the parameters that influence the fabric properties at high temperature are investigated. The results show that the breaking load of the fabrics is clearly influenced by the high temperature. The oxidation of the fibres leads to a decrease in tensile force at break on fibre, yarn and fabric level. Sagging tests show that single bed structures better resist the cyclic loading at high temperature, compared to double bed structures. It was found that a combination of structural fabric parameters influences the sagging properties. By investigating the five structural parameters, two specific warp knits are found interesting for further research: the combination of a pillar stitch and a stitch with long underlap (e.g. satin), or the combination of a pillar stitch and inlay threads over at least four needles. The main parameter influencing these structures is the underlap or inlay length, determined by the guide bar movement. The most important properties for the application as an HRSM fabric are the elongation at 49 N and the ratio between wale and course elongation. The sample with inlay threads has the most promising values for elongation, ratio w/c and sagging. Following the results of this thesis, a patent application was filed regarding the use of the satin-pillar and pillar-inlay pattern for HRSM applications. The unidirectional tensile tests performed in this study cannot account for the interaction from the other fabric directions. Therefore, biaxial tensile testing of the fabrics is recommended for further research. General conclusion From the results it can be concluded that it is possible to knit a fabric from stainless steel fibre yarns by warp knitting technology. Two patterns have been identified as promising for use as an HRSM fabric. When compared to weft knitted fabrics, certain properties are improved but others remain better for weft knitted fabrics. Further research into these structures, with additional mould testing, will show if the warp knitted fabrics are a feasible alternative to weft knitting technology. 69

82 Appendix A: Warp knitted structures In this appendix the detailed schematics of the structures defined in the DOE are given. The first section includes an overview of the DOE samples in a table. The second section gives a detailed description of every structure, including the chain link notation with drawn structure, as well as 3D images generated by the TexMind program. Overview Single needle bed structures Name GB 1 [# needles lapped] GB 2 [# needles lapped] Tricot-pillar 2 1 Cord-pillar 3 1 Satin-pillar 4 1 Tricot-tricot 2 2 Cord-tricot 3 2 Satin-pillar 4 2 Pillar-inlay 1 Inlay over 4 Double needle tricot-tricot Double needle bed structures Name GB 1 [# needles lapped] GB 2 [# needles lapped] Double face Double tricot 2 2 Double cord Double pillar with inlay 1-1 Inlay over 4 70

83 Patterns Single needle bed patterns Tricot-pillar (TRPI) Lapping diagram and chain link notation 3D Images Technical face Technical back 71

84 Cord-pillar (COPI) Lapping diagram and chain link notation 3D Images Technical face Technical back 72

85 Satin-pillar (SAPI) Lapping diagram and chain link notation 3D Images Technical face Technical back 73

86 Tricot-tricot (TRTR) Lapping diagram and chain link notation 3D Images Technical face Technical back 74

87 Cord-tricot Lapping diagram and chain link notation 3D Images Technical face Technical back 75

88 Satin-tricot Lapping diagram and chain link notation 3D Images Technical face Technical back 76

89 Pillar-inlay (PINL) Lapping diagram and chain link notation 3D Images Technical face Technical back 77

90 Double needle tricot-tricot Lapping diagram and chain link notation 3D Images Technical face Technical back 78

91 Double needle bed patterns Double face (DNDF) Lapping diagram and chain link notation 3D Images Technical face Technical back Cross-section in course direction 79

92 Double tricot (DNDT) Lapping diagram and chain link notation 3D Images Technical face Technical back Cross-section in course direction 80

93 Double cord (DNDC) Lapping diagram and chain link notation 3D Images Technical face Technical back Cross-section in course direction 81