Analysis of mechanical properties of non woven fabrics Prepared by: jemal assres

Analysis of the Mechanical Properties of Woven and Nonwoven Fabrics as an Integral Part of Compound Fabrics
Nonwovens are normally made from continuous filaments or from staple fibre webs or batts strengthened by bonding using various techniques such as adhesive bonding, mechanical interlocking by needling or fluid jet entanglement, thermal bonding and stitch bonding [3]. With the proper selection of fibres, technological process of web production, and methods of web bonding and finishing, the desired properties and applicability of nonwovens can be achieved. The mechanical properties of nonwovens depend on the constructional parameters of the web (the basic structure element of the web, the fibres composition and orientation, the method of web production), and the type of web bonding and finishing [4]. In the case of needle punch production, it has been stated that the tensile properties of needle punched fabrics are above all highly dependent upon the frictional properties of the constituent fibres, as well as the way in which the fibres are arranged [5, 6, 7]. Comparing the mechanical properties of nonwoven and woven fabrics, the mechanical properties of the latter are shown to be much better. The use of woven fabric in the compound fabric structure is therefore inevitable in order to achieve proper mechanical properties. Woven fabric as an integral part

of compound fabric has a reinforcing role. Woven reinforcement exhibits good stability in the warp and weft direction, and offers the highest cover or yarn packing in relation to fabric thickness [2]. When engineering woven fabric as reinforcement in the compound fabric for specific end use (conveyor belts), the following project demands should be taken into consideration: high breaking strength, low elongation at break, high Young s module level, high tensile force by 1% of elongation, and fabric thickness lower then 1 mm [8]. Woven fabric mechanical properties are the function of the mechanical properties of the fibres used, the mechanical properties of constituent yarns, the woven fabric geometry (yarn linear density, thread density, type of weave) and the technological parameters of spinning, weaving and finishing processes [9]. By studying the load-extension properties of woven fabrics, most researches focused on the basic weaves, rather then the fabric s tightness [10]. In the case of Malimo fabric, which can also be used as a reinforcement instead of as a woven fabric, the mechanical properties depend on the mechanical properties of longitudinal, transversal and stitching yarns, as well as on the type of stitching. The mechanical properties of a compound fabric are not simply the arithmetical sum of the properties of a separate layer in a compound fabric. When engineering needle-punched NWN or NMN compound fabric, not only woven or nonwoven fabric geometry parameters must be taken into account. Another very important parameter is the number of reinforcement layers of fabric in the Abstract In this paper the difference between the mechanical properties of woven and nonwoven fabrics is presented, and the important role of constructional parameters (particular the cover factor) of reinforcement fabric in the compound fabric is stressed. Some basic mechanical properties, such as tensile strength, elongation at break, work to rupture and the force by 1% of elongation were measured and analysed. The stress/strain diagrams show the differences in the behaviour under defined loads of different woven and nonwoven fabrics. The results show that the fabric with good mechanical properties is not always suitable as reinforcement, where the needle punching is used as a method of linking layers into compound fabric. Key words: technical textiles, compound fabrics, construction parameters, cover factor, reinforcement fabric, mechanical properties.

Analysis and Measuring Surface Roughness of Nonwovens Using Machine Vision Method Abstract Concerning the measurement of friction properties of

textiles and fabrics using Kawabata Evaluation System (KES), whose output is constrained to the surface friction factor of fabric, and no other data would be generated; this research has been conducted to gain information about surface roughness regarding its surface friction factor. To assess roughness properties of light nonwovens, a 3-dimensional model of a surface has been simulated with regular sinuous waves through it as an ideal surface. A new factor was defined, namely Surface Roughness Factor, through comparing roughness properties of simulated surface and real specimens. The relation between the proposed factor and friction factor of specimens has been analyzed by regression, and results showed a meaningful correlation between them. It can be inferred that the new presented factor can be used as an acceptable criterion for evaluating the roughness properties of light nonwoven fabrics. Keywords Surface roughness, Nonwoven, Machine vision, Image processing. I. INTRODUCTION ODAY, image analysis techniques are used in many aspects of engineering works, these techniques have been used in textile and nonwoven technologies to measure and control their different features. Specifically in nonwovens, image analysis is a quite reliable and reputable technique to measure uniformity, cover factor, surface roughness, etc. Some research topics were dedicated to measuring nonwoven mass uniformity and homogeneity as it plays an important role

in these fibrous structures [1,2]. There are researches were undertaken to analyze structure of nonwoven fabrics using image analysis [3,4]. In the present work, nonwoven fabrics were modeled into a 3 dimensional structure for measuring their roughness; in fact modeling textile layers to a 3-D format to measure their properties is not a new idea and was also used in other researches [5,6,7]. In a research nonwoven fabrics were simulated in a 3-D virtual structures and a geometric model was considered to simulate nonwoven webs. In another attempt, textiles were modeled to 3 dimensional form to measure their wrinkles by the photometric stereo technique. In addition, Sul et al. measured surface roughness using three dimensional profile data. In their system they used box counting method to calculate fractal dimension and then reconstructed an image for obtaining surface roughness data. Moreover, some researchers presented new methods on objective measurement of fabric surface roughness by image Dariush Semnani is with Textile Engineering Department of Isfahan University of Technology 84156-83111, Iran (e-mail: d-semnani@cc.iut.ac.ir). analysis. They introduced two parameters derived from fabric images which are fractal dimension calculated by wavelet fractal method and surface average mean curvature. These two parameters describe surface roughness [8,9]. In another study a solution which provides a method to measure roughnessfriction criteria of tested surface introduced [10].

Surface roughness in Kawabata is measured by pulling a Ushaped steel wire with 0.5 mm diameter through the fabric length under a normal force of 10 gf. In fact, height is measured by this system, and a height profile of fabric through its length concluded from this experiment. Mean deviation of this profile is considered as surface roughness in Kawabata system. The problem in Kawabata surface roughness measuring is twofold. First, changes in height along the fabric width are not considered as the height is measured along the fabric width in this system. Second, friction of fabric is not a point of issue in measuring roughness, which plays an important role in roughness. This is mostly because objective of Kawabata from measuring surface roughness is evaluating fabric handle not measuring roughness itself, as a matter of fact, friction coefficient is considered as a criterion of fabric handle but not surface roughness in KES. In a research presented by Govindaraj, et al. a novel method had been proposed to measure fabric roughness which simulate woven fabric surface using two signals. These signals had been generated from friction profile of fabric in two direction of warp-wise and weft-wise, and then these two orthogonal signals have been combined to reconstruct a 3D surface profile of fabric [11]. This method is constrained to fabrics with regular surface i.e. woven fabrics. Hence random

structured fabrics such as nonwovens couldn t be analyzed in these systems from surface roughness point of view; furthermore, a need emerged to measure surface roughness accurately, and simply. These conditions led this research to an image processing based method. As random textile structures become more and more applicable so today many researches were focused on developing new, simple and fast measurement method for evaluating different characteristics of these layers. As a result, presenting a novel method which can evaluate surface roughness by image processing techniques is essential for the nonwoven industries. In this research an image processing based method was used to model surface of nonwoven layer by turning their 2-D images into a 3-D structure, and then its surface roughness Analysis and Measuring Surface Roughness of Nonwovens Using Machine Vision Method Dariush Semnani, Javad Yekrang, and Hossein Ghayoor T World Academy of Science, Engineering and Technology 57 2009 543 had been measured. Finally its relation with their friction coefficient was investigated. II. METHODS To evaluate surface roughness a reference is needed which

fabric surfaces could be compared to it. In fact this reference is nothing but a layer which presents most pleasant roughness for human tactile sensation, this layer is called ideal surface in this paper. Ideal surface is a surface with regular sinuous waves with least amplitude and wavelength which could be sensed by human tactile sensation, this surface has the most comfortable handle sense for textiles; moreover, it presents the ideal friction of textiles for human body. As the friction coefficient of textiles are generally related to their touch

Characterization of Non-Woven Fabrics Using DSC W.J. Sichina, International Marketing Manager INTRODUCTION One important class of polymer materials is non-wovens fabrics. These fabrics or fabric-like materials are made from semi-crystalline thermoplastics. The processing of the polymer non-wovens results in the generation of a material with significant open spaces or voids.

Applications of polymer nonwovens include: Disposable diapers Shipping materials Disposable garments Sanitary napkins Carpet backing Weed blocks for gardening and landscaping Construction covers Envelopes Floppy diskette liners Road construction components Filtration media Floating row covers Since polymer non-woven materials are produced from semi-crystalline thermoplastics (such as PET or polypropylene), it is important to know their critical thermo-physical properties, including: Melting temperature Heat of melting Percent crystallinity Glass transition temperature(s)

 Heat set or processing temperatures Differential scanning calorimetry (DSC) provides an easy-to-use means of characterizing the important thermal properties associated with polymer non-woven materials. PerkinElmer offers different high performance DSC instruments which cover a wide range of applications and needs. DSC provides valuable information on the processing and the development of structure of nonwoven fabrics. Spunbonding is one of the most widely used means of producing non-woven fabrics. This involves extrusion of the polymer, followed by non-isothermal extensional flow of the melt, crystallization and orientation. The filaments are sucked through a venturi (high velocity low pressure zone) and a distributing chamber, which ensures entangling of the

filaments. The entangled filaments are deposited as a random web on a moving belt for thermal calendaring [1]. In order to bond the filaments, the web is thermally treated. This generally involves moving the web through heated rollers. The resulting bonding of the filaments is influenced by the temperature and the pressure used as the web passes through the rollers during calendaring. The analysis of the resulting bonded, non-woven fabric by DSC provides valuable information on the morphology, or structure, achieved by the polymer during processing. EXPERIMENTAL In this study, the thermo-physical properties of two different nonwovens were characterized using differential scanning calorimetry. Two PET (polyethylene-terephthalate) copolymer nonwovens, labeled as T and U were analyzed. The

following experimental conditions were used: application note Instrument Pyris 6 DSC Sample mass Approx. 7 mg Sample pan Crimped aluminum pan Purge gas Nitrogen Temperature program Heat from 30 to 300 C at 10 C/min Cool back to 30 C at 20 C/min Heat from 30 to 300 C at 10 C/min PETech-57 RESULTS Displayed in Figure 1 are the DSC results generated on non-woven fabric sample T, in its as-received state (first heat). The DSC plot shows heat flow as a function of sample temperature and an endothermic response (heat

absorbed) is oriented upwards. The sample exhibits its main melting transition at 257.8 C with a heat of melting of 54.4 J/g. There is a small peak observed below the main melting transition at 189.9 C and this most likely reflects a thermal annealing or heat set processing step. When a polymer, like PET, is exposed to thermal conditioning after spinning or extrusion, this results in the formation of imperfect crystallites. These crystallites have a melting point significantly lower than that of the main melting temperature of the more stable crystalline component. This small, endothermic, pre-melt peak is useful for assessing the effective heat set conditions to which the polymer was exposed during production. Another small endothermic peak is observed at approximately 40 C for the as received sample, and this reflects the particular processing

conditions to which the non-woven material was exposed. Non-woven sample T was cooled back to 30 C and then reheated through the melt and the results of the second heating are displayed in Figure 2. When a polymer, that has been exposed to processing (via elevated temperatures and orientation), is melted and cooled, the original structure is destroyed. The polymer then receives a new thermal history and the DSC results obtained during the reheat will often be significantly different than those generated for the as received sample. As a result of its new thermal history, sample T now exhibits a glass transition event, Tg, at 73 C, as reflected by the stepwise change in heat flow that occurs in this temperature region. The main melting transition is now observed at 250 C with a heat of melting of 45.2 J/g. . The decreases in the peak temperature and heat

of melting (or crystallinity) during the second heat are reflective of the loss of crystalline perfection that the as received sample possessed. Figure 1. DSC results obtained on as received non-woven sample T Figure 2. DSC results obtained on sample T during second heating PETech-57 Displayed in Figure 3 is an overlay of the DSC results obtained on nonwoven sample T during the first and second heating segments. The large differences in the sample s thermal properties, due to processing, are exemplified in this figure. Non-woven sample U was analyzed in the same manner and the results of the first heating experiment (as received) are presented in Figure 4. Non-woven U yields its main melting peak at 255.1 C with a heat of melting of 63.9 J/g. Sample U has a larger heat of melting than T (63.9 versus 54.4 J/g) which indicates that U is more crystalline than T in the as received state.

Sample U yields a pre-melting peak shoulder at 213.2 C, which is most likely due to the heat set conditions. In addition, non-woven U shows two lower temperature transitions at 46.3 C and 82.2 C, indicative of the sample s particular processing. The transition at 82,2 C is most likely the Tg with an overlapping enthalpic relaxation peak. Sample U was cooled back to room temperature and then reheated at a rate of 10 C/min to eliminate the effects of processing on the sample s morphology. Displayed in Figure 5 are the DSC results obtained during the second heating segment for sample U. Sample U yields its main melting peak at 253.3 C with a heat of melting of 53.9 J/g, during the reheat experiment. The glass transition is now clearly observed at 82.3 C as a stepwise change in the sample s heat flow. Figure 3. Overlay of 1st and 2nd heating segments for sample T

Figure 4. DSC results obtained on non-woven U (as received) Figure 5. Results obtained on non-woven U (second heat) PETech-57 Shown in Figure 6 is a direct overlay of the first and second heating segments for sample U. The differences in the sample s thermal history, due to processing, are very apparent from these results. A comparison of the DSC results generated on the as received samples T and U is presented in Figure 7. The differences in the pre-melting peaks reflect clear differences in the processing histories of non-woven samples T and U. SUMMARY DSC is a valuable technique for characterizing the thermo-physical properties of polymeric non-woven fabrics. DSC provides useful information on the following main properties of non-wovens: Melting temperatures Heats of melting and percent

crystallinities Glass transition temperature (Tg) Heat set or processing temperatures PerkinElmer offers a variety of high performance DSC instruments for outstanding results on polymers and non-woven fabrics. In addition to DSC, thermogravimetric analysis (TGA), thermomechanical (TMA) and dynamic mechanical analysis (DMA) would provide valuable characterization information on the degradation, dimensional and mechanical properties of non-woven materials. REFERENCE

The formation of hydroentangled fabrics relies on the impact of high-speed water jets to bond fibres (or filaments) in a web to produce a coherent fabric structure with the required properties. It is known that the influences of water pressure and specific energy influence the tensile properties of hydroentangled fabrics, but little is known about how the structure of hydroentangled fabrics determines the fabric properties and how both processing parameters and fibre properties influence the formation of the fabric structure. The limited information about the fundamental mechanisms involved in hydroentanglement hinders the engineering design of functional fabric structures. Therefore, it is desirable to have structureprocessproperty relationships for hydroentangled fabrics that links hydroentanglement processing parameters, fabric structure and fabric properties (e.g., tensile strength and permeability) in order to engineer hydroentangled fabrics to meet the requirements of specific product applications.

In this paper, modeling of the basic hydroentangled fabric formation process is summarized based on a consideration of the fibre deformations induced by water jet impact. A relationship between the fabric permeability and the deflection depth of fibre segments in hydroentangled fabrics is introduced. Microstructural analysis of non-woven fabrics using scanning electron microscopy and image processing. Part 2: application to hydroentangled fabrics E Ghassemieh*, M Acar and H K Versteeg Mechanical and Manufacturing Engineering, Loughborough University, Leicestershire, UK Abstract: The image analysis techniques developed in Part 1 to study microstructural changes in nonwoven fabrics are applied to measure the bre orientation distribution and bre length distribution of hydroentangled fabrics. The results are supported by strength and modulus measurements using samples from the same fabrics. It is shown that the techniques developed can successfully be used to assess the degree of entanglement of hydroentangled fabrics regardless of their thickness. Keywords: non-woven, fast Fourier transform, Hough transform, microstructure, scanning electron microscope 1 INTRODUCTION The mechanical properties of non-woven fabrics and brous assemblies depend greatly on their microstructure. The arrangement of the bres in the assembly to a large extend determines the mechanical properties of the fabrics. Predictive

models of the mechanical behaviour of non-wovens in terms of their microstructure have been studied by various researchers. Komori et al. modelled the compressibility of brous assemblies using bre contact theory [1]. In further work they improved their model by introducing a modied bre contact theory [2]. The main structural parameters used in their study were the orientation function, bre crimp and elasticity of the bres. Patel and Warner have modelled the bending stiffness of point-bonded fabrics using the properties of the constituent bres and the fabrics microstructure [3]. In micromechanical modelling of brous assemblies, the microstructural parameters form the basis of the analysis. Therefore, it is of fundamental importance to be able to assess the microstructural parameters such as bre orientation distribution and curl of the bres for modelling purposes. In this study, image processing techniques have been developed to investigate the microstructural properties of fabrics made by the hydroentanglement process [4], and the mechanical properties of fabric samples, such as tensile strength and modulus, have been measured. 2 HYDROENTANGLEMENT Hydroentanglement is a purely mechanical bonding technology used for non-woven fabric production. It uses very ne high-speed jets of water that impact the web, penetrating

into its inner planes, resulting in reorientation and entanglement of the constituent bres. The impact of the high-speed water jets causes the bres and bre bundles to twist, bend and rotate around themselves and other bres to form a series of small, interlocking entanglements [5]. The process uses successive stages of entanglement either by deploying multiple water jet stations or feeding the fabric through the machine several times. Both sides of the fabric are normally processed, either by impacting both sides of the fabric by water jets in one pass or by ipping the fabric over in alternate passes, depending on the congurations of the machine used. Thus, the consolidated fabric is bonded by friction, resulting in a soft yet relatively strong fabric. One of the main factors that changes through the hydroentanglement process is the friction, which plays a predominant role in determining the mechanical behaviour of fabrics as reported by Hearle and Sultan [6]. The energy required at each progressive stage of the hydroentanglement process depends on the interbre friction and the entanglement sites that have already been imparted to the fabric in The MS was received on 20 November 2001 and was accepted after revision for publication on 16 April 2002. *Corresponding author:Wolfson School of Manufacturing and Mechanical Engineering, Loughborough University, Loughborough, Leicestershire,

LE11 3TU, UK. 211 L04201 # IMechE 2002 Proc Instn Mech Engrs Vol Downloaded from pil.sagepub.com at Bahir Dar University on November 6, 2011 216 Part L: J Materials: Design and Applications

the previous stage. Through this energy transfer, the microstructure of the brous assembly progressively changes and its mechanical properties are consequently improved. Web properties such as area density, bre arrangement and structure, determined by the method used in the web production (e.g. parallel or cross-laid), and the properties of the constituent bres such as modulus, bending stiffness, length and interbre friction are some of the factors that determine the amount of energy needed for a desirable entanglement [7]. The main microstructural changes that occur in the fabric through the process of hydroentanglement are manifested by: (a) a change in bre orientation distribution through rearrangement of the bres; (b) a change in the apparent length of the straight segments of bres through the curling and migration of bres into the inner planes of the brous assembly. The main microstructural parameters assessed here are therefore the bre orientation distribution and the distribution

of the straight segments of the bres in the fabric. These parameters can be applied to estimate the level of bre contact points and the friction between bres, which are important parameters in the study of the fabric microstructure. The bre orientation distribution can also be used as a measure of the anisotropy of the fabric. The modulus and strength of a non-woven fabric directly depend not only on the physical properties of the bres but also on the arrangement of the bres in the fabric, e.g. the bre orientation, the curl and the friction between the bres at the points of the contact. These mechanical properties can relate to the microstructural properties of the non-woven fabrics. 3 EFFECT OF PROCESSING ON THE BONDING STRENGTH AND MICROSTRUCTURE OF THE FIBRE ASSEMBLIES In this study, scanning electron micrographs and twodimensional Fourier and Hough transforms of the image have been used to estimate the bre orientation distribution, and these two methods have been compared. The distribution of the length of the straight bre segments has been found using the Hough transform (HT) of the fabric images. The mechanical properties of the fabrics (i.e. strength and modulus) have also been measured. The relationship between these microstructural variables and the mechanical

properties has been analysed. 3.1 Fibre orientation distribution using the fast Fourier transform (FFT) In this study, two different fabrics were used as supply webs. These webs were hydroentangled and tested to examine the effect of the hydroentanglement process on the bre orientation distribution using the FFT method. These two webs are: (a) web 1: 70:30 viscose polyethylene terephthalate (PET) bre blend, 120 g/m2 cross-laid web; (b) web 2: 70:30 viscose PET bre blend, 110 g/m2 parallel-laid web. The processing speed was 10 m/min. The pressure prole in different consecutive passes of the fabric through the hydroentanglement machine is given in Tables 1 and 2 respectively. The rst and the second sides of the fabric were processed through four and three different passes respectively. Numbers 1.1 to 1.4 refer to the fabric samples obtained after processing the rst side of the fabric, and numbers 2.1 to 2.3 correspond to the fabric samples obtained after processing the second side of the fabric. The rst test was performed on fabrics made of a 70:30 viscose PET bre blend, 120 g/m2 cross-laid web. This web was processed under the pressure prole reported in Table 1. Figure 1 shows scanning electron microscope (SEM)

images of the tested fabric after various stages of hydroentanglement. Figure 2 shows the tensile strength and modulus of the fabric samples, measured using the EDANA standard. Details of the experimental procedure have already been reported [7]. The SEM images of fabric samples (Fig. 1) after the rst side processing illustrates a much looser brous structure for fabric sample 1.1 than for fabric sample 1.3 (Figs 1a and b respectively). Inspection of fabric sample 2.1 (after the rst pass of the second side) reveals a tightly consolidated fabric structure already achieved at this stage. In contrast to the rst side samples, the difference between second side fabric structures 2.1 and 2.3 is relatively small, as shown in Figs 1c and d respectively. Successive processing of the fabric increases the number of entanglement points at a gradually diminishing rate owing to increasingly tightly packed brous structures resisting further bre movement. Therefore, the effect of processing of the rst side of the fabric on the entanglement and the fabric microstructure is much greater. This explains the higher rate of increase in the tensile strength and modulus for the fabric samples from the rst side processing compared with the second side processing (Figs 2a and b). In these gures, MD, CD and DD indicate machine direction, cross-direction and diagonal direction (45 to MD and DD) respectively.

In the example of a fabric made from cross-laid web, the large increase in the fabric strength in the cross-direction indicates an increase in the degree of entanglement, while Table 1 Processing conditions using the 120 g/m2 viscose PET cross-laid web (machine speed 10 m/min) First side pass number Second side pass number 1.1 1.2 1.3 1.4 2.1 2.2 2.3 60 bar 100 bar 130 bar 130 bar 110 bar 130 bar 130 bar Proc Instn Mech Engrs Vol 216 Part L: the increase of strength in the machine direction a shows change in bre orientation towards the machine direction. The relative number of bres in any direction generally correlates with the strength and modulus of the bres in that direction. The bre orientation distribution estimated using the Fourier transform (FT) for fabrics 1.1 and 1.3 are compared in Fig. 3a. A similar comparison is made for fabrics 2.1 and 2.3, as shown in Fig. 3b. The predominant orientation of the bres is in the cross-direction (90), as expected from a cross-laid web. An increase in bre distribution in the machine direction is observed throughout the entire entanglement process. However, the change in the orientation distribution from fabric 1.1 to fabric 1.3 is more signicant than the change from fabric 2.1 to fabric 2.3. In processing the rst side of the fabric, since there is less initial bonding between the bres, they can be reoriented with less effort.

As the hydroentanglement process successively moves from one stage to the next, the number of entanglement points between the bres increases progressively, hence at each stage rendering the movement and migration of bres more difcult. The second test was performed on fabrics made of a 70:30 viscose PET bre blend, 110 g/m2 parallel web. This web was processed under the pressure prole reported in Table 2. In this case, for the parallel-laid web, most of the bres are originally oriented in the machine direction, and therefore the modulus and strength in the machine direction are higher for all samples than those in the cross-direction. The strength and modulus of the machine direction increase signicantly from fabric sample 1.1 to 1.3 and from 2.1 to 2.3, as shown in Figs 4a and b, indicating gradual increase in the entanglement of the bres. Figure 4 shows a very slight increase both in the strength and modulus of the fabric in the cross-direction, indicating little change in the direction of bre orientation Table 2 Processing conditions using the 110 g/m2 viscose PET parallel-laid web (machine speed 10m/min) First side passes Second side passes 1.1 1.2 1.3 1.4 2.1 2.2 2.3 50 bar 100 bar 120 bar 120 bar 100 bar 120 bar 120 bar Fig. 1 SEM images of viscose PET cross-laid fabrics (Table 1)

L04201 # IMechE 2002 Proc Instn Mech Engrs Vol 216 Part L: J Materials: Design and Applications MICROSTRUCTURAL ANALYSIS OF NON-WOVEN FABRICS. PART 2 213 Downloaded from pil.sagepub.com at Bahir Dar University on November 6, 2011J Materials: Design and Applications L04201 # IMechE 2002The bre orientation distribution estimated for the fabric samples obtained after the rst and third passes of the rst side processing, 1.1 and 1.3 respectively, are compared in Fig. 5a. Figure 5b shows a similar comparison for the fabric samples obtained after the rst and third passes of the second side processing, 2.1 and 2.3 respectively. Since the main orientation of the bres forming the web is in the machine direction, the maxima of all the distribution curves occur at 0 (180), i.e. the machine direction. As indicated by the results for each fabric sample, the predominant orientation of the bres remains the machine direction, with little increase in the bre orientation in the cross-direction. There is a greater increase in the bre orientation in the crossdirection after the rst side processing than after the second side processing, as expected. The overall change in the bre orientation in the crosslaid fabric is greater than that of the parallel-laid fabric. This can be concluded by comparing the results of fabrics made of cross-laid web (Fig. 3) with those of fabrics made of parallel web (Fig. 5). This is due to the nature of the hydroentanglement process in that the impinging jets continuously move relative to the fabric in the machine

direction, dragging bres to migrate from the cross-direction to the machine direction with more ease. 3.2 Fibre orientation distribution using the Hough transform It has been shown that the bre orientation distribution can also be estimated by the HT method [4], as the orientation of the lines is one of the direct HT results. On the other hand, in the FFT method, the orientation of the line is derived by the parameter of the rate of change in grey scale of the original image. Therefore, the HT result might be expected to be more accurate. We have carried out a number of tests to compare FFT and HT results in evaluating the bre orientation distribution. The rst test used a fabric made of cross-laid viscose PET web (Table 1). The HT result is highly affected by the intervals chosen to pick up the local maximum and by the discretization of the Hough space. Figure 6a compares the Fig. 2 (a) Tensile strength and (b) tensile modulus of viscose PET cross-laid fabrics (Table 1): CD, cross direction; MD, machine direction; DD, diagonal direction Fig. 3 Fibre orientation distribution of viscose PET cross-laid fabrics (Table 1) (a) on the rst side and (b) on the second side, evaluated by the FFT method Proc Instn Mech Engrs Vol 216 Part L: J Materials: Design anorientation distribution using different numbers of cells (50, 100 and 200) in Hough space. The accuracy of the method

in estimating the bre orientation distribution increases with an increasing number of cells in Hough space. On the other hand, increasing the number of cells also increases the computational time. Therefore, there is a need for a compromise between accuracy and the demand on computation time. For the purpose of the present investigation, numerous tests have shown that a 100-cell Hough space is the optimum size cell discretization for acceptable accuracy and computing time. This is discussed in more detail in Part 1 of this paper [4]. The bre orientation distribution obtained by the FFT method and the HT method using 100-cell Hough space are compared in Fig. 6b. The good agreement between these results is evident. The second test made a similar comparison between the bre orientation distribution for a hydroentangled fabric produced from parallel-laid viscose PET web (Table 2) as estimated by the FFT and HT methods (Fig. 7). The normalized results show once again a reasonable agreement between the outcome of the two methods. The HT method was used to investigate the effect of rst side and second side processing on the bre orientation distribution. This test was performed on a viscose PET cross-laid web which was processed under the pressure proles shown in Table 3. The results of the HT estimation are shown in Figs 8a

and b. The solid lines in these gures are the curves tted to show the distribution function. The discrete points show the immediate outcome of the HT for orientation distribution. The main orientation of the bres is in the cross-direction, as expected from a cross-laid web. Processing the fabric through more passes caused fabric 1.3 to have a more uniform distribution compared with fabric 1.1. Figure 8b shows the bre orientation distribution estimated by the HT method for fabric samples after pass 2.1 and 2.3. The orientation distributions of these two fabric samples are much more similar. The bre orientation distribution estimated by the HT method, gives similar results to that estimated by the FFT method, presented earlier. In this nal test, a comparison of the FFT and HT techniques was made using a fabric sample with more curled bres. This fabric was made of a viscose PET cross-laid web, which was processed under the pressure prole reported in Table 4. Under this high pressure, much Fig. 4 (a) Tensile strength and (b) tensile modulus of viscose PET parallel fabrics (Table 2): CD, cross direction; MD, machine direction Fig. 5 Fibre orientation distribution of viscose PET parallel fabrics (Table 2) (a) on the rst side and (b) on the second side, evaluated by the FFT method L04201 # IMechE 2002 Proc Instn Mech Engrs Vol 216 Part L: J Materialsd Applications L04201 # IMechE

2002The FFT method does not have this restriction. Because of this property, the outcome of the HT method is only comparable with the FFT method when it is applied to skeletonized images. Figure 9 compares the bre orientation distributions evaluated by the HT method using a skeletonized image and by the FFT method using both skeletonized and binarized images. While the FFT results using skeletonized images are in reasonable agreement with the HT results, the FFT estimation of the binarized images is considerably different. This can be attributed to the fact that the curl of the bres causes the skeletonized image to be considerably different to the binarized images. This shows that, as the curl of the bre in a fabric increases, the accuracy of the HT method in estimating bre orientation distribution reduces. Fig. 6 Effect of Hough space discretization on the bre orientation distribution evaluated for viscose PET cross-laid fabric (Table 1) and (b) comparison of the FFT and HT results for the bre orientation distribution of the fabric Fig. 7 Comparison of the FFT and HT results for the bre orientation distribution of viscose PET parallel fabric (Table 2) Table 3 Processing conditions using the 120 g/m2 viscose PET

cross-laid web (machine speed 10 m/min) First side passes Second side passes 1.1 1.2 1.3 1.4 2.1 2.2 2.3 50 bar 70 bar 100 bar 100 bar 80 bar 100 bar 100 bar Fig. 8 Fibre orientation distribution of viscose PET cross-laid fabric (Table 3) (a) on the rst side and (b) on the second side, evaluated by the HT method Proc Instn Mech Engrs Vol 216 Part L: J Materials: Design and Applications L04201 # IMechE 2002 length index is used, as the values on the horizontal axis do not show the exact length of the bres but are proportional to the length of the bres. Figure 10a shows the bre length distribution in fabric samples 1.1, 1.2 and 1.3, all rst side processing. As the length distribution for these bres veries, the length of the straight segments of the bres decreases from sample 1.1 to sample 1.2 to sample 1.3, as indicated by the bre length indices. The results of the second side for fabrics 2.1 and 2.3 are presented in Fig. 10b. A similar trend of change is also observed, as the length of the straight segments of the bres decreases from fabric sample 2.1 to sample 2.3. It should be noted that the actual length of the bres does not change through the process. However, as the entanglement process proceeds, bres are moved from one plane to another, and at the same time they bend and curl. This results in a reduction in the length of the straight segments

of bres, as detected by HT analysis. Therefore, the most frequent bre length in each distribution curve correlates with the degree of bonding achieved through entanglement. The lower the most frequent bre length, the higher is the level of entanglement. The bre length distribution can therefore be used as a measure of the bre entanglement in the hydroentangling process. 4 CONCLUSIONS The image processing techniques developed in Part 1 using FFT and HT methods in conjunction with SEM images have successfully been used in evaluating microstructural changes in hydroentangled fabrics, such as bre orientation and bre length distribution. Mechanical fabric properties such as strength and modulus also support the validity of the above techniques, which can conveniently be used to estimate the degree of entanglement in hydroentangled fabrics. The results of the FFT and HT methods have been compared. It has been shown that the results of the skeletonized images by the FFT and HT techniques are in very good agreement. The outcome of the HT analysis is highly affected by the Hough space discretization. This needs to be optimized for a given fabric sample. The image processing techniques developed have been successfully tested for fabrics up to 120 g/m2. This shows the applicability of the methods for heavy and thick fabrics.

In cases where the thickness of the fabric is higher than the depth of eld of the SEM, a two-dimensional analysis would be less accurate. In these cases, a three-dimensional analysis including the depth of the fabric is required.