Structural Analysis by X-Ray Microtomography of a Strained Nonwoven Papermaker Felt

  • Published on

  • View

  • Download


  • Research Journal online version of this article can be found at:

    DOI: 10.1177/004051750207200603

    2002 72: 480Textile Research JournalX. Thibault and J.-F. Bloch

    Structural Analysis by X-Ray Microtomography of a Strained Nonwoven Papermaker Felt

    Published by:

    can be found at:Textile Research JournalAdditional services and information for Alerts:

    What is This?

    - Jun 1, 2002Version of Record >>

    at Gebze Yuksek Teknoloji Enstitu on April 27, 2014trj.sagepub.comDownloaded from at Gebze Yuksek Teknoloji Enstitu on April 27, 2014trj.sagepub.comDownloaded from

  • 480

    9. Hearne, E. R., and Nossar, M. S., Behavior of LooseFibrous Beds During Centrifuging, Part I Textile Res. J. 52,609-614 (1982).

    10. Katsube, N., The Constitutive Theory for a Fluid Filled PorousMaterial J. Appl. Mechan. ASME 84-WA/APM-48 (1984).

    11. Pan, N., and Camaby, G. A., Theory of the Shear Deformation ofFibrous Assemblies, Textile Res. J. 59, 285-292 (1989).

    12. Sebastyen, E., and Hickie, T. S., The Effect of CertainFiber Parameters on the Compressibility of Wool, J. Tex-tile Inst. 62, 545-560 (1971).

    13. Shinohara, K., Tamara, K., Gotoh, K., and Tanaka, K.,

    Kagaku Kogaku 31, 287 (1967).

    14. Skempton, A. W., The Pore Pressure Coefficients A and B,Geotechnique 4, 143-147 (1954).

    15. Van Wyk, C. M., Note on the Compressibility of Wool, J.Textile Inst. 37, T285-T292 (1946).

    16. Verruijt, A., The Theory of Consolidation, in "Fundamen-tals of Transport Phenomena in Porous Media," J. Bear andM. Yavuz Corapcioglu, Eds., NATO ASI Series, Series F,Applied Science No. 82, 1984.

    17. Virto, L., and Naik, A., Frictional Behavior of Textile

    Fabrics, Part I, Textile Res. J. 67(11), 793-802 (1997).

    Manuscript received August 6, 2001; accepted November 16. 2001. ,

    Structural Analysis by X-Ray Microtomography of a StrainedNonwoven Papermaker Felt


    Paper Physics Department, Ecole Franaise de Papeterie et des Industries Graphiques,38402 Saint-Martin dHres, France


    Textiles are used in many industrial applications, especially in papermaking. During thepressing of paper, felts are used to recover the water expressed from the wet sheet. Toimprove this operation, felt structures have to be characterized. X-ray microtomographyis a powerful tool for reconstructing the complex organization of such fibrous materials.This method is first described, then examples of measurements are shown and discussed.This nondestructive technique appears to be an excellent tool for investigating woven andnonwoven structures. The fine description of the structure permits the characterization ofstructural parameters and numerical modeling of physical phenomena of this true three-dimensional structure.

    Textiles are used in many industries for different ap-

    plications. The technique we present here may be used inalmost any case, but we will focus our attention in this

    paper on one particular sector: papermaking. In particu-lar, the experimental results will illustrate the ability ofour set-up for felts used in the pressing section.

    Paper pulp contains only roughly 5% mass fiber and 95%mass water. Thus the main aim of the whole process is

    essentially to remove the water from the wet fibrous pad toobtain a paper sheet. At the beginning of paper making, thepulp is laid on a fabric, and the suspension is concentratedby filtration. The drainage is improved by depressor ele-ments such as vacuum boxes or foils in contact with the

    fabric. Next, the sheet supported by the felt is pressed

    between rolls. Finally, the wet sheet is dried by heating therolls to remove the excess water and to reach a siccity (ratioof dry and humid masses) of 0.95. The energy necessary toremove the same quantity of water by drying is six times

    higher than by pressing. Hence, optimization of the pressingsection impacts the entire paper mill economy.

    Different means exist to improve pressing efficiency. Inparticular, one way is to improve sheet consolidation byoptimizing the structure of the felt. Often in pressing stud-ies, the flow resistance due to the felt structure is eitherrelated to physical parameters following an empirical law ornot taken into account at all. In order to justify such laws,experimental work must consider felt flow resistance. Ex-perimentally, when the porosity goes from 0.65 to 0.30

    at Gebze Yuksek Teknoloji Enstitu on April 27, 2014trj.sagepub.comDownloaded from

  • 481

    [12], the resulting permeability may run from 10-10 to10- 12 2 m2 following an exponential relationship with theporosity. However, recent works tend to show that feltpermeability in modem pressing techniques is not a majorfactor for the efficiency of unit operation and may beneglected [6]. But the structure of the felt remains at theforefront of pressing physics because of its water stockcapacity, the problems of rewetting, or phenomena relatedto the uniformity of pressure application. A 3D reconstruc-tion of felt has never been published to our knowledge.Some works have focused on fabric or scrim reconstruction

    [ 13, 10, 8] but never on a whole felt because of the difficultyof accounting for fiber needling.

    Characterization of fibrous structures, and felt in partic-ular, has always been within the scope of many scientificworks [1, 2, 7, 9]. For example, confocal laser scanningmicroscopy has been used to obtain a three-dimensionalstructure [4]. Indeed, it is very well known that the pressingfelts structure has a huge influence on its physical proper-ties. Therefore, many experimental techniques such as mi-croscopy have been used to visualize the complex organi-zation of the fibers inside porous networks. However, it hasbeen experimentally very difficult to obtain the whole struc-ture with an acceptable accuracy, that is to say, less than afew microns on a large sample. Furthermore, the sampleshave to be carefully prepared, and the structure may bedamaged.

    In this work, we construct a 3D characterization of thefelt structure using a local x-ray microtomography tech-nique. We first briefly describe this tool and then presentsome examples to illustrate our experiments.


    The technique used here is called microtomography. Itdiffers from x-ray computed tomography only because theyhave different spatial resolutions. X-ray absorption com-puted tomography is widely used for medical imaging andis accomplished by reconstructing a 2D or 3D image of theobject from attenuation measurements at different angularpositions. This technique is greatly improved by the use ofsynchrotron radiation. Unlike laboratory x-ray sources, syn-chrotron radiation offers the possibility of selecting x-rayswith a small energy bandwidth from the wide, continuous

    energy spectrum while retaining a sufficiently intense beam.It allows high spatial resolution images to be generated.Moreover, hardening artefacts, which often occur withpolyenergetic beams in classical tomographic imaging, areattenuated. Synchrotron x-ray computed microtomography,which provides high-resolution 3D images, is well suitedfor studying the porous structure of felt.The fine description of the structure is the main aim of

    this study. The computed sample diameter is 6.7 mm,

    and the images are recorded with a 6.6 micron pixel size.Another possible technique involves x-ray phase contrasttil, 51. but we did not use this technique here becausetransmission gives excellent results in our experiments.Furthermore, this technique is much faster than phasecontrast. In classical scanners, the x-ray source and thedetector rotate around a fixed sample. This is obviouslynot possible with a synchrotron radiation source. There-fore, the rotation is applied to the sample itself (seeFigure 1 ). The sample to be analyzed is mounted on atranslation/rotation stage, allowing precise alignmentwith the beam. A 2D detector records the beam attenu-ation produced by the sample for different angular posi-tions, as shown in Figure 1. A typical scan involves 900projections of the sample over 180. Different lenses andscintillators may be used to adjust the pixel size to thesample size. Moreover, we have used a monochromatorin the experimental set-up. The detector is of primaryimportance in the set-up, since it determines the spatialresolution of the image. The detector we used is based onthe Frelon CCD (charge couple device) camera developedby the ESRF Detector Group. There are two cco sizes-1024 by 1024 elements and 2048 by 2048 elements. Athin scintillation layer deposited on glass converts x-raysto visible light. Light optics (associated with each cam-era) magnify the image on a screen and project it onto theCCD. The CCD camera is mounted perpendicular to thex-ray beam in order both to protect it and to avoid directinteractions that cause noise in the recorded images.

    FIGURE 1. Schematic view of the experimental set-up.

    A 3D back projection algorithm is then used to rooonsduaa 3D image of the sample from the series of 2D projections.A volume classically has a size of 3.5 gigabytes, and thecomputing time per sample is two hours using parallelizedcomputing. The main idea of the reconstruction is based onthe Beer-Lambert law (or attenuation law),

    where A(x, y) represents the value of the linear attenu-ation coefficient at the point ( x, y). Measuring the num-

    at Gebze Yuksek Teknoloji Enstitu on April 27, 2014trj.sagepub.comDownloaded from

  • 482

    ber No of photons emitted by the source and the numberN, of photons transmitted throughout a single line acrossthe sample allows us to calculate the integral of tL alongthe considered path.

    During the experiments, the sample thickness is about3 mm and the image is 6.67 mm wide. Repeating such ameasurement along a sufficient number of straight lineswithin the same slice delivers the Radon transform of the

    object. The inverse of the Radon transform leads to thewhole map of the attenuation coefficient A. Because

    attenuation values of the component media are well

    separated, reconstruction of the structure is possible (seeFigure 2). Other details of the x-ray microtomographytechnique are presented by Coles [3].

    FIGURE 2. Example of binarization: the left picture used 256 graylevels, the right picture only 2 (black and white).


    In order to illustrate the x-ray computed microtomog-raphy technique, we present 3D reconstructions of felt.The felt we tested was a batt-on-mesh type with a singlescrim and a total basis weight of 1240 g - m - 2. Attenu-ation measurements are shown for a single sample forthree different levels of deformation. After the recon-

    struction, the attenuation map is usually scaled on 2;2gray levels. In our case, that range of attenuation mea-

    surements is too large for our two-phase media. There-fore, we have rescaled the image and use only 2g graylevels. Finally, relative to the gray level distribution ofthe picture, the structure is binarized (Figure 2). How-ever, because the felt sample may have slipped in rota-tion inside the cell, its angular position may have to bemodified for successive levels of deformation. This en-

    ables direct computation of local strain using twostrained, reconstructed volumes.The reconstructed binarized volume is presented in

    order to illustrate the ability of the microtomographytechnique. The first picture illustrates the felt samplewith no applied strain, so the fibers of the upper surfacemay be isolated. The path of the needles through the

    thickness is brought to the fore (Figure 3). The size of thevolume is 3.84 mm wide and 3.16 mm high. The feltscrim is made of monofilament yarns in the warp direc-

    tion and multifilament yarns in the filling direction.

    FIGURE 3. Felt with no strain,.

    Figure 4 shows the structure at the second level ofstrain imposed on the felt. The size of the volume is 3.84mm wide and 1.54 mm high, instead of 3.16 mm ini-tially. The region of interest is chosen inside the com-pressing surfaces at a distance of 3 pixels, representingroughly 20 microns from them. The batt is dramaticallycompressed and the scrim seems lightly strained. In orderto study the influence of compression on the structure.

    FIGURE 4. Felt reconstruction under the second level of strain.

    at Gebze Yuksek Teknoloji Enstitu on April 27, 2014trj.sagepub.comDownloaded from

  • 483

    FIGURE 5. Porosity versus thick-ness position for each level of strain:no strain.

    the porosity is plotted versus the position in the feltthickness (Figures 5 and 6). The porosity for each hori-zontal section is computed by taking into account thenumber of black pixels that represent a void space. Theposition reference is set to the bottom fixed surface. Foreach level of strain, the porosity is higher in the batt thanin the scrim. The total porosity is calculated as theaverage of the porosity values for each section. Thecorresponding values are obtained by increasing thestrain-0.63, 0.51, and 0.37.

    Due to the high compressibility of the batt comparedto the scrim, the variations in porosity are greater in thebatt than in the scrim. When there is no strain applied tothe felt, the porosity ranges from 0.31 to 0.99. This highporosity is the consequence of the absence of contact onthe upper surface. Different cross sections of both the

    scrim and the batt are presented hereafter. The study of.on one hand, the porosity curves. and on the other hand,these pictures allows us to distinguish roughly two sep-arate mechanisms that occur during compression: first.only the batt is strained, and second, the batt is verycompressed and the yams of the scrim bend.The porosity is higher in the batt for each strained

    state. The biggest local void area is created in the scrimby the woven structure. This can be verified in Figures 7.8, and 9. Figures 8 and 9 illustrate the bending of thewarp monofilament yams inside the scrim for twostrained states. The difference between the two pictures,considering the yam strain, is very small compared to thebatt compression.

    In order not to have to account for the compressiveset-up, we chose the region of interest for the recon-

    FIGURE 6. Porosity versus thick-ness position for each level of strain:second strain level.

    at Gebze Yuksek Teknoloji Enstitu on April 27, 2014trj.sagepub.comDownloaded from

  • 484

    FIGURE 7. Different cross sections for each strain state.

    struction to be smaller than the size of the compressivecell. We also wanted to compare the size of the rebuilt

    volume and the felt thickness. Therefore, we firstestimated the thickness of the volume according to thepixel size. Second, assuming the computed porosityfor the volume to be representative of the whole feltsample, we obtained the thickness value from the basisweight and the computed porosity (Equation 2). Theresults are summarized in Table I. Both values are

    close except in the case of the highest strain, becausethe felt is slightly inclined from the x-ray beam direc-

    FIGURE 8. Felt section for the first level of strain.

    FIGURE 9. Felt section for the second level of strain.

    tion. This may be explained by contact between thecompression set-up and the rotation set-up, which is

    perhaps not perfectly plane:

    where e is the thickness of the felt, W is the basis weight( 1240 g - m-2), s is the porosity of the felt, and p is thedensity of nylon ( 1.13 kg - m-;).

    TABLE I. Comparison of thickness values.

    The size of the rebuilt sample is often too small tobe representative of the media. However, the size ofthe representative element volume depends in our casemostly on the heterogeneous batt, i.e., on the order of10 cm2 by the felt thickness. Therefore, we have todetermine if the physical properties that describe thestructure are intrinsic to the material.


    Synchrotron x-ray microtomography is an excellenttool for investigating felt structures. Using an adaptedset-up, we have analyzed a 3.8 mm sample with 6.6micron accuracy. The maximum dimension of the

    acquisition set-up here is governed by the chosenaccuracy. This technique may be used to study differ-ent felt structures without any special preparationother than carefully cutting the sample. Note also thatit is a nondestructive technique, considering that the

    at Gebze Yuksek Teknoloji Enstitu on April 27, 2014trj.sagepub.comDownloaded from

  • 485

    reconstructed structure here is only the center part ofthe 10-mm-diameter original sample. Although com-plex, synchrotron x-ray microtomography providesnew opportunities for fine descriptions of porousstructures. The next step is to deduce from the struc-tural description the different physical properties suchas, for example, the hydraulic diameter or the specificsurface. Other computations of physical phenomena,such as flow through the obtained structure, will bemade in the future.


    We would like to thank E. Boller and J. Baruchel fromESRF for their help during the experimental work.

    Literature Cited

    1. Ballard, J., Press Felt Characterisation, Tappi J. 70(7)57-61, ( 1986).

    2. Chevalier, P., Contribution loptimisation des feutres depresses humides de papeterie, Doctoral thesis, Grenoble,INPG, 1988.

    3. Coles, M. E., X-ray Imaging, in "Experimental Methods inthe Physical Sciences," vol. 35, P.-Z. Wong, Ed., Aca-demic Press, NY, pp. 301-336.

    4. Goel, A., Tzanakakis, M., Huang, S., Ramaswamy, S.,Choi, D., and Ramarao, B. V., Confocal Laser ScanningMicroscopy to Vizualize and Characterize the Structure of

    Paper, Fundamentals and Numerical Modeling of UnitOperations in the Forest Products Industries, AIChE Symp.Ser. 324, 96, 75-80 (2000).

    5. Gureyev, T. E., Evans, R., Stevenson, A. W., Gao, D.,Wilkins, S. W., X-ray Phase-Contrast Microscopy of Pa-per, Tappi J. 84(2) (2001).

    6. IAnson, S., and Ashworth, T., Differentiated PermeabilitySurface Layer Pressing Theory, Tappi J. 83(11) (2000).

    7. Kershaw, T. W., The Three Dimensions of Water Flow inPress Felts, Tappi J. 55(6). 880-887 (1972).

    8. Lu, W. M., Tung, K.-L., and Hwang, K.-J., Fluid FlowThrough Basic Weaves of Monofilament Filter Cloth, Tex-tile Res. J. 66(5), 311-323 (1996).

    9. Macklem, J.-E., A Study of the Resistance of Woven WoolFelts to Liquid Flow, Tappi J. 44(8), 535-544 (1961).

    10. Peirce, F. T., Geometrical Principles Applicable to theDesign of Functional Fabrics, Textile Res. J. 18(3), 123-147 (1947).

    11. Samuelsen, E. J., Houen, P. J., Gregersen, O. W., Helle, T.,Raven, C., Three-dimensional Imaging of Paper by Use ofSynchrotron X-ray Microtomography, in "Proc. Tappi Int.Paper Conf.," San Diego, 1999, pp. 307-312.

    12. Thibault, X., Bloch, J.-F., Serra-Tosio, J.-M., Chave, Y..Permeability of Strained Felt: Application to PapennakingProcess, in "Proc. 8th World Filtration Congress," Brigh-ton, vol. 2, 2000, pp. 1011-1014.

    13. Toney, M. M., Computer Modeling of Fibrous Swctures,Part 3. J. Textile Inst. 91(3), 133-139 (2000).

    Manuscript recei>wd June 25. 2001: accepted August 19. 2001.

    at Gebze Yuksek Teknoloji Enstitu on April 27, 2014trj.sagepub.comDownloaded from