UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Experimental studies on creping and its influence on mechanical properties of tissue paper products Das, Ratul 2019

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


24-ubc_2019_february_das_ratul.pdf [ 22.85MB ]
JSON: 24-1.0376053.json
JSON-LD: 24-1.0376053-ld.json
RDF/XML (Pretty): 24-1.0376053-rdf.xml
RDF/JSON: 24-1.0376053-rdf.json
Turtle: 24-1.0376053-turtle.txt
N-Triples: 24-1.0376053-rdf-ntriples.txt
Original Record: 24-1.0376053-source.json
Full Text

Full Text

Experimental Studies on Creping and Its Influence onMechanical Properties of Tissue Paper ProductsbyRatul DasB.Tech., Indian Institute of Technology Roorkee, 2016A THESIS SUBMITTED IN PARTIAL FULFILMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMaster of Applied ScienceinThe Faculty of Graduate and Postdoctoral Studies(Mechanical Engineering)The University of British Columbia(Vancouver)January 2019c© Ratul Das 2019The following individuals certify that they have read, and recommend to the Faculty ofGraduate and Postdoctoral Studies for acceptance, a thesis/dissertation entitled:Experimental Studies on Creping and Its Influence on Mechanical Properties of Tissue PaperProductssubmitted by Ratul Das in partial fulfillment of the requirements forthe degree of Master of Applied Sciencein Mechanical EngineeringExamining Committee:Dr. A. Srikantha PhaniCo-supervisorDr. Sheldon GreenCo-supervisorDr. Gary S. SchajerSupervisory Committee MemberDr. Mauricio PongaSupervisory Committee MemberiiAbstractTissue papers are softer, stretchier, and more water absorbent than regular writing or packagingpaper products. Creping of an adhesively bonded low-density paper web from the surface ofa rotating Yankee drum is the key manufacturing technique in tissue production. Creping de-densifies and weakens the paper by partially damaging the fibers and the inter-fiber bonds inthe fiber network. The process also imparts a signature microstructure, called crepe structure,to the tissue paper. Tissues thus produced have a high specific volume (bulk to basis weightratio), work to rupture, failure strain (stretch), softness and absorbency. Mechanical propertiesof tissues are governed by the creping process. Therefore, a scientific understanding of thecreping process, and its impact on the structural and mechanical properties of the tissue paper isimportant.The present research approaches the highly complex problem from an experimental perspec-tive, with a view to complement ongoing physics based numerical models to simulate creping.Experimental techniques are developed to visualize the high speed creping process, quantify thecrepe structure, and finally understand the influence of the crepe structure on the uni-axial ten-sile response of the tissue. A novel surface imaging based structural quantification technique isdeveloped and successfully demonstrated on a commercial tissue machine. The surface imagebased quantification technique is also validated by micrographic observations of the tissue crosssection under a Scanning Electron Microscope (SEM). This work lays the foundational tech-niques and protocols for future studies in the laboratory and opens the opportunity to observecrepe structure in real time for quality and process control.The surface imaging techniques are then used to observe the evolution of the creping micro-structure under a tensile load. Local two dimensional strain fields are quantified using DigitalImage Correlation (DIC) to gain insight into failure mechanisms at the macroscopic networklevel. Micro tensile tests are conducted under SEM to gain further insight into the deformationand failure mechanisms operative at fiber length scales. The studies showed the impact of thecreping structure, formation, and inter fiber bonds on the tensile response of the tissue paper,specifically along machine direction.iiiLay SummaryTissue paper is a complex material. During tissue manufacture, pulp is first dried on a chemical-coated rotating drum until it is 95 percent dry. It is then pushed off at very high speed by a sharpcreping blade, creating hundreds of microscopic folds that give tissue its softness, flexibility,tearing resistance and strength. To scientifically understand how the creping folds affect a tissuepaper is a top concern for the paper industry. This research aims to quantify the creping foldsand relate them to the tissue properties. A novel surface image based creping fold measurementsystem is developed. This enables process monitoring on a tissue machine. The visualizationtechnique is then used to observe and analyze the change in tissue micro-structure under load.This research offers insight into the quantification of creping structure and its influence on me-chanical properties of tissue papers.ivPrefaceParts of Chapter 2 and Chapter 4 of this thesis have been included in a research paper [1] :”An Elastoplastic Creping Model for Tissue Manufacturing”. The paper is under review forpublication in International Journal of Solids and Structures.In Chapter 2, the Line Laser and surface imaging setup are designed by the author in theDynamics and Applied Mechanics Lab at The University of British Columbia. Sample prepa-ration for Scanning Electron Microscope imaging is done at Materials Engineering Departmentand Bioimaging Facility at the University of British Columbia by the author, under supervisionof Mr. Jacob Kabel and Mr. Derrick Horne respectively.In Chapter 3, both high speed visualization of creping mechanism and high speed imagingof tissue surface are done by author with help from Mr. Kui Pan, Ph.D. candidate from the samelab as the author.In Chapter 4, tensile characterization of tissue materials are done by author with help fromMr. Kui Pan and Mr. Charly Jeunot, a former summer intern. The microtensile tests with insitu SEM imaging are done by author under supervision of James Drummond. Both the tensilecharacterization and microtensile test experiments were done at FP Innovation, Vancouver.vContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiLay Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vContents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viList of tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiiList of figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ixGlossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Tissue Production and Creping . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Mechanism of Creping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.3 Crepe Structure Quantification . . . . . . . . . . . . . . . . . . . . . . . . . . 71.4 Tensile Properties of Paper . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91.5 Objectives and Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 Tissue Structure Quantification . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132.2 Through-thickness Quantification . . . . . . . . . . . . . . . . . . . . . . . . 152.2.1 Cross section Imaging Methods . . . . . . . . . . . . . . . . . . . . . 152.2.2 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162.2.3 Result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182.3 Surface Image Quantification . . . . . . . . . . . . . . . . . . . . . . . . . . . 222.3.1 Surface Imaging Setup . . . . . . . . . . . . . . . . . . . . . . . . . . 222.3.2 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232.3.3 Result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26vi2.4 Discussion and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303 Visualization and Online Measurement in Creping . . . . . . . . . . . . . . . . . 323.1 Visualization of Creping Mechanism . . . . . . . . . . . . . . . . . . . . . . . 323.1.1 Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323.1.2 Creping Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . 343.2 Online Crepe Structure Quantification . . . . . . . . . . . . . . . . . . . . . . 353.2.1 Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353.2.2 Result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 374 Tensile Response of a Creped Tissue . . . . . . . . . . . . . . . . . . . . . . . . . 384.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 384.2 Tensile Characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 384.2.1 Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . 384.2.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 404.3 Visualization of Fiber Network Deformation . . . . . . . . . . . . . . . . . . . 434.3.1 Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434.3.2 Results and Observations . . . . . . . . . . . . . . . . . . . . . . . . . 434.4 Structural Evolution with Load . . . . . . . . . . . . . . . . . . . . . . . . . . 494.4.1 Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494.4.2 Evolution of Crepe Structure and Local Strain Field . . . . . . . . . . . 494.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 555 Conclusions and Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . 565.1 Conclusions and Major Findings . . . . . . . . . . . . . . . . . . . . . . . . . 565.2 Limitations and Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . 57Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58A Microtensile Stage Extension Rate Calibration . . . . . . . . . . . . . . . . . . . 61viiList of tables2.1 Comparison of crepe structure quantification methods . . . . . . . . . . . . . . 132.2 Details of Grade 3-6. All the measurements were done on a 2 ply creped tissue 142.3 Details of an image obtained from line laser and from SEM technique . . . . . 162.4 Mean dominant wavelengths and crepe counts of Grades 1-6 based on throughthickness imaging. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182.5 Mean dominant wavelengths and crepe counts of Grades 1-6 based on surfaceimaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264.1 Analysis results of the machine direction load extension and stress strain plotsshown in Fig. 4.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 404.2 Analysis results of the cross direction load extension and stress strain plotsshown in Fig. 4.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414.3 Important parameters of Digital Image Correlation analysis . . . . . . . . . . . 54viiiList of figures1.1 Schematic of tissue manufacturing process and impactful parameters (1-9) . . . 31.2 Creping structure as seen from a (a) surface image and a (b) cross-section imageof a high basis weight creped tissue. . . . . . . . . . . . . . . . . . . . . . . . 41.3 Definitions of various angles related to creping mechanism. . . . . . . . . . . . 51.4 Schematic of the mechanism of creping. . . . . . . . . . . . . . . . . . . . . . 61.5 Stress strain behaviour of paper before and after creping. . . . . . . . . . . . . 102.1 Tissue cross section imaging techniques. . . . . . . . . . . . . . . . . . . . . . 152.2 Typical tissue cross section images and analysis methodology. . . . . . . . . . 172.3 Crepe structure quantification of Grade 1 and 2 based on cross section images. . 192.4 Crepe structure quantification of Grade 3 and 4 based on cross section images. . 202.5 Crepe structure quantification of Grade 5 and 6 based on cross section images. . 212.6 Tissue surface imaging setup . . . . . . . . . . . . . . . . . . . . . . . . . . . 232.7 A typical creped tissue surface image and quantification methodology. . . . . . 242.8 Typical surface images and crepe frequency histograms for Grade 1 and Grade 2 272.9 Typical surface images and crepe frequency histograms for Grade 3 and Grade 4 282.10 Typical surface images and crepe frequency histograms for Grade 5 and Grade 6 292.11 Through thickness and surface imaging crepe count comparison of Grade 1-6 . 303.1 Creping rig with high speed cross-section imaging setup . . . . . . . . . . . . 333.2 Four stages of creping fold formation in the high speed creping rig . . . . . . . 343.3 Online crepe count quantification setup in a real tissue machine (left). Compo-nents of the high speed imaging setup (right). . . . . . . . . . . . . . . . . . . 363.4 Variation of crepe count with time in a running tissue machine. . . . . . . . . . 374.1 (a) Experimental set up for tensile test of tissue papers. (b) Dimension of testedtissue samples. Tests are done according to ISO: 12625-4 protocol [2] . . . . . 394.2 Load extension (a) and corresponding stress strain plots (b) for multiple samplesof Grade 3-6 along machine direction . . . . . . . . . . . . . . . . . . . . . . 404.3 Load extension (a) and corresponding stress strain plots (b) for multiple samplesof Grade 3-6 along cross direction . . . . . . . . . . . . . . . . . . . . . . . . 41ix4.4 Microtensile stage (DEBEN Microtest 200N Tensile Tester; Serial number :MT10129) for tensile test of tissue paper outside SEM chamber . . . . . . . . 434.5 Microscopic deformation of tissue fiber network under tensile load for Grade 3 444.6 Microscopic deformation of tissue fiber network under tensile load for Grade 4 454.7 Microscopic deformation of tissue fiber network under tensile load for Grade 5 464.8 Microscopic deformation of tissue fiber network under tensile load for Grade 6 474.9 Evolution of crepe structure and local principal strain field with tensile loadalong machine direction for Grade 3 . . . . . . . . . . . . . . . . . . . . . . . 504.10 Evolution of crepe structure and local principal strain field with tensile loadalong machine direction for Grade 4 . . . . . . . . . . . . . . . . . . . . . . . 514.11 Evolution of crepe structure and local principal strain field with tensile loadalong machine direction for Grade 5 . . . . . . . . . . . . . . . . . . . . . . . 524.12 Evolution of crepe structure and local principal strain field with tensile loadalong machine direction for Grade 6 . . . . . . . . . . . . . . . . . . . . . . . 53A.1 Microtensile stage extension rate calibration methodology. . . . . . . . . . . . 61xGlossaryApparent density The mass of tissue material occupying unit apparent volume. This volumeincludes volume of fibers, pores and voids.Basis weight Areal density of a paper or fabric product, that is, its mass per unit of area.Bulk Inverse of density of paper products.Bulk softness A perception of easily crumpling a structure lacking firmness.Caliper Measure of thickness of paper products.Felt Woven belt of wool, cotton or synthetic fibres used to transport web of paper.Formation The manner in which paper fibres are mixed in a sheet of paper. Physical distri-bution and orientation of fibres and other solid constituents in the structure of a sheet ofpaper that affects its appearance and other physical properties.Forming fabric Finely woven fabric fitted at the wet end of the tissue machine to support andprovide drainage for the pulp suspension as it becomes a paper web.Hardwood Pulpwood from broad leafed dicotyledonous deciduous trees, such as aspen, mapleetc.Softwood Wood obtained from evergreen, cone bearing species of trees, such as pines, spruces,hemlocks, etc., which are characterized by having needles.Stock Various pulps, dyes, additives and other chemicals blended together in liquid form forpaper-making.Stretch The amount of strain a paper sheet can undergo before breaking.Wire mark The impression made on the bottom side of a paper web by the surface contour offabric pattern.xiAcknowledgementsFirst, I would like to thank my research supervisors: Dr. Anasavarapu Srikantha Phani andDr. Sheldon Green for their time, patience, and faith in me. I cannot emphasize enough howmuch their guidance and advice oriented me to be a critical thinker, as a researcher should be.Srikanth’s sincere attention to every step of my research and Sheldon’s innovative ideas andperspectives are the stepping stones of this thesis.I would like to thank Anna Karin, Ashleigh Ward, Daniel Ricard, Ho Fan, James Drum-mond, Jimmy Jong, Timothy Patterson, and Xuejun Zou for their support and for sharing valu-able industrial knowledge and experience. Guidance, training, and assistance provided by Der-rick (SEM specialiast, UBC Bioimaging Facility) and Jacob (Former Electron Microscopist,UBC Materials Engineering) are sincerely appreciated.I am thankful to the members of my examination committee: Dr. Gary S. Schajer and Dr.Mauricio Ponga, for generously offering time to evaluate and constructively criticize my work.My lab mates, Manav, Kui, Masoud, Reza and Aashish are the most sincere researchersI have come across. They create an academic yet enjoyable environment in my lab, whichconstantly motivates me to perform better. The mutual insightful discussions there are really acrucial part of my research experience.My friends: Claire, Nirmalendu, Harsh, Somesh, Prashant, Miguel, Adriana, Patrick, Faisal,Aishwarya, Mike, Ankita, Saurav, Juuso, Sarah, Fagun, Ketaki, and Shahzaib are always withme, through thick and thin. Special thanks to my past, present and future roommates (readfriends): Kristian, Mielle, Keith, Meredith, Seth, Marina and Moshe for supporting me andbeing the awesome roommates you always are.Last, but not the least, my mother Rinku Das, father Bhabananda Das, and late grandmotherLaxmi Das are always in my thoughts and have an important role in this achievement.xiiTo my family.xiiiChapter 1IntroductionLight-weight tissue-paper products like bathroom tissues, facial napkins and kitchen towels arean integral part of modern life. The apparent density of these products is generally below 300kg/m3 and the basis weight is less than 50 gram per square meter (gsm) [3]. For comparison, thegrammage of regular copy paper is 80 gsm and can be as high as 300 gsm for card boards. Unlikeprinting and packaging paper materials, tissue papers need to be more soft, stretchy, toughand water-absorbent. These properties are achieved by a specialized manufacturing techniquecalled creping. Creping increases bulk, stretch and toughness by damaging the paper sheet in acontrolled fashion and by introducing a strong micro-structure, known as the creping structure[3, 4, 5, 6]. Combined with damaged fibers and network characteristics, the creping structure isbelieved to govern the mechanical properties and quality of tissue products. Therefore, a carefulunderstanding of creping mechanics, related manufacturing parameters, and how creped paperstructural properties affect tissue quality is important.Recently Pan et. al. developed a one dimensional discrete particle model [7] to simu-late creping. This model was able to explain the underlying physics, quantify post-crepingmicro-structure, and predict the stress-strain behaviour of creped tissue. The theoretical workis complemented with experiments in this thesis. The primary focus of the thesis is developingexperimental techniques to visualize creping and to quantify creping structure. The tensile prop-erties of commercial tissue are also determined from experiments and visualization techniquesare developed to observe the network structure deformation with load to understand the mecha-nism of deformation and failure. Section 1.1 is a brief description of creping in context of tissueproduction. Section 1.2 is an overview of the physics of creping. Section 1.3 describes crepestructure quantification methods. Section 1.4 is a synopsis of theories describing the stress-strainbehaviour of paper products. Finally, section 1.5 lists the objectives, and describes the structure,of this thesis.1.1 Tissue Production and CrepingTissue production is a complex process involving various techniques. A simple schematic oftissue manufacturing process with related process parameters is shown in Fig. 1.1. The entireprocess can be divided into four major sub-processes: forming, dewatering, creping, and finallyreeling and converting (see Fig. 1.1 top). Tissue production usually starts from a headbox at the1forming section (see the triangular shaped feature towards the left hand side in Fig. 1.1). It con-tains a very dilute slurry of mechanically or chemically refined pulp fibers, also known as stock,at 0.15%-0.5% solid content. The composition of fibers (see parameter 1 in Fig. 1.1) in the stockis important; more short hardwood fibers (aspen, maple etc.) are preferable for softer productslike facial tissues whereas, longer softwood fibers (pine, beech, spruce etc.) are required forstronger products e.g. kitchen towels. In the forming section, one or multiple layers of stockfibers (together known as pulp mat) is(are) deposited and dewatered in between a forming fabricand a dewatering felt (sometimes another forming fabric is used instead of a dewatering felt).The topography of the forming fabric imparts some initial structure (see parameter 2 in Fig. 1.1)to the pulp mat. The pulp mat is then transferred to a water absorbing felt in the dewatering sec-tion, where the solids content of the mat is increased to about 20%. A pressure nip presses thismat against a rotating cast iron cylinder at this point. The pressing action not only expels waterfrom the mat, but also densifies the mat and strengthens the bonds between fibers (see parameter3 in Fig. 1.1). Finally, the wet mat, which has a similar consistency to cooked lasagna noodles,is transferred to the surface of Yankee dryer (the big circular feature at the right side in Fig. 1.1).The rotating (40-50 rotations per minute) yankee drum is precoated with a mixture of ad-hesive chemicals, release agents and modifiers sprayed from a nozzle (see under the Yankeedryer in 1.1). The coating spray intimately adheres the wet mat to the yankee drum and forms aprotective layer of chemicals on the yankee surface (see parameter 5 in Fig. 1.1). The internallysteam-heated yankee and drying hoods rapidly increase the solid content of the web to 95% bythe time it reaches the other end of the drum (see parameter 4 in Fig. 1.1), at which location,a sharp doctor blade scrapes the dry web off the surface. The violent interaction between thesheet and blade, known as dry creping, damages the network of fibers, resulting in foreshorten-ing of the sheet, improved stretch, bulk and softness with concomitant tensile degradation. Thestrength and uniformity of adhesion between yankee surface and web is extremely important forefficient creping [6]. If the adhesion is strong and even, fiber-fiber bonds will break when thepaper web hits the blade, increasing bulk, softness and absorbency. Too strong adhesion mayresult in sheet break whereas, weak adhesion will result in sheet falling off the yankee cylinderbefore it hits the blade.Once creped, the damaged, compressed web is reeled downstream for converting and em-bossing at a lower speed than the yankee surface speed (see parameter 6 in Fig. 1.1). Thedifference between yankee dryer surface speed and reel speed divided by yankee surface speedis known as the creping ratio (see parameter 7 in Fig. 1.1). Since reeling speed of tissue is lowercompared to yankee surface speed, more material is packed in a given length after creping. Thisincreases the overall bulk of the tissue. The creping ratio is generally held between 15%-20%where bulk increases with increasing creping ratio [4]. High bulk is correlated with greater bulksoftness and water absorption capacity.2Figure 1.1: Schematic of tissue manufacturing process and impactful parameters (1-9)3Figure 1.2: Creping structure as seen from a (a) surface image and a (b) cross-section image of a highbasis weight creped tissue. The basis weight of the uncreped paper was 25 gsm and was creped in a crepesimulator. The surface image is taken under oblique diffuse light. The cross section is illuminated bya line laser and imaged subsequently. The creping folds can be seen clearly from both the surface andcross-section image. However the crepe folds are much smaller in a commercial tissue.Observation of creped tissue surface reveals a series of folds, aligned roughly parallel tothe cross direction (CD) (see Fig. 1.2 (a)). The folds are visible as approximately parallelbands of bright and dark regions. The actual micro-structure of these folds is seen clearly in across-section image (see Fig. 1.2 (b)). These folds are produced during creping and are key toincrease machine direction (MD) stretch and bulk in the final tissue. Uniform and short crepefold structure is usually, but not necessarily, desired in a good quality tissue. A multitude offactors e.g. forming fabric induced wire mark pattern (same as parameter 2 in Fig. 1.1), basisweight, web-yankee adhesion strength and creping blade geometry affect this fold structurelength-scale and its distribution [8]. The distinct wire mark pattern generates finer crepe folds,which are generally in phase with the wire marks [4]. If the wire mark is too heavy, it createsholes in the sheet. High web basis weight and strong sheet-yankee adhesion generally createsfiner crepe folds [5, 9].A schematic of the geometry of creping blade is shown in Fig. 1.3. The angle between cre-ping blade and yankee surface is known as creping angle. Creping angle also plays an importantrole by varying the creping force that doctor blade applies to scrape off the web from the yankee4Figure 1.3: Simplified 2D schematic of doctor blade and Yankee dryer contact and various angles relatedto creping mechanism. The size of the doctor blade is exaggerated for clear presentation. In reality theblade size is much smaller compared to the size of the Yankee. The various angles shown are crepingangle (δ ), blade bevel angle (β ), and blade contact angle (α). Nomenclature of the angles vary amongindustries and literature.surface. In real tissue machines this angle is generally 85o−95o. An increase in creping anglegenerally creates a fine and uniform crepe structure [5]. Understandably, an optimal combina-tion of these parameters is needed to produce a quality tissue. However, there is no systematicway of choosing values for these parameters because of the lack of scientific understandingof creping mechanism and the effect of the process parameters on the tissue quality. We willapproach this problem with the objective to develop experimental techniques to understand cre-ping mechanism, and to quantify tissue structural and mechanical properties. Data from theseexperiments will compliment analytical and numerical works.1.2 Mechanism of CrepingTissue production is extremely rapid. 30 meters of creped paper sheet containing hundreds ofthousands of crepe folds can be manufactured each second. with each fold generation occurringwithin a few microseconds. High speed imaging studies close to the creping point and numericalsimulations are two major approaches taken by previous researchers [5, 7, 9, 10, 11] to under-5Figure 1.4: Four stages (a-d) of a creping fold formation. Only a cross-section of the tissue, blade andyankee is shown. The yankee drum surface is approximated as a linear surface. Wire mark patterns beforecreping and internal sheet damage are not shown in this schematic.stand creping. High speed imaging in a real tissue machine, however, is notoriously difficultand expensive because of accessibility issues and the dusty, steamy environment. All publishedexperimental studies are done either in small scale creping simulators or pilot tissue machines atsignificantly low production speed. In the last two decades, more emphasis has been placed onnumerical simulation of creping based on the previous phenomenological observations. Fromthese experimental and numerical studies, creping is understood as a fracture-driven buckle-delamination process.Hollmark imaged the blade-sheet interaction during creping in a pilot tissue machine oper-ating at 140 meters per minute and observed how the folds were generated [5]. Creping ratio inthis experiment was held at 32%. According to this study, once the sheet hits the blade and getsseparated from the surface, a loop forms at the edge of the blade. A second loop forms on topof the first one as more web is fed. This loop formation continues until the pile collapses andis pushed off from the yankee surface. This study gives a phenomenological description of thefold generation, but does not explain the underlying mechanics involved in this process. Rama-subramanium et al. built a laboratory scale creping device [9] and made similar observations atthe blade-sheet interface. Typical yankee speed in their experiments was around 144 meters perminute but there was no creping ratio. They explained the mechanism of individual fold gener-ation in four stages as shown in Fig. 1.4 (a) - (d). In the step (a), stress develops in the web andadhesive layer as the sheet hits the blade. This stress cracks the sheet-adhesive interface and the6crack propagates until the debonded segment reaches a critical length in step (b). Afterwards instep (c) and (d), the debonded segment of web buckles to produce a single fold and gets pushedoutwards. This process repeats itself to give the continuously folded structure that is similar towhat is observed in commercial tissues.A significant drawback of all the experimental works was that the yankee surface-speed wasan order of magnitude lower compared to a real tissue machine. Yankee speed is an importantparameter because it controls the drying time of partially wet web on the yankee surface. Vari-ation in drying time can alter moisture content and elastic properties of both web and adhesivelayer as well as sheet yankee adhesion just before creping. Also the effect of inertia of web oncreping force is not known. The mechanism of crepe at high speed has not been explored exten-sively and whether it is significantly different or not is still an open question. This literature-gapwill be partially addressed in this thesis.Ramasubramanium et. al developed a mechanics of material model of creping [10] basedon strength based failure criterion of the adhesive layer and Euler buckling analysis of the papersheet. This model predicted results qualitatively consistent with experimental data and knownindustrial observations. However it was a quasi static analysis and did not try to predict themechanism at higher speeds. More recently, Pan et. al. applied a discrete element model[7] to simulate creping of a single layer of elastic web bonded to a rigid yankee surface atrealistic yankee surface speed. They observed similar stress development, delamination andbuckling phenomenon in the numerical simulation of creping fold formation. Creping angle andweb thickness before creping were shown to be most significant factors affecting creping foldwavelength and the effect of yankee speed was shown to be negligible. But it is important topoint out that in the model, the effect of changing yankee speed on properties of the sheet andadhesive layer was not taken into account. Hence the conclusion about effect of creping speedis likely not to be realistic. Also, this model was limited by the one dimensional simplificationof paper, its homogeneity and elastic response.It is evident from the previous works that creping significantly alters the the planar fiber-network structure of uncreped paper. It plastically bends individual fibers, introduces fiber axialmicro-compression, and causes internal fiber-fiber bond damage [3]. High bulk, stretch, soft-ness, water absorption capability of tissue comes from this altered structure. Therefore, anunderstanding of tissue micro-structure and related quantifiable parameters is important to eval-uate creping-efficiency and tissue-quality. This is discussed in the next section.1.3 Crepe Structure QuantificationCreped tissue paper is primarily a network of plastically deformed loosely bonded fibers, prefer-ably aligned with the machine direction, with a folding pattern. The in plane and out of planedimensions of this folding pattern varies in both machine and cross direction as a result of vari-ation in formation, sheet yankee-adhesion and many other factors. Although the wave-like pat-7terns are not exactly periodic, two quantifiable structural parameters can be attributed to them.First, the number of folds per unit machine direction length of unstretched tissue paper is knownas the crepe count. There is no direct correlation between tissue quality and crepe count, butuniform small crepes are generally an indication of stable production and high-quality product.Nonuniform and bigger folds indicate non uniform or inadequate adhesion between sheet andyankee surface, poor formation or excessive blade wear. So, crepe count is often determinedin a tissue machine during production (online) or in laboratory (offline) after creping to assesstissue quality and predict the time of blade change. Out of plane dimension of the crepe folds isknown as creping amplitude, which increases tissue caliper and bulk. This section is devotedto explaining the techniques available to quantify the structural aspects of crepe folds, namelycrepe count and creping amplitude.There is no standard protocol for crepe count measurement, partially because it is not a well-defined parameter in tissue industry. Generally creped tissue surface is imaged in laboratory at60X–70X magnification to identify peaks and valleys. Distance between consecutive peaks orvalleys is measured to find fold wavelength and reciprocal of the measurement is recorded ascrepe frequency [12]. Alternatively, a 3D image of tissue surface structure can be reconstructedby confocal microscopy. Cross-sections along machine direction can be extracted from the3D structure to determine crepe count. In another technique, a mechanical profilometer canbe used to measure surface profile of tissue and dominant frequency can be extracted usingFourier Transformation. The dominant frequency is believed to come from crepe folds and usedas a measure of crepe count [12]. Recently, photometric stereo method based surface profilecharacterization has also been used instead of profilometer [13]. Microtome sectioning of tissuecross-section can be done to take images of cross-section along machine direction. These imagescan be analyzed to extract crepe frequency as well as out of plane amplitude of crepe folds[14]. All these laboratory based offline methods are time consuming and a large number ofmeasurements are required for a statistically meaningful measurement. These techniques arehard to implement in a tissue-machine to measure crepe count or caliper in real time duringproduction.Sabater et. al. demonstrated a way to measure the tissue surface state index or undulatoryvariations of tissue surface just after creping in a running tissue machine [15]. In this method,a laser light is directed on the moving creped surface and a photoelectric device is placed at anangle to collect the back-scattered laser light. Using optical proximity detection method andsuitable signal analysis, geometric data of the creped sheet along a straight line in machine di-rection can be extracted. This method can be employed online to continuously measure crepefold pitch and the average amplitude of the crepe folds. But this method does not give informa-tion on cross directional variability of crepe count since it measures crepe pitch along machinedirection in a straight line.In the past few years, measurement of crepe structure using surface images has been in-vestigated because of its simplicity and potential of online measurement across the paper web[16, 17] In this method, bright light is shined from an oblique angle relative to the plane of8the paper, preferably along machine direction. Hills in the crepe structure are perceived asbright regions and the valleys as darker regions. However, the grey-scale periodicity seen insurface images is a perception of the folds and is not related to actual physical depth. Imagesare processed using suitable signal processing methods e.g. spectral analysis to extract domi-nant frequency of the folds. Further analysis has also been done to quantify small scale surfaceroughness [18]. Recently due to a drop in the price of high-speed imaging equipment and an in-crease in computational power, these techniques are incorporated in online tissue quality controlsystems [18]. However, the spatial pixel resolution of surface images are about 60-100 µm. Forcommercial grade fine tissue papers, crepe folds can be as small as 200 µm. A better resolutionimaging system can improve quality of measurements significantly. In Chapter 2 of this thesis,development and application of a surface imaging-based crepe-count measurement techniqueare discussed. A high resolution commercial camera and a diffuse light source is used to build alaboratory scale crepe-count device first. Image pixel resolution is about 7 µm. Instead of com-puting crepe-count as a single number to characterize tissues, emphasis is given on determiningthe distribution of crepe-counts. Distribution or variability in crepe count measurement from asurface image can tell us about the uniformity of creping, a highly desired characteristic of goodquality tissue. To prove reliability of the surface image based crepe count measurement tech-nique, Scanning Electron Microscope images of tissue cross-sections are analyzed to determinecrepe-count and compared against surface imaging measurements. Good correlation is found.In chapter 3, the surface image based quantification method is used in a real tissue machine toextract crepe count of tissue online. Images of tissue sheet moving at 25 m/min are taken usinga high speed flash (pulse width 1 µs) synchronized with a commercial camera and quantifiedonline. The measurements are compared with offline crepe counts. Again good correlation wasfound between the measurements. Capability of the surface imaging system as a online processmonitoring tool is established. If this system is integrated with a tissue machine, systematicstudies on effects of blade wear, furnish, forming fabric, and adhesive coating can be done tocomplement theoretical studies and numerical simulations of creping.1.4 Tensile Properties of PaperTensile behaviour of paper materials change significantly after creping. Fig. 1.5 shows typicalstress strain curves for paper before and after creping [4]. Evidently creping increases stretchand work to rupture with a corresponding degradation in strength and elastic modulus, especiallyin machine direction (MD). The entire stress strain curve in both machine and cross direction fortissue and low-density grades is important for several reasons. The strength characteristics alongmachine direction directly influences runnability [3]. Cross direction stretch is important forsuccessful converting operations and embossing. Elastic modulus of dry creped tissue is widelybelieved to govern bulk softness. Recently attempts have been made to produce soft tissuepaper without creping, but those attempts have not been particularly successful. Therefore, an9Figure 1.5: Stress strain behaviour of paper before and after creping. Creping increases stretch and workto rupture and lowers strength in both machine direction (MD) and cross direction (CD). However, theimpact is higher along MD. Reconstructed from [4]understanding of the principal mechanisms that govern the tensile response, elastic modulus,strength and stretch is important to successfully design a tissue as per consumer-requirementwith efficient production speed.Tensile response of paper materials, creped or uncreped, have significant nonlinearity. Inpast half a century, two-dimensional network theories [19, 20, 21] have been developed to par-tially explain tensile behaviour of uncreped paper materials. Not much has been done to un-derstand the tensile behaviour of creped tissues. Cox’s analytical expression of elastic modulusof a plane fibrous network is the foundation of network theory for paper [19]. He models thefiber network as a homogeneous mat of systematically oriented infinitely long, straight, linearlyelastic fibers. He assumes perfect inter-fiber bonding free of friction and sliding and consid-ers only axial stiffness of the fibers in his analysis. Assuming an initial homogeneous stressfield, he derives an analytical expression of elastic modulus of the network and concludes thatunder the assumption of randomly oriented fibers, elastic modulus of the network is one-third10of elastic modulus of fiber. This conclusion is experimentally verified for high density papermaterials [20] made of straight fibers free from any gross kinks and cramps. These fibers have alinear tensile response. High density or intimate fiber fiber bonding is important to create a near-homogeneous stress field. However in reality, elastic modulus of paper materials is less than onethird of average elastic modulus of fibers. This is because of two reasons. First, most pulp fibersused for paper and tissue production have a lower elastic modulus after pulping because of thenature of pulping process and the fibers are often oriented preferentially than randomly in thesheet. Second, for low density paper products, stress distribution through fiber length can varygiving rise to inhomogeneous stress field and inefficient inter fiber stress transfer. These factorsare not taken care of in Cox’s theory. Also, Cox’s analytical model does not give insight intothe viscoelastic nature of paper materials and cannot be extended to plastic regime of the tensileresponse.Page et. al. modified the analytical expression of elastic modulus derived by Cox. Based onexperimental observations they concluded that entire tensile response of fibers, degree of inter-fiber bonding and presence of kink, curl and micro compression in fibers affect elastic modulusof paper [20, 21, 22]. Three factors were incorporated in the expression of elastic modulus. Thefirst depends on average fiber elastic modulus. The second one takes into account distributionof fiber orientation and the third one depends on fiber-fiber bond density or relative bondedarea. The same argument of three governing factors was extended to plastic region of the tensileresponse. Non-linearity of the stress strain curves was believed to be largely related to the non-linear tensile response of fibers [23, 24]. Another theory behind the strain-softening behaviourand plasticity of paper is based on the concept of progressive micro-damage and rupture of fiber-fiber bonds with load from the very beginning of straining process [25]. However the secondview, also known as a structuralist view, is largely discredited by experimental evidence [23].However, in both Cox’s work and Page’s modified theory, the underlying assumption ofa planar mat of straight fibers does not hold true for creped tissue papers. Creped tissue, asdescribed before, consists of loosely bonded fibers that are deformed plastically out of plane.So, the applicability of the analytical theory developed by Cox and modified network theoriesbuilt upon it is questionable for creped tissue. Also tissue paper network structure can undergosignificant deformation to accommodate large strain before failure, which changes the fiber ori-entation factor of elastic modulus. Because of the complex nature of creped tissue paper, notheoretical work has been done to understand the entire stress-strain behaviour. Observation ofstructural evolution of tissue papers is a starting point to understand how the structure respondsmechanically to load. In Chapter 4 of this thesis, imaging techniques are used to observe de-formation of tissue fiber network at various levels of strain to delineate the underlying networkbehaviour.111.5 Objectives and OutlineBased on the above literature review and the gaps found in understanding creping mechanismand mechanical properties of creped tissue, the following objectives have been identified for thisthesis.1. To develop and validate a surface image based visualization technique and processingalgorithm to quantify crepe structure of tissue paper.2. To demonstrate the capability of high speed imaging techniques for imaging and quanti-fying tissue structure in a running machine.3. To visualize creping mechanism at high speed regime (>1000 m/min).4. To characterize tensile response of crepe tissue with simultaneous observation of the net-work deformation.In Chapter 2, a surface image based crepe count measurement technique is developed andvalidated using measurements from through thickness cross-sectional images of tissue. In Chap-ter 3, it is demonstrated that the surface imaging technique can be integrated with a high-speedimaging setup to do online crepe count measurement. Also a line laser based cross-section imag-ing technique integrated with a high speed camera is used to observe creping mechanism at highspeed. In Chapter 4, structural evolution of tissue network under tensile load along machinedirection is explained. Finally conclusion and future works are presented in Chapter 5.12Chapter 2Tissue Structure Quantification2.1 IntroductionExperimental techniques to visualize and quantify creped tissue structure have been exploredbefore [14, 15, 16]. These techniques can be broadly classified in four different types: crosssection imaging, surface imaging, reconstruction of line profile, and reconstruction of surfaceprofiles of tissues (see Section 1.3 for more details). A brief comparison of these techniques isgiven below in Table 2.1.Table 2.1: Comparison of crepe structure quantification methodsCross section imaging and surface profile reconstruction [14] can accurately quantify bothcrepe count and creping amplitude. However, data acquisition in these techniques is time con-13suming, making it difficult to implement them in a running tissue machine. Line profile recon-struction method based on laser based optical proximity detection is faster and can be integratedin a running tissue machine [15]. Recently, the surface imaging technique has been used ina tissue machine to extract crepe count and caliper in real time [18]. However, the accuracyof the crepe count measurement based on surface images has not been investigated before. Inthis chapter, tissue structure quantification technique based on cross section imaging (through-thickness) of tissue paper is described first (section 2.2) followed by a surface imaging basedtechnique (section 2.3). The results are then compared in section 2.4 to investigate the accuracyand robustness of surface imaging methods in determining crepe count and its distribution. Asystematic protocol of imaging and analysis is also established from this exercise.To image cross sections, a line laser and a Scanning Electron Microscope (SEM) is used.The line laser based illumination technique (combined with a commercial DSLR camera) toimage tissue cross section is very useful because of its minimal invasion on crepe structure.This technique works well for tissues with high basis weight (>25 gsm per ply) and with coarsecrepe structure (>500 µm creping wavelength). However for low basis weight commercialtissues with finer crepes, this technique is seen to have inadequate image resolution. For thesetissues, SEM is used. SEM has very high image resolution and depth of focus, which ensuresclear images required for finer crepe structure. Suitable image analysis methods are used toextract the creping pattern from the images. Finally quantification techniques are developed toquantify the crepe count. An experimental setup is then built to image tissue surface. An imageprocessing and quantification algorithm is developed to extract dominant crepe count. Thoughthe imaging concept is same as any other offline surface-imaging method, this technique canquantify crepe count with an estimation of its variability in a tissue. Degree of variability increpe count is important to tissue manufacturers for quality control purpose.Two high basis weight tissue paper samples produced in a crepe simulator and four low basisweight commercial grade tissues are characterized in this experiment. These samples, namedGrade 1-6 were collected immediately after creping without any embossing or converting. Grade1 and Grade 2 are high basis weight samples. Single ply laboratory made paper strips werecreped in a crepe simulator (see section 3.1.1 for more details) and collected. Basis weight ofthe uncreped sheet was 25 gsm and caliper was 63.923 µm. Grades 3-6 are low basis weightcommercial tissues produced in a real tissue machine. Details of the Grade 3-6 is in Table 2.3.Table 2.2: Details of Grade 3-6. All the measurements were done on a 2 ply creped tissueName Basis Weight Caliper Softness Nature of(gsm) (µm) (Handfeel) end useGrade 3 14.3 100 78 bathroom tissueGrade 4 20.1 158 80 bathroom tissueGrade 5 16.9 116 90 facial tissueGrade 6 17.9 151 85 bathroom tissue142.2 Through-thickness QuantificationImaging of tissue cross section is challenging because tissue paper is very thin and delicate. Veryhigh image resolution and a sample preparation technique with minimal effect on tissue structureare imperative to get an image that can clearly represent the undeformed micro-structure. Twodifferent ways of visualizing tissue cross section, namely line laser and SEM, are explored inthis section. Suitable image processing and analysis methods are used to extract and quantifythe creping structure.2.2.1 Cross section Imaging MethodsFigure 2.1: Tissue cross section imaging techniques based on (a) Line laser illumination and (b) ScanningElectron Microscope (SEM) imagingFig. 2.1 (a) illustrates the line laser based illumination technique used to image tissue crosssection. A low power (500 mW) line laser is used to illuminate a narrow cross section (seefigure) along the machine direction. The illuminated cross section is imaged subsequently usinga high-resolution commercial camera (not shown) placed at an angle to the tissue surface. Thedirection of imaging is shown in the figure. This method was successful in imaging high basisweight grades (Grades 1-2) with high caliper, less porosity and large crepe fold size. To reiterate,these were not commercial grade tissues, but rather produced in a laboratory scale creping rig.Caliper and the dominant length-scale of crepe folds are significantly smaller for low basisweight commercial grade bathroom and facial tissues. So, the image resolution in line-laserbased imaging is not enough to capture the micro structure adequately. Low magnificationScanning Electron Microscope imaging is used for these low-density commercial grade tissues15(Grades 3-6). However, SEM imaging is not as straightforward and non-invasive as the previ-ously discussed line laser imaging, but rather involves a complex sample preparation technique[26]. One very common technique is to impregnate delicate samples in a suitable resin block.For tissues, anhydrous Spurrs Resin is used to completely fossilize multiple rectangular stripsof tissue paper in a cylindrical plastic mold. For two-ply tissues, each ply is separated beforeputting them in the mold. Bleached cardboard strips are placed between two consecutive pliesto prevent them from sticking together. The card boards strips also hold the samples in placeduring the baking of the resin block at 70o C. After 24 hours the cylindrical resin block is takenout of the oven and separated from the mold. Fig. 2.1 (b) shows one side of such a resin blockwith 3 fossilized tissue plies. These layers are visible as very thin white horizontal lines (shownin inset) in the middle of the block. The short thick white lines close to the edges are parts ofcardboard strips that were used to hold the tissue samples in place. The entire surface is pol-ished and chemically etched to reveal several microns of unconstrained tissue structure. A verythin (∼5 nm) conductive layer of gold palladium coating is deposited on the entire surface andimaged with a Scanning Electron Microscope at 100X magnification.2.2.2 AnalysisTypical cross section images based on line laser illumination and SEM are shown in Fig. 2.2 (a)and (b) respectively. The magnification is much higher in the SEM image compared to the linelaser technique. This makes the SEM images much detailed compared to those obtained fromline laser. This can be appreciated from the level of details that can be seen in SEM images (seeFig. 2.4 top). Image details of a typical line laser and an SEM image are given below.Table 2.3: Details of an image obtained from line laser and from SEM techniqueName Spatial dimension Spatial resolution Cross section length(mm × mm) (µm / pixel) along MD (mm)Line laser 10 × 4.67 25.39 10SEM 1.28 × 0.59 1 1.28The superimposed blue lines on the cross section images approximately follow the crepingfold structure. Image analysis methods are used to extract these patterns. For images takenwith line laser illumination (Fig. 2.2 (a)), greyscale-intensity weighted average distance of thepixels from the bottom edge is used as a measure to represent the creping pattern. In ScanningElectron Microscope images, top and bottom peripheral lines (black lines in Fig. 2.2 (b)) of thecross section are extracted first using edge detection. The mid line of top and bottom peripherallines is then extracted to represent the creping pattern. The creping patterns are smoothed usinga third order Savitzky Golay filter.Clearly, the observed creping structures are not exactly periodic. Wavelength and amplitudeof the folds vary as we go along machine direction. However, dominant wavelength and fre-16Figure 2.2: Typical tissue cross section images and analysis methodology. Images obtained from Linelaser imaging (a) and SEM imaging (b) with extracted creping patterns (in blue). The dominant crepingfrequency of the creping fold pattern shown in (b) is obtained from its autocorrelation function in (c).quency of these folds can be extracted from their autocorrelation function. The advantage ofusing autocorrelation function over popular Fourier Transformation or power spectral densityis its versatility in determining dominant periodicity for any kind of signals, harmonic or non-harmonic. Also this technique is less sensitive to image noises, adaptable to any signal length,and computationally more efficient.Part of the autocorrelation function of the extracted cross section shown in Fig. 2.2 (b) isgiven in Fig. 2.2 (c). The inherent periodicity of the cross section is well represented by theperiodic nature of the autocorrelation function. Position of the first peak (indicated by a blackcircle in figure) is used as a measure of dominant fold wavelength of the cross section. Itsinverse is used as a measure of dominant crepe frequency of the cross section. For each grade oftissue, several cross section images are taken and analyzed for a statistically robust estimationof crepe count. Each image is taken at a different location (several meters apart) of the tissue for17independent and unbiased sample images. Therefore dominant frequency of each cross sectionis independent from the others. The mean of dominant crepe frequencies of all the cross sectionsis calculated to estimate the dominant crepe count of one grade. Sample standard deviation ofthe dominant creping frequencies is used as a measure of variation in crepe count. Results arepresented in next section.2.2.3 ResultGrades 1-6 are characterized using the cross section quantification technique. Dominant meancreping wavelengths and crepe counts are tabulated in Table 2.4. The number of images ana-lyzed (sample size, n) are given in the second column. Dominant creping wavelength and crepecount for all the grades are given in third and fourth column respectively. It can be readilyseen that Grade 1 and 2 have much lower dominant crepe count, therefore larger creping wave-lengths, compared to Grades 3-6. This is expected since the commercial grade tissues (Grades3-6) are much finely creped than the samples produced in the crepe simulator. Among grades3-6, grades 3 and 5 have similar dominant crepe count, which is significantly higher than grades4 and 6. This indicates that there is a difference in length scale of crepe folds among the com-mercial grades as well. This difference can be visually observed from the cross section images(discussed later). There is also a large variation associated with each crepe count measurement.This is evident from the standard deviation values, which means that the creping fold structureof a tissue grade varies spatially. The source of this variation is the stochastic nature of cre-ping process, inhomogeneity in the base sheet and sheet yankee adhesion, which in turn affectthe creping pattern. This variation is not associated with the accuracy of measurement, but anindication of randomness of the crepe structure in tissue.The summarized results for all six grades are explored in Fig. 2.3 - 2.5 in more detail,visually and statistically. For clear presentation, 2 grades are grouped together in one figure.For both grades, typical cross section images are shown at top. In each image, ”MD” representsmachine direction and ”ZD” represents thickness direction. From Fig. 2.3, it can be seen thatTable 2.4: Mean dominant wavelengths and crepe counts of Grades 1-6 based on through thicknessimaging.Name Number of Mean dominant Mean dominantimages , n wavelength (µm) crepe count (folds/in.)Grade 1 71 671 ± 143 40 ± 11Grade 2 44 583 ± 145 47 ± 12Grade 3 22 246 ± 60 109 ± 27Grade 4 16 321 ± 68 82 ± 17Grade 5 18 252 ± 61 106 ± 25Grade 6 17 342 ± 61 77 ± 1418Figure 2.3: Typical cross section images (top) obtained from line laser based imaging and correspondinghistograms (bottom) of crepe count for Grade 1 and Grade 2. Mean (µ), sample standard deviation (σ )of crepe counts, and sample size (n) are indicated in each histogram. Y axes of the histograms have beennormalized with respect to sample size (n).)19Figure 2.4: Typical cross section images (top) obtained from SEM imaging and corresponding his-tograms (bottom) of crepe count for Grade 3 and Grade 4. Mean (µ), sample standard deviation (σ )of crepe counts, and sample size (n) are indicated in each histogram. Y axes of the histograms have beennormalized with respect to sample size (n).)20Figure 2.5: Typical cross section images (top) obtained from SEM imaging and corresponding his-tograms (bottom) of crepe count for Grade 5 and Grade 6. Mean (µ), sample standard deviation (σ )of crepe counts, and sample size (n) are indicated in each histogram. Y axes of the histograms have beennormalized with respect to sample size (n).)21the wavelength of crepe folds in Grade 1 and 2 is approximately half a mm (compare with thescale bar). This is in alignment with the mean dominant wavelength found for these 2 grades(see Table 2.4, column 3). For grades 3-6 the creping folds are smaller than grade 1 and 2 (seeSEM images in Fig. 2.4 and 2.5 and compare with scale bar). It can also be seen that grade 3and 5 have smaller folds compared to grade 4 and 6, which is again consistent with the previousdiscussion of dominant creping wavelengths and crepe counts.The histograms of dominant crepe frequencies are shown below the cross section images ineach figure. The non-periodic nature of the folds and spatial variation are evident here from thewide distribution of the creping frequencies in the histograms. However, the number of sampledcross sections is statistically insignificant (except for Grade 1 and 2) to draw conclusions aboutthe nature of distribution of these dominant crepe frequencies. To get a more statistically robustvalue of mean crepe count and a better understanding of the distribution, more cross sectionimages need to be analyzed. However, the large amount of effort needed to take cross sectionimages limits the number of images that can be taken. This problem can be alleviated by ana-lyzing surface images. This will be discussed in the next section. To conclude, in spite of thelimited number of samples, the cross section imaging can differentiate between tissue gradeswith significantly different crepe counts, both visually and quantitatively.2.3 Surface Image QuantificationSurface image based quantification of crepe structure for Grade 1-6 is carried out to estimatecrepe count and then compared with the results obtained using the through thickness technique.The goal is to demonstrate the capability of the surface imaging technique as a potentially morepowerful tool to accurately estimate mean crepe count and the distribution.2.3.1 Surface Imaging SetupThe tissue surface imaging setup is shown in Fig. 2.6. The setup consists of a tensile stage withtwo grips that hold a tissue sample under suitable tension, a diffuse light source (VELA oneflash), a highly uniform back-light source (SCHOTT A08925 and A20500 ACE 1 EKE lamp)and a DSLR camera (Nikon D7000) placed perpendicular to the tissue surface. Tissue samplesof 100 mm length along machine direction and 25.4 mm width along cross direction are cutand placed between the jaws in the imaging area. The tissue holder grip on left hand side isfixed. The other grip can be linearly displaced along the stage. It is attached to a pulley andweight system (not shown in figure) for tension loading. It is important to apply a low tensileload (< 5 N/m) along machine direction to avoid wrinkling and deflection of tissue surfaceunder its own weight while imaging. The wrinkles and deflection of surface stands in the wayof uniform illumination, which is imperative for a good quality surface image. The diffuselight used to illuminate the tissue surface is tilted at an angle θ to the normal of tissue surface22Figure 2.6: Tissue surface imaging setupto highlight the crepe folds which are not clearly visible under ambient light. This angle isgenerally held between 50 to 70 degrees. It is seen that varying this angle does not producesignificant differences in measurement. In a different imaging configuration, the uniform back-light can be placed below the sample to expose the internal network structure.2.3.2 AnalysisAnalysis methodology to extract mean dominant crepe count and corresponding variability fromsurface images is shown in Fig. 2.7. A typical surface image of a commercial tissue (Grade-6)is shown in Fig. 2.7 (a). The crepe structures are visible as alternate bands of bright and darkregions aligned roughly with cross direction. There are some very large scale intensity variationin some areas of the image. This comes from large scale physical undulation in surface. Theyare not related to creping structure. Dimensions of the image along machine and cross directionare denoted as lMDim and lCDim respectively. The thin rectangular box in the image indicates atypical cross section along machine direction. Length of this cross section (lMDcs ) along machine23Figure 2.7: Analysis methodology to extract mean dominant crepe count from one surface image. Typicalsurface image (a) of a commercial tissue. The image has 2 segments. 160 cross sections can be extractedfrom each segment. A typical cross section (white rectangle) is shown in segment 1. Autocorrelationfunction of the cross section is shown in (b). Cloud of autocorrelation functions of multiple cross sectionsin segment 1 is shown in (c) and histogram of crepe frequencies of all the cross section in the image isshown in (d). Note that sample size of the histogram (n=310) is less than ncs/im=320. 10 cross sectionswere ignored as the dominant crepe frequencies found from them were statistical outliers.direction is taken as 5 mm. There are typically at least 10 or more crepe folds in this length fora commercial grade tissue. However, this length may be changed for other grades, dependingon the coarseness of crepe structure. Tissues will coarser crepe folds may need a larger crosssection length to adequately represent the crepe structure. A typical cross section width shouldbe an order of magnitude smaller than the cross direction width of crepe folds. Typically, widthof crepe folds are in the length scale of mm. Therefore, one typical cross section width (lCDcs ) canbe around 100 µm or less. Multiple rows of pixels are averaged spatially along cross-directionover the width to extract one cross section. The spatial averaging also reduces noise in greyscaleintensity of the image.Since the image in Fig. 2.7 (a) is much longer in machine direction than the length of24cross section, it can be segmented in at least 2 parts in machine direction. Number of segments(nseg/im), therefore, depends on the machine direction dimension (lMDim ) of each image and de-sired cross section length (lMDcs ). The number of segments can be calculated from the followingequation.nseg/im = f loor(lMDimlMDcs) (2.1)where ”floor (.)” is the function that takes as input a real number x and gives as output thegreatest integer less than or equal to x. The image in Fig. 2.7 (a) is segmented in 2 parts (seeimage).Multiple cross sections can be extracted from one segment. Number of extracted crosssection from each segment can be expressed asncs/seg = f loor(lCDimlCDcs) (2.2)Total number of extracted cross sections from each image (ncs/im) can be expressed asncs/im = ncs/seg×nseg/im (2.3)320 cross sections are extracted in this manner from the image in Fig. 2.7 (a). Each crosssection is analyzed using autocorrelation to extract the dominant wavelength and crepe fre-quency. The autocorrelation function of the cross section indicated in the image is shown inFig. 2.7 (b). Again, the first peak (denoted by a blue circle) is detected to extract dominantwavelength (346 µm) and crepe frequency (73 folds/in.) of the cross section. Many other au-tocorrelation functions obtained from other cross sections are shown in 2.7 (c). It is readilyseen that the shape of the functions are not similar to each other, rather vary from one crosssection to another. All the autocorrelation functions are analyzed to first determine all the dom-inant crepe frequencies. The reported mean crepe count is calculated after ignoring the outliersthat lie outside 2 standard deviation from the mean crepe frequency of all the cross sections.The histogram of dominant crepe frequencies of these cross sections (see Fig. 2.7(d)) is thenconstructed. Mean dominant crepe count (µ), standard deviation (σ ), and the number of crosssections (n, after ignoring the outliers in the first calculation) are indicated in the histogram.One very obvious advantage of the surface image based quantification is that a larger areaof tissue, therefore many cross sections can be analyzed from one image. The amount of effortrequired to take one surface image is an order of magnitude lower than imaging one cross sec-tion. Therefore multiple images can be taken to analyze an even larger area which will give amore reliable estimate of dominant crepe count than cross section imaging.25Table 2.5: Mean dominant wavelengths and crepe counts of Grades 1-6 based on surface imagingName CS per image, Images Mean dominant Mean dominant(ncs/im) (nim) wavelength (µm) crepe count (folds/in.)Grade 1 400 6 685 ± 103 38 ± 7Grade 2 400 6 548 ± 130 49 ± 12Grade 3 320 5 261 ± 54 102 ± 22Grade 4 320 5 326 ± 75 83 ± 22Grade 5 320 5 264 ± 61 102 ± 24Grade 6 320 5 346 ± 80 78 ± 212.3.3 ResultGrades 1-6 were analyzed using surface imaging technique. Mean and standard deviations of thedominant creping frequencies are tabulated in table 2.5. Number of images used for the analysis(nim) and number of cross sections (ncs/im) extracted from each image are also given. All thecross sections extracted are analyzed to compute the mean and standard deviations of crepingwavelength and crepe count after excluding the outliers. Again, it is shown that Grade 1 and2 have larger crepe folds compared to Grade 3-6. Also Grades 4 and 6 have larger crepe foldscompared to Grade 3 and 5. The same observations were made from cross section quantification.The detailed histograms are presented in Fig. 2.8 - 2.10 with a typical surface image. Again,2 grades are grouped together for better presentation in one figure. For each grade, part of onetypical surface image is shown (top) for visual interpretation. Machine and cross directionsare indicated as ”MD” and ”CD” in the image. The actual images used for the analysis arelarger in dimension. However, all the surface images shown here have the same dimensionfor better visual comparison. Comparing the surface images in Fig. 2.8 with Fig. 2.9 andFig. 2.10, it is visually obvious that the Grades 1 and 2 have larger crepe folds compared toother grades. It is hard to distinguish the commercial grade images from one another visually.However, the histograms showing the distribution of crepe frequencies vary from one grade toanother. Since the sample size of cross sections analyzed are sufficiently large (see n of eachhistogram), these distributions are more reliable and accurate. Again, the standard deviationof the creping wavelength and crepe count is seen to be considerably large, which comes fromthe large spatial variation of crepe folds. However, the nature of distribution is not explored indetail in this thesis. It can be seen that the histograms are not normally distributed, rather skewed(see histograms Grade 2, 4, and 6) like beta distribution or lognormal distributions. Therefore,other parameters like skewness can be determined to better represent these distribution on topof mean and standard deviation. In conclusion, the surface imaging method can distinguish thedifference in mean dominant creping wavelength length scale among the grades and also give anestimation of variation of the creping frequencies. The dominant crepe count and correspondingvariations from cross section imaging and surface imaging are compared in the next subsection.26Figure 2.8: Typical surface images and crepe frequency histograms for Grade 1 and Grade 227Figure 2.9: Typical surface images and crepe frequency histograms for Grade 3 and Grade 428Figure 2.10: Typical surface images and crepe frequency histograms for Grade 5 and Grade 6292.4 Discussion and ConclusionFig. 2.11 shows the comparison of mean dominant crepe count between the through thicknessand surface imaging technique. Surface image mean crepe count (Y axis) with correspondingvariation is plotted against cross section imaging mean crepe count (X axis). High basis weightGrades 1 and 2 (marked in red circles) are separated from the commercial grades 3-6 (markedin blue squares). Good agreement of mean crepe count is found between the 2 techniques.Also the amount of variation in creping frequencies are approximately of the same order. Alinear best fit of the data points is shown in a black dashed line, which deviates from the lineof exact similarity (black solid line). An exact one to one correlation (x=y) should not beexpected considering the stochastic behaviour of the folds. Also the cross section images andsurface images are based on finite length of samples imaged at different parts (several metersapart) of the tissue. Considering these sources of variability, the agreement is acceptable, whichestablishes the capability of surface imaging to accurately measure mean crepe count.Figure 2.11: Through thickness and surface imaging crepe count comparison of Grade 1-630Quantitative structural characterization of two high basis weight and four low basis weighttissue paper is done and compared. Significant conclusions are as follows.1. A simple laboratory-scale experimental setup to capture crepe tissue surface images isdeveloped. A systematic protocol of analyzing the surface images to extract quantifiablestructural parameters of tissue, for example crepe count, is also established.2. The results are compared with cross section imaging techniques for six different grades oftissues. The analysis proves that surface imaging is as accurate as cross section imagingbased crepe count quantification. The added advantages of surface imaging over crosssection imaging are its simplicity, robustness and possibility of statistical analysis for amore reliable quantification.31Chapter 3Visualization and Online Measurementin CrepingThe creped tissue structure visualization and quantification methods developed in Chapter 2 areused to visualize creping mechanism and quantify crepe structure online. Imaging experimentson creping have been done previously on a pilot tissue machine [5] and on a laboratory scalecreping rig [9] to delineate the mechanism of crepe fold formation. However, yankee surfacespeed in both experiments was 140 m/min, which is about an order of magnitude lower thanthe real tissue machine yankee speed. In the current work, creping mechanism is visualized at avery high yankee speed (> 1000 m/min) in a crepe simulator using a high speed camera and aline laser. The images are used to verify the creping mechanism at high yankee speed in section3.1. Though mechanism of creping is believed to be well established among researchers, it wasthe first step towards confirming the mechanism at realistic creping speed.Recently, other researchers have used very short duration pulsed light sources with machinevision cameras to capture tissue surface images and quantify crepe count in real tissue machine[18]. In this research, the surface imaging technique was integrated with a high speed flash toquantify crepe count in a tissue machine. Instead of a pulsed Laser Diode used by previousresearchers, a simple, inexpensive high speed LED flash is used with a commercial DSLR cam-era. Tissue images were taken and quantified in a running machine over a three hour period.Tissue samples were collected during the experiment and quantified in lab to verify the crepecounts obtained online. Online measurements agree well with offline measurements. Details ofthe experiments are presented with results in section Visualization of Creping Mechanism3.1.1 Experimental SetupThe lab-scale creping rig that was used to replicate and visualize creping at a high speed isshown in Fig. 3.1. It has five main parts; heated yankee cylinder, transfer roll, sheet moisturecontrol system, an adhesive spray and an instrumented creping blade. The sheet moisture controlsystem and adhesive spray are not shown in the figure. Details of each part are given below.• Yankee : A cast iron shell of 16 in. diameter. Heating is provided internally.32Figure 3.1: Creping rig with high speed cross-section imaging setup• Transfer roll : Green coloured roll above the yankee. It is used to transfer the wet sheet toyankee.• Sheet moisture control system (not shown in figure) : A mechantronics system used to setthe moisture content of sheet at desired level.• Adhesive spray (not shown in figure) : Sprays controlled amount of adhesive to yankeesurface.• Instrumented crepe blade : Section of a standard crepe blade with a 3D force transducermounted on a fixture.The yankee keeps dwelling at very low (2 rpm) speed to maintain a uniform surface tem-perature. The temperature can be varied from room temperature to an excess of 130oC. Atthe beginning of each experiment, an uncreped sheet is applied to the transfer roll and broughtto the desired moisture content. The sheet length is about 27 in. A preset amount of coatingpackage (a mixture of adhesive, release agent, modifier and plasticizer) is sprayed onto the yan-kee drum simultaneously. The transfer roll presses and transfers the sheet on the yankee aftera preset amount of curing time. At this stage the yankee starts to rotate at higher speed and33when a desired yankee speed is reached, the creping blade is engaged to crepe the sheet fromyankee surface. The device is capable of speeds up to 1524 meter/min and can be used withany basis weight sheet from low to high. A 25 gsm paper sheet of 63.9 µm caliper was crepedat 1120 m/min for the experiment. Creping angle was set to 80o. A line laser was used to il-luminate a cross-section of tissue very close to right edge. The imaging area is shown in Fig.3.1. A high-speed camera was used to capture the creping mechanism at 67000 frames per sec-ond. The camera was triggered by the force sensor attached to the blade to capture the crepingmechanism.3.1.2 Creping MechanismFigure 3.2: Four stages of creping fold formation in the high speed creping rig34Fig. 3.2 shows four images corresponding to four stages of fold formation captured in theexperiment. In the first stage, paper sheet hits the blade edge as shown in 1. The sheet bucklesand forms a fold in stages 2 and 3. Then the fold gets pushed off the edge and the process repeats.This mechanism is similar to the well established mechanism found in literature [5, 7, 9, 10] .However, there are some limitations. Imaging of the tissue was done at edge of the paper. It waslater observed that the crepe folds towards the edges were larger in size compared to the crepefolds in the middle. Therefore the folding observed may not be exactly representative of foldingin midsection.3.2 Online Crepe Structure QuantificationThe surface image quantification technique developed in Chapter 2 is integrated with a highspeed static imaging system to measure crepe count online in a real tissue machine. Imagesof crepe tissue surface were captured at regular intervals for 3 hours and analyzed online usingthe surface image analysis technique developed in section 2.3.2. Tissue samples were collectedduring the experiment for offline crepe count analysis. Crepe count measurement from bothonline and offline imaging were compared to verify the accuracy of quantification based ononline images.3.2.1 Experimental SetupThe experiments were carried out in a paper machine of a tissue manufacturing plant. Fig. 3.3(a)shows the high speed imaging set-up placed between the yankee drum and reeling section of apaper machine inside the plant. Images of tissue were taken under a sheet stabilizer to avoidsheet-fluttering. Fig. 3.3(b) shows the components of the high speed imaging setup. It has threemain components: a commercial Digital Single Lens Reflex (DSLR) camera (Nikon D 7000),a high speed Light Emitting Diode (LED) flash (Vela one) and a small table fan placed insidea box. The high speed flash is synchronized with camera via a wireless triggering system. Theflash can illuminate the tissue surface with a high intensity pulse of light for 1 microsecond.This short intense burst of light limits sheet movement to about 15 microns for a tissue machineoperating at 1000 m/min.High speed camera was not used for this type of imaging. The dusty, steamy environmentaround the imaging area makes it impossible. Further, the narrow space available for imagingis very small to accommodate the intense diffuse light and wires needed for the high speedcamera to work. The entire system with DSLR camera is compact and capable of wirelesslytransmitting the captured images to a laptop placed outside the tissue machine. Whole box iscovered except the top of lens to protect the equipment from dust and humidity. A small fan isused for circulation of air to prevent moisture condensation and dust buildup as shown in Fig.3.3(b).35Figure 3.3: Online crepe count quantification setup in a real tissue machine (left). Components of thehigh speed imaging setup (right).3.2.2 ResultFig. 3.4 shows the variation of crepe count with time, over a 3 hour window. Several onlineimages were taken at regular interval of approximately 15 minutes. They were analyzed usingsurface image quantification method explained in section 2.3.2. Mean dominant crepe countsextracted from those images are plotted against time in blue circles. The errorbars at each datapoint represents the standard deviation in crepe frequency of each image. The variability comesfrom the spatial variation of crepe fold length scale in one image, which is a property of theimage and not a measurement error. The measured mean dominant crepe counts are repeatable(see the close cluster of mean dominant crepe counts at different times intervals). It can beseen that the dominant crepe count is around 90 folds/in., which is typical for a commercialgrade tissue. Several sample of tissue were collected at different times during the experimentfor offline analysis. They were analyzed offline using the surface imaging setup described insection 2.3.1. Offline results are plotted at the time they were collected in the experiment in redsquares on the same plot. It can be observed that the crepe count obtained offline are similar tomean crepe count measured online. It is not realistic to expect exactly similar numbers becausethe images analyzed are not taken exactly at same area of tissue, but probably several thousandmeters away from each other. Also during online imaging, the sheet is under tension for reel-ing. Exact value of this tension is unknown and may change the crepe structure slightly when360 50 100 150 200Time (min.)6080100120Crepe Count (folds/in.) OnlineOfflineFigure 3.4: Comparison of online and offline mean dominant crepe count, and standard deviation withtime. The standard deviation at each data point is shown as a two sided vertical error bar. This varia-tion represents the spatial variability of crepe fold length scale in one surface image, and not related tomeasurement accuracy.detached from tissue machine and reeled. With these factors in consideration, the agreement iswithin the statistical dispersion. This leads us to believe that the surface imaging technique iscapable of quantifying crepe count from images taken at high speed.3.3 ConclusionHigh speed video and static imaging techniques are used to visualize and quantify creping inthis chapter. Significant conclusions are as follows.1. For the first time, mechanism of creping is observed through the thickness of a tissuepaper at a speed over 1000 m/min in a realistic creping rig. However The mechanismis observed at the edge of tissue paper, which is generally loosely bonded to the yankeedrum. The same technique can be used in future to observe creping at the middle portionof the tissue.2. The laboratory scale surface imaging technique (see Chapter 2) is successfully integratedwith the high speed flash to measure crepe count during production in a real tissue ma-chine. The potential of this device to work as an inexpensive online process monitoringtool is significant.37Chapter 4Tensile Response of a Creped Tissue4.1 IntroductionThe tensile testing of tissue papers is ubiquitous among tissue manufacturers, who use it forquality control. For manufacturers, the strength and stretch of tissue are important parameterswhereas they pay little heed to the shape of the tensile curve. Sufficiently high dry and wet ten-sile strength along the machine and cross direction are crucial to ensure tissue machine runnabil-ity and successful conversion or embossing operations. High stretch and low elastic modulusalong machine direction is important to obtain a softer (bulk softness) tissue. However, there is atrade-off between strength and softness. Tissues with higher softness tend to be weaker. There-fore, to create a sufficiently strong tissue structure with high softness, a basic understanding ofthe deformation mechanism and failure mode of tissue fiber network is needed. In section 4.2 ofthis chapter, stress strain response of low density commercial grade tissue papers (Grade 3-6) isdetermined along both machine and cross direction. Tensile strength, breaking stretch and initialYoung’s modulus are quantified according to existing standard [2] from the stress strain data.In another experiment, the fiber network is imaged under a microscope during tensile testing toobserve the deformation of the network structure and failure mode in section 4.3. Crepe foldsstructures in tissue are believed to decrease stiffness of paper materials along machine direction.Investigation of crepe fold deformation with machine direction tensile loading can give insightinto the network stiffness at low strain, which is related to bulk softness of tissues. Therefore,evolution of crepe fold structure with machine direction tensile load is quantified in section 4.4.Fiber-fiber bond strength is known to govern tensile strength of paper [23]. It has been reportedthat large local strains are precursors to bond failure [24]. In low density paper like tissue, largestrain localization can occur due to various reasons like sparse fiber network, and poor forma-tion. Analysis of principal strain field at various global network strain is done along machinedirection to quantify the uniformity and observe the degree of strain localization in Tensile Characterization4.2.1 Experimental SetupThe experimental setup to determine in-plane uniaxial tensile stress strain behaviour of tissuesamples is shown in Fig. 4.1 (a). An INSTRON 5969 universal testing machine is used in tensile38Figure 4.1: (a) Experimental set up for tensile test of tissue papers. (b) Dimension of tested tissuesamples. Tests are done according to ISO: 12625-4 protocol [2].testing mode to do the experiments. The machine is equipped with two pneumatic pressure jawsto hold the tissue samples firmly while applying tensile strain. A rubber coating is used on theinner face of each jaw to avoid any slippage during the test. The bottom jaw is fixed to the baseof the machine while the top jaw can move upwards at a constant velocity (1-100 mm/min) tostrain the samples. An encoder is used to measure the upper jaw displacement. A 500 N loadcell (0.5% accuracy at 0.5 N) is attached to the top jaw to measure load. Both displacement andload are measured and recorded simultaneously in real time for postprocessing.Tensile response of paper products can vary with dimensions of the sample [27], strain rate,sample defects, formation, moisture content and temperature [28]. Humidity and temperatureof the testing environment can also affect the results. Consistency of all these parameters is ofimmense importance for repeatability and consistency of each test. ISO 12625-4 tensile testprotocol for tissue papers is followed for these tests [2].15 samples along both machine and cross direction, each 150 mm long and 25.4 mm wide,were prepared beforehand for tensile tests. The width of the samples were not 50 mm as pre-scribed in the protocol because the jaws of the INSTRON machine were only 25.4 mm wide.39However, this change was acceptable as per protocol as long as mentioned in the report. Asingle edge paper cutter was used to cut the samples from parent roll. For preconditioning, thesamples were kept at the testing environment at 23◦C and 50% relative humidity for 48 hoursbefore testing. Samples were examined to eliminate any specimen with obvious defects in thetesting area. Though 150 mm long samples were prepared, only 100 mm was used as gaugelength. 25 mm from top and bottom of the sample was used to grip the samples by pneumaticjaws. To eliminate any slack in the tissue, each sample was preloaded with 5 N/mm force. Thesamples were elongated at a uniform extension rate of 50 mm/min. At this extension rate, theinertial effects on the load extension plots are negligible. 10 specimen from each grade weretested for statistical robustness.4.2.2 ResultsFigure 4.2: Load extension (a) and corresponding stress strain plots (b) for multiple samples of Grade3-6 along machine directionTable 4.1: Analysis results of the machine direction load extension and stress strain plots shown in Fig.4.2Name Stretch at break Tensile strength Initial modulus Tensile energy absorption(%) (N/m) (MPa) (J/m2)Grade 3 18.32 ± 2.02 97.95 ± 7.05 15.81 ± 1.33 10.17 ± 1.49Grade 4 23.89 ± 0.88 107.14 ± 9.11 6.13 ± 0.27 12.34 ± 1.13Grade 5 21.14 ± 1.13 111.76 ± 11.56 10.92 ± 0.47 11.62 ± 1.44Grade 6 22.72 ± 0.81 85.4 ± 6.07 4.32 ± 0.14 9.26 ± 0.81Fig. 4.2 shows the load extension (left) and stress strain (right) plots for Grade 3-6 along40Figure 4.3: Load extension (a) and corresponding stress strain plots (b) for multiple samples of Grade3-6 along cross directionTable 4.2: Analysis results of the cross direction load extension and stress strain plots shown in Fig. 4.3Name Stretch at break Tensile strength Initial modulus Tensile energy absorption(%) (N/m) (MPa) (J/m2)Grade 3 4.2 ± 0.69 38.89 ± 4.13 22.52 ± 1.81 1.14 ± 0.25Grade 4 4.02 ± 0.81 37.99 ± 5.09 14.42 ± 1.87 1.06 ± 0.31Grade 5 4.36 ± 0.85 30.08 ± 3.5 13.95 ± 1.22 0.94 ± 0.3Grade 6 5.5 ± 0.7 23.74 ± 1.33 7.04 ± 0.24 0.96 ± 0.18machine direction. Load at zero extension is positive because of preloading to eliminate slack inthe samples. Stress strain plots are calculated from the load extension data. Stress is calculatedas load per unit cross-sectional area of the undeformed sample. Cross section area of unde-formed sample is calculated from the caliper and width of tissues. Strain is the extension of thesample divided by the gauge length. Tensile behaviour of tissue papers shows significant non-linearity along machine direction for all grades. Along machine direction, the tensile responsehas three stages of nonlinearity as seen in Fig. 4.2 (a) and (b). It begins with a initial elasticregion (approximately before 2% strain) followed by a strain-softening region (approximately7% - 12% strain) and finally a hardening region towards the end before failure. Explanation ofthe exact reasons of these transitions is out of scope of this thesis. However, a partial answer willbe given in subsequent sections. All grades show a variation in breaking strength and breakingstretch close to the end of the test. This is believed to be a result of the inherent sparse natureof the fiber network and fiber fiber bonds, which govern the strength of the tissue. Grades withuniform formation and good quality fibers should show less dispersion with that argument andgrades with poor formation and low quality fibers should perform poorly in terms of repeatabil-41ity. From Fig. 4.2, it can be seen that Grade 3 has more dispersion in stretch at break and overall shape compared to other grades. It was confirmed from the relevant industry that Grade 3was indeed a low quality tissue grade. The shape of stress strain plots along machine directionare fairly consistent and repeatable within other grades.Load extension and stress strain data are analyzed to extract tensile strength (N/m), stretchat break (%), initial modulus (MPa) and tensile energy absorption (J/m2). Tensile strengthis the highest load carried by the sample per unit width. This load occurs generally just beforethe onset of fracture of samples. The higher the tensile strength, the stronger is the tissue. Thestrain at the breaking load is known as stretch at break. Tensile energy absorption is the workneeded to break the sample per unit area of the undeformed sample. It is the area under theload extension plot until failure divided by the undeformed area of the sample. Initial modulusis calculated as the slope of the stress strain plot between 0% and 2% strain. The mean andstandard deviation of these parameters for all grades are tabulated in Table 4.1. From Table 4.1,it can be seen that Grade 3 shows higher variation in all the four parameters compared to othergrades, which reiterates the fact that formation and quality of fibers are important to produce ahigh quality and consistent tissue.Similar tensile characterization is done along cross direction. The load extension and stressstrain plots are shown in Fig. 4.3 (a) and (b). The analysis results are tabulated in Table 4.1.In cross direction, the tissue paper shows monotonic progressive softening behaviour with load(see Fig. 4.3) with more visible dispersion in the entire shape for all Grades, except possiblyfor Grade 6. This shape is entirely different than the machine direction tensile response, whichindicates that crepe folds govern the shape of the tensile response. Crepe folds do not contributein the tensile response along cross direction since they are aligned perpendicular to the direc-tion of applied load. Therefore, the shape of tensile response along cross direction is entirelygoverned by formation and furnish of tissue.Comparing Table 4.1 and 4.2, it can be concluded that stretch at break in machine directionis higher than cross direction for all 4 grades. The machine direction stretch is higher thancross-direction stretch because creping compresses and packs more material as series of foldsin machine direction. While load is applied, these folds are removed which gives rise to thehigh stretch. Also the tensile strength is higher along machine direction because of preferentialalignment of fibers along machine direction during formation. Because of high stretch at breakand strength, the tensile energy absorption is also higher along machine direction. However, theinitial modulus in machine direction is lower than cross direction. Creping folds are believedto cause this deterioration of modulus along machine direction. The role of creping folds onthe overall physics of tensile deformation, specially in the low strain region will be explored insection Visualization of Fiber Network Deformation4.3.1 Experimental SetupFigure 4.4: Microtensile stage (DEBEN Microtest 200N Tensile Tester; Serial number : MT10129) fortensile test of tissue paper outside SEM chamberFurther experiments are done to observe the deformation and failure mechanism of the fibernetwork under machine direction load. A microtensile tester (DEBEN Microtest 200N TensileTester; Serial number : MT10129) with in situ SEM imaging capability is used for this experi-ment. Fig. 4.4 shows the microtensile stage outside SEM chamber with attached tissue sampleready to test. Tissue samples were cut from parent roll using a single edge blade and no pre-conditioning was done in this experiment. Each sample had a gauge length of 10 mm alongmachine direction and 15 mm width along cross direction. After carefully attaching the sampleto the device, the entire setup was placed inside an SEM chamber and the sample was elongatedat 2.37 mm/min (see AppendixA for calibration details) until failure. Entire load-extension plotwas recorded with a video of the tissue sample until failure. About 1 mm2 area of the samplewas visualized to observe the deformation.4.3.2 Results and ObservationsFig. 4.5 - 4.8 shows the deformation of tissue’s fiber network with load for all grades. Images at3 different stress levels ((b)-(d)) along with post failure images (e) of tissue samples are shown43Figure 4.5: Microscopic deformation of tissue fiber network under tensile load for Grade 3. Tissuemicrostructure at approximately end of elastic region (c), at half of breaking strain (c), at breaking strain(d) and after network failure (e). Approximate locations of these pictures on stress strain plot are shownin top.44Figure 4.6: Microscopic deformation of tissue fiber network under tensile load for Grade 4. Tissuemicrostructures at approximately end of elastic region (b), at half of breaking strain (c), at breaking strain(d) and after network failure (e) are shown. Approximate locations of these pictures on stress strain plotare shown in top.45Figure 4.7: Microscopic deformation of tissue fiber network under tensile load for Grade 5. Tissuemicrostructures at approximately end of elastic region (b), at half of breaking strain (c), at breaking strain(d) and after network failure are shown. Approximate locations of these pictures on stress strain plot areshown in top.46Figure 4.8: Microscopic deformation of tissue fiber network under tensile load for Grade 6. Tissuemicrostructures at approximately end of elastic region (b), at half of breaking strain (c), at breaking strain(d) and after network failure (e) are shown. Approximate locations of these pictures on stress strain plotare shown in top.47for all grades. Approximate state of stress and stain of the entire tissue sample for each imageis marked in the stress-strain plot above. There stress strain plot from the microtensile stage iscompared with the family of characteristic tensile response (obtained from INSTRON as shownin Fig. 4.2) of each grade in Fig. 4.5 (a) - 4.8 (a) first and the locations of the images aremarked on the microtensile stage data. Stress strain plots from the microtensile stage are shownfor all grades, except for Grade 4 (Fig. 4.6 (a)) in which no data was obtained from the deviceunfortunately. However, approximate locations of images are shown on the INSTRON stressstrain plot in this case. Though the testing environment inside the SEM (ambient temperatureand 5% relative humidity), sample size and strain rate were different in these experiments, thestretch, strength and shape of tensile curve were similar to the characteristic tensile responseuntil fracture, confirming similar mechanism of deformation. On careful observation, one mightsee that the stress strain data obtained from the microtensile stage is very noisy. This is becauseof the inferior load cell accuracy of the microtensile stage compared to the load cell attached tothe INSTRON machine.Though the shapes of the curves are similar, the post failure behaviour is significantly dif-ferent. The drop in load upon fracture is much abrupt in INSTRON than the microtensile stage.This is explained by the strain rate in each experiment. In INSTRON, the tissue samples werestrained at a strain rate of 50% / min (50 mm/min extension rate at 100 mm gauge length). In themicrotensile stage the samples were strained at a much lower strain rate of 23.7% / min (2.37mm/min extension rate at 10 mm gauge length). Because of a lower strain rate, the fracturepropagated slowly showing a less abrupt load drop after failure.Images obtained from the SEM can show the quality of formation of each grade. Grade 3has a more porous structure compared to other grades as seen in Fig. 4.5 (b) as a result of poorformation. This is also confirmed by the industry and discussed in Section 4.2.2 before. Theseholes and pores in tissue network can deform to a large extent to accommodate large strainaround them. Evidence for this phenomenon is shown in Fig. 4.5 (b) - (d) and also in 4.8 (b) -(d) for another grade. The indicated holes (surrounded by white dashed lines) in those pictureschange shape and size until failure to distribute the strain concentration around them. Other thanthis, curled and kinked fibers are shown to be straightened out with load in Fig. 4.6 (b) - (d)and 4.7 (b) - (d). Some fibers are indicated as white dashed lines. Based on these observationsof the tissue deformation, it can be concluded that the fiber network can undergo significantrearrangement of to accommodate large strain. It can be seen that the flexible fiber fiber bondsrotate and slide to rearrange the network and also the curled fibers are straightened towards theend of the tensile curve, but prior to fracture. Straight fibers are much stiffer than curled fibers,and this stiffens the entire network before fracture. This argument qualitatively explains thehardening of the tissue before failure in machine direction. Also it is observed that very fewfiber fiber bonds rupture during the straining process. However at failure, the inter fiber bondsfail together abruptly, resulting in the failure of structure. Pulled out unbroken fibers can be seenclearly in the post-fracture images of Fig. 4.5 (e). Creping folds are not distinguishable in anyof these images and how these folds behave under tensile load is not clear from this experiment.484.4 Structural Evolution with Load4.4.1 Experimental SetupTo observe the evolution of crepe folds, one tissue specimen from each grade was loaded inthe machine direction in small load-steps (approximately 0.1 N) and imaged at each step withoblique diffuse light to highlight the crepe structure. The experimental setup for surface imag-ing described in section 2.3.1 is used to image tissue surface at each load step. The pulley andweight system of the setup is used to load the samples using calibrated weights. The loading isdone slowly to avoid any sudden impact-load on the structure. Once loaded, the tissue is allowedto relax for 30 seconds to reach the maximum elongation, at which point the surface images aretaken. Length and width of each tissue sample specimen are 100 mm and 25.4 mm respec-tively. These tissue samples are not preconditioned in a constant temperature and humidityenvironment, and the experiments are done at ambient temperature and humidity. Stress strainplots obtained from these experiments are compared with the characteristic tensile response tovalidate each test.4.4.2 Evolution of Crepe Structure and Local Strain FieldTensile evolution of crepe folds and local strain field are shown in Fig. 4.9-4.12. Each figurecorresponds to a separate grade. Stress strain plots were constructed for all grades from theapplied load and calculated global extension. Graph (a) in each figure shows the stress-straindata (discrete loading since weights were applied in steps) obtained from the experiment super-imposed on the characteristic stress-strain plot obtained from previous tensile characterizationexperiment in section 4.2.2. In Fig. 4.11, the breaking strain in discrete loading experiment forGrade 5 is almost half of the characteristic breaking strain. This indicates that the sample brokeprematurely, which invalidates the experiment. The stress strain plot obtained from discreteloading for other grades reach similar breaking stress and strain as the characteristic response,but shows a systematic deviation. More specifically, the strain in discrete loading experiment,is higher than the characteristic strain at a given load. This happens because tissue paper is aviscoelastic material [29, 30] and slowly stretches with time if left under a constant load. Inthese experiments, photo of the tissue surface was taken at least 30 seconds after the load wasapplied. This time difference caused the tissue to elongate more and give a higher strain thanexpected. With this in mind, the agreement between the stress strain plots is satisfactory. 3 datapoints ((b) - (d)) are highlighted in each these plots in Fig. 4.9 (a) - Fig. 4.12 (a) and the analysisresults of the corresponding surface images are shown below.At each step of (b) - (d), a 2D Fourier Transformation of the surface image (left) and theprincipal strain field (right) are shown. Fourier Transformation of the reflection images showthe frequency distributions of the crepe fold structure and wire mark patterns along machineand cross directions. These are indicated in each figure. 2D FFT of the surface image at low49Figure 4.9: Evolution of crepe structure in 2D Fourier space (bottom left) and local principal strain field(bottom right) with increasing machine direction tensile load approximately at the end of elastic region(b), half of breaking strain (c) and at breaking strain (d) for Grade 3. Global stress and strain associatedwith each step are shown in the stress strain plot above.50Figure 4.10: Evolution of crepe structure in 2D Fourier space (bottom left) and local principal strain field(bottom right) with increasing machine direction tensile load approximately at the end of elastic region(b), half of breaking strain (c) and at breaking strain (d) for Grade 4. Global stress and strain associatedwith each step are shown in the stress strain plot above.51Figure 4.11: Evolution of crepe structure in 2D Fourier space (bottom left) and local principal strain field(bottom right) with increasing machine direction tensile load approximately at the end of elastic region(b), half of breaking strain (c) and at breaking strain (d) for Grade 5. Global stress and strain associatedwith each step are shown in the stress strain plot above.52Figure 4.12: Evolution of crepe structure in 2D Fourier space (bottom left) and local principal strain field(bottom right) with increasing machine direction tensile load approximately at the end of elastic region(b), half of breaking strain (c) and at breaking strain (d) for Grade 6. Global stress and strain associatedwith each step are shown in the stress strain plot above.53tensile load (image (b) of each figure) shows a cloud of frequency components aligned alongmachine direction with some other distinct frequency components spread across machine andcross direction frequency space. These distinct frequency components appear from the formingfabric induced wire mark pattern embedded in the tissue whereas the cloud of frequency alongmachine direction represents creping fold pattern (indicated in Fig. 4.9 (b) - Fig. 4.12 (b)).Observation of the 2D FFT just before failure (image (d) of each figure) shows that the crepefold induced frequencies have almost disappeared, but the wire mark patterns have not. Thisis not very clear for grade 6 (see Fig. 4.12 (b) - (d)). Even the creping fold cloud is not wellvisible there. The reason is: creping folds appear almost in phase with the wire marks forGrade 6, hence the creping frequency is not well separated from the wire-mark patterns andthe disappearance of the creping folds seems unclear. However, based on Grade 3-5, this studyshows that the creping folds are unfolded during the tensile test governing the shape of thecurve until failure. But close to failure, wire mark patterns or formation become an importantstructural feature because of their distinct presence.The local strain field information at each step is extracted using two dimensional DigitalImage Correlation (DIC). Digital Image Correlation is an algorithm to extract displacement andstrain information at multiple points of an undeformed reference image as it is progressivelydeformed. A number of images of the area of interest with progressive deformation is needed.In this technique, first, the reference image is subdivided into many smaller regions (subsets)separated by some distance (subset spacing). Then the subsets are tracked down in the followingdeformed images based on structural similarity to extract displacements. The structural similar-ity generally comes from an unique speckle pattern, which is generally spray painted externallyon the surface of interest. In this case, there was no additional need to spray speckle patterns onthe surface. Rather the corrugated tissue surface under reflective light had enough salt and pep-per patterns for the optical tracking. An opensource 2D DIC MATLAB software called Ncorr isused to perform the DIC computation. Relevant details about the DIC calculation can be foundin Table 4.3.Table 4.3: Important parameters of Digital Image Correlation analysisImage pixel spatial resolution 8.62 µm / pixelSubset radius 50 pixel = 431.2 µmSubset spacing 10 pixel = 86.2 µmStrain radius 3Green-Lagrangian principal strain fields are obtained from the displacement fields on thetissue surface at three stages of the stress strain plot. The strain fields are plotted on the unde-formed tissue surface (Fig. 4.9 (d) right Fig. 4.12 (d) right). Analysis of principal strain fieldsshows significant high strain localization (notice the red spots in the strain field) just beforefailure. Interestingly, the strain localization areas form distinct bands approximately parallel to54cross direction for all grades. The strain localization is not very distinct in Grade 5 because thesample failed prematurely before reaching the ultimate stress. Length scale of strain localizationareas in all grades is in millimeter length scale which is larger than the length scale of crepingfolds, indicating that the reason is related to formation of the tissue. It is well known that forma-tion of tissues change over a length scale of several millimeters. Also the regions of high strainlocalization start to appear from the beginning of the tensile curve, during elastic deformation(carefully notice the similar high strain pattern in (d) right and see similar bands in (b) right inall grades). These indicate that the strain localization is not the result of plastic damage entirely,but also related to the network deformation in the elastic regime, which depends on formation.4.5 ConclusionThe experiments have provided insight into the stress-strain behaviour of creped tissue materials,especially along machine direction. The conclusions can be summarized as follows.1. Characterization of tensile behaviour along machine direction shows initial linear elasticregime followed by a strain softening and a strain hardening before failure. This type ofresponse is not seen in the cross-direction, which indicates that the three stage nonlinearresponse along machine direction is governed by the creping structures.2. The tissue fiber network deforms significantly to accommodate large strain before fail-ure when load is applied along machine direction. During this deformation, the inter-fiber bonds rotate and slide. This network deformation also straightens the plasticallydeformed, curled fibers. Straightening of fibers stiffens the tissue structure, which ex-plains the hardening of tissue before failure in machine direction. The mechanism offailure is understood as a avalanche of bond disruption and energy release upon reachingthe breaking load. However, the mechanism of deformation is not investigated along crossdirection.3. When a tensile load is applied along machine direction, the crepe folds gradually unfoldand disappear, but the forming fabric induced wire mark patterns do not. Unfolding of thecrepe folds explains the origin of high stretch of creped tissues along machine direction. Itis understood that creping folds govern the nonlinear shape of the machine direction ten-sile response. However, wire mark patterns become important close to failure. Thereforeit is possible that wire mark patterns affect the tensile strength.4. Principal strain fields immediately before tensile break show significant strain localizationat a length scale much larger than the crepe fold wavelength. The strain localization isrelated to the formation of creped tissue and distribution of creping folds.55Chapter 5Conclusions and Future Work5.1 Conclusions and Major FindingsMajor achievements and findings of this thesis are summarized below.1. A robust and simple surface imaging technique to quantify tissue creping structure isdeveloped and demonstrated on a commercial tissue machine at high speeds under adverseoperating conditions.2. It is shown that the surface image based quantification of crepe structure is as accurateas cross-section imaging using, for example, SEM. The added advantages are simplicityand easier possibility of statistical analysis over a larger tissue sample for a more reliablequantification.3. Mechanism of creping is observed at a speed over 1000 m/min on a realistic creping rigusing a line laser and high speed camera. The through-thickness visualization of crepingprocess is believed to be obtained for the first time, to our knowledge.4. Uniaxial tensile response of tissues along machine direction is characterized by 2 transi-tions from linear elastic to strain softening and from strain softening to hardening beforefailure. No such transition exists in the cross-directional tensile response. Hence, defor-mation of creping folds are important to understand tensile properties of a tissue paper.5. Microscopic, qualitative insights about tissue structure deformation are gained from ma-chine direction tensile tests under an SEM. Tissue structure shows an exceptional capa-bility of fiber fiber bond rotation and arrangement to accommodate the large global strainuntil failure. However, the role of crepe structures remain unclear from the qualitativeobservations.6. Surface imaging techniques are used to understand the influence of crepe structure onthe tensile response at macroscopic length scales. Creping folds are shown to governthe Young’s modulus of tissue. However, formation and wire mark patterns induced byforming fabric are dominant at failure. A uniform formation is important for a moreconsistent tissue product.567. Advancements in the experimental techniques made in this work are expected to be ofinterest to tissue manufacturers and engineers alike.5.2 Limitations and Future WorkThis work contributes to the development of experimental tools to quantify crepe structure andto understand the deformation of a tissue under load. However, many of the directions emergingfrom the study could not be pursued in this work and can be explored further. They have beendescribed briefly in the sections below.1. The crepe structure quantification techniques developed so far are limited to the extractionof creping wavelengths, but, out of plane amplitude information is also important. Furtherdevelopment, using photometric stereo methods for example, is needed to simultaneouslymeasure creping wavelength and amplitude directly from surface images. Photometricstereo method works by capturing surface images under reflective light conditions frommultiple angles. A camera with multiple colour sensors can be used for this purpose.2. The proof-of-the-concept apparatus developed for online imaging can be made more ro-bust to perform better under adverse environmental conditions, through careful engineer-ing design. It is possible for this set-up to traverse the surface of the tissue by to and fromovement along cross direction of the web on a tissue machine.3. Much work remains to be done to understand the shape of nonlinear tensile response,specifically along the machine direction. Here, a combined numerical and experimentaltechniques have to be pursued.4. The importance of forming fabric induced wire mark patterns remains to be fully un-derstood. Our results clearly point to the influence of wire mark patterns on the tensileresponse. Such pre-patterned base sheets before creping are also common to base sheetsproduced by Through-Air-Drying (TAD). A systematic study of different wire mark pat-terned base sheet before and after creping can be pursued in future.5. Blade wear is known to play an important role during creping. A systematic study oftemporal variation of the crepe structure and its distribution can be done to understand theimpact of blade wear on the physics of creping.57Bibliography[1] K. Pan, R. Das, A. S. Phani, and S. Green, “An elastoplastic creping model for tissuemanufacturing,” International Journal of Solid and Structures, no. IJSS-D-18-00836,Under review.[2] PN-EN ISO 12625-4:2016 Tissue paper and tissue products – Part 4: Determination oftensile strength, stretch at maximum force and tensile energy absorption. 2016.[3] M. Ramasubramanian, “Physical and mechanical properties of towel and tissue,” inHandbook of Physical Testing of Paper, pp. 683–896, CRC Press, 2001.[4] H. Hollmark, “Mechanical properties of tissue,” vol. 1, pp. 497–521, 1983.[5] H. Hollmark, “Study of the creping process on an experimental paper machine,”STFI-meddelande, Serie B, no. 144, 1972.[6] J. F. Oliver, “Dry-creping of tissue paper-a review of basic factors,” Tappi, vol. 63, no. 12,pp. 91–95, 1980.[7] K. Pan, A. S. Phani, and S. Green, “Particle dynamics modeling of the creping process intissue making,” Journal of Manufacturing Science and Engineering, vol. 140, no. 7,p. 071003, 2018.[8] W. McConnel, “The science of creping,” Tissue World Americas, Miami Beach, FL, 2004.[9] M. K. Ramasubramanian and D. L. Shmagin, “An experimental investigation of thecreping process in low-density paper manufacturing,” Journal of Manufacturing Scienceand Engineering, vol. 122, no. 3, pp. 576–581, 2000.[10] M. K. Ramasubramanian, Z. Sun, and G. Chen, “A mechanics of materials model for thecreping process,” Journal of Manufacturing Science and Engineering, vol. 133, no. 5,p. 051011, 2011.[11] S. S. Gupta, “Study of delamination and buckling of paper during the creping processusing finite element method-a cohesive element approach.,” Ph.D. Thesis, North CarolinaState University, Raleigh, NC, 2013.58[12] D. M. Rasch, T. A. Hensler, and D. J. Daniels, “A papermaking belt,” July 11 1995. USPatent 5,431,786.[13] H. Ihalainen, K. Marjanen, M. Ma¨ntyla¨, and M. Kosonen, “Developments in camerabased on-line measurement of paper,” in PaperCon 2012 Growing the Future, ControlSystems 2012 , April 22-25, New Orleans, Louisiana, USA, PaperCon, TAPPI, 2012.[14] C. Crosby, A. Eusufzai, R. Mark, R. Perkins, J. Chang, and N. Uplekar, “A digitizingsystem for quantitative measurement of structural parameters in paper,” Tappi, vol. 64,no. 3, pp. 103–106, 1981.[15] J. Sabater, J. C. Kerneis, and S. Bauduin, “Device for the continuous determination of asurface state index for a moving creped sheet,” Dec. 18 1990. US Patent 4,978,861.[16] W. A. Von Drasek, S. L. Archer, and G. S. Furman Jr, “Method and apparatus to monitorand control sheet characteristics on a creping process,” Dec. 26 2017. US Patent9,851,199.[17] J. P. Raunio, “Quality characterization of tissue and newsprint paper based on imagemeasurements; possibilities of on-line imaging,” Thesis for the degree of Doctor ofScience in Technology, Tampere University of Technology, 2014.[18] M. Kelloma¨ki and A. Paavola, “Apparatus and method for measuring caliper of crepedtissue paper based on a dominant frequency of the paper and a standard deviation ofdiffusely reflected light including identifying a caliper measurement by using the imageof the paper,” Apr. 5 2016. US Patent 9,303,977.[19] H. L. Cox, “The elasticity and strength of paper and other fibrous materials,” BritishJournal of Applied Physics, vol. 3, no. 3, pp. 72–79, 1952.[20] D. H. Page, R. S. Seth, and J. H. De Grace, “The elastic modulus of paper. i. thecontrolling mechanisms,” Tappi, vol. 62, no. 9, pp. 99–102, 1979.[21] D. H. Page and R. S. Seth, “The elastic modulus of paper. ii. the importance of fibermodulus, bonding, and fiber length,” Tappi, vol. 63, no. 6, pp. 113–116, 1980.[22] D. H. Page and R. S. Seth, “The elastic-modulus of paper. iii. the effects of dislocations,microcompressions, curl, crimps, and kinks,” Tappi, vol. 63, no. 10, pp. 99–102, 1980.[23] R. S. Seth and D. H. Page, “The stress-strain curve of paper,” The Role of FundamentalResearch in Paper Making, Trans. VIIth Fund. Res. Symp. Cambridge, vol. 1,pp. 421–452, 1981.[24] S. Borodulina, A. Kulachenko, M. Nyga˚rds, and S. Galland, “Stress-strain curve of paperrevisited,” Nordic Pulp & Paper Research Journal, vol. 27, no. 2, pp. 318–328, 2012.59[25] H. F. Rance, “The formulation of methods and objectives appropriate to the rheologicalstudy of paper,” Tappi, vol. 39, no. 2, pp. 104–115, 1956.[26] G. J. Williams and J. G. Drummond, “Preparation of large sections for the microscopicalstudy of paper structure,” Journal of Pulp and Paper Science, vol. 26, no. 5, pp. 188–193,2000.[27] A. Hagman and M. Nyga˚rds, “Investigation of sample-size effects on in-plane tensiletesting of paperboard,” Nordic Pulp & Paper Research Journal, vol. 27, no. 2,pp. 295–304, 2012.[28] D. Caulfield, “Effect of moisture and temperature on the mechanical properties of paper,”Solid Mechanics Advances in Paper Related Industries, pp. 50–62, 1990.[29] D. Roylance, P. McElroy, and F. McGarry, “Viscoelastic properties of paper,” FibreScience and Technology, vol. 13, no. 6, pp. 411–421, 1980.[30] T. Uesaka, K. Murakami, and R. Imamura, “Two-dimensional linear viscoelasticity ofpaper,” Wood Science and Technology, vol. 14, no. 2, pp. 131–141, 1980.60Appendix AMicrotensile Stage Extension RateCalibrationThe extension rate controller of the microtensile stage was not calibrated during the tensile tests.It was discovered later during data analysis and a calibration factor had to be multiplied withthe apparent extension rate to correct it. The calibration factor was calculated by measuring theactual relative rate of separation of the jaws while the apparent extension rate was kept fixed at1 mm/min from the software. The relative jaw separation was optically tracked with time forover 3 minutes. The initial and final configurations of the microtensile stage are shown in Fig.A.1 (left). A series of images were taken in between and analyzed to extract the jaw separationwith time; the rate was calculated from the data (see A.1 (right)). The actual extension rate wasfound to be 1.589 mm/min. Therefore the extension rate calibration factor was 1.589.During the previous tensile tests , the apparent extension rate was set to be 1.5 mm/minfrom the software. Therefore, the actual extension rate was 2.37 mm/min (multiply 1.5 with the1.589).Figure A.1: Microtensile stage extension rate calibration methodology. A series of image is taken atregular intervals between the initial (left top) and final (left bottom) configuration of the stage. Therelative distance between the jaws is plotted against time (see right). The extension rate is calculated fromthe slope of the linear fit line.61


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items