UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Development and modeling of a novel particle mixture for the core configuration of particulate wood composites Sackey, Emmanuel Kuuku 2010

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

Item Metadata

Download

Media
24-ubc_2010_spring_sackey_emmanuel.pdf [ 3.38MB ]
Metadata
JSON: 24-1.0069234.json
JSON-LD: 24-1.0069234-ld.json
RDF/XML (Pretty): 24-1.0069234-rdf.xml
RDF/JSON: 24-1.0069234-rdf.json
Turtle: 24-1.0069234-turtle.txt
N-Triples: 24-1.0069234-rdf-ntriples.txt
Original Record: 24-1.0069234-source.json
Full Text
24-1.0069234-fulltext.txt
Citation
24-1.0069234.ris

Full Text

DEVELOPMENT AND MODELING OF A NOVEL PARTICLE MIXTURE FOR THE CORE CONFIGURATION OF PARTICULATE WOOD COMPOSITES  by EMMANUEL KUUKU SACKEY B.Sc., University of Science and Technology, 1989 Diplom Holzwirt, University of Hamburg, 2001 M.Sc., University of British Columbia, 2003  A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Forestry)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) February 2010  © Emmanuel Kuuku Sackey, 2010  Abstract Particleboard (PB) is the most utilized reconstituted wood composite panel for furniture production. However, edge smoothness and screw withdrawal resistance (SWR) have been limiting factors in some of its applications. In view of mounting wood and resin costs, this study focused on benchmarking the mechanical properties of Canadian-made furniture grade ReadyTo-Assemble MS and M2 particleboard. Of the five plants surveyed, over 80% of the 30 furniture grade particleboards tested were below the ANSI A208.1 standard for edge SWR of 900N. Particle geometry was examined by hydrolyzing samples from the commercial panels to disintegrate them into individual particles which were then characterized in terms of particle length, aspect ratio (AR), and slenderness ratio (SR). Based on maximum likelihood estimation a 2-parameter lognormal distribution function was found to be the best fit for particle length, AR, and SR followed by the 2-parameter Gamma distribution function. AR was found to be the best indicator of edge SWR. A novel particle mixture was formulated by repartitioning the core particles of commercial, as-received PB furnish into core-fine, medium, and coarse particles, which were then remixed in different proportions for the core furnish to fabricate low density panels: the IB strength was 40% higher than the control panels and the edge SWR 18% higher. A response surface model based on a mixture design was developed for macro-voids in the core of simulated particle mats and the macro-voids ratio in pressed panels was also found to increase exponentially with the void fraction in randomly packed loose particle mats. The study concluded that there are too many fine particulates and dust in the core of commercial PB. Increasing the amount of fines in the panel surface by decreasing fine and dust in the panel core through repartitioning the particles into fine, medium, and coarse has the potential of increasing surface smoothness, IB strength, SWR, and consequently reducing density and raw material cost. ii  Keywords: aspect ratio, gamma distribution, lognormal distribution, mixture design, particle, particle distribution, particle mixture, particleboard, random packing, response surface methodology, slenderness ratio, X-ray CT.  iii  Table of contents Abstract .......................................................................................................................................... ii Table of contents ............................................................................................................................ iv List of tables ................................................................................................................................. viii List of figures .................................................................................................................................. x List of symbols ............................................................................................................................. xiii Acknowledgements ....................................................................................................................... xv Dedication ................................................................................................................................... xvii Co-authorship statement ............................................................................................................ xviii Chapter 1 Project introduction ........................................................................................................ 1 1.1  Introduction ...................................................................................................................... 1  1.2  Rationale......................................................................................................................... 10  1.3  Objectives ....................................................................................................................... 12  1.4  Structure of dissertation ................................................................................................. 13  1.5  Literature cited ............................................................................................................... 14  Chapter 2 Properties variation study of furniture grade M2 particleboard manufactured in Canada1 .......................................................................................................................... 22 2.1  Introduction .................................................................................................................... 22  2.2  Materials and methods ................................................................................................... 25  2.2.1  Sample sourcing particleboards and sample preparation ........................................ 25  2.2.2  Experimental design and statistical analysis ........................................................... 27  2.2.3  Measurement of panel density ................................................................................ 29  2.2.4  Measurement of SWR ............................................................................................. 30  2.3  Results and discussion.................................................................................................... 31  2.3.1  Summary of effects ................................................................................................. 31  2.3.2  Physical properties: density, core density, MC, and IB .......................................... 32  2.3.3  Face SWR ............................................................................................................... 35  2.3.4  Edge SWR ............................................................................................................... 37  2.3.5  Flexural strength (MOR and MOE) ........................................................................ 40  2.4  Conclusions .................................................................................................................... 42  2.5  Literature cited ............................................................................................................... 43 iv  Chapter 3 Properties comparison of furniture grade MS and M2 particleboard products manufactured in Canada2 ............................................................................................... 46 3.1  Introduction .................................................................................................................... 46  3.2  Materials and methods ................................................................................................... 49  3.2.1  Sample procurement and preparation ..................................................................... 49  3.2.2  Experimental design and statistical analysis ........................................................... 50  3.2.3  Specimen testing ..................................................................................................... 52  3.3  Results and discussion.................................................................................................... 52  3.3.1  Main effects and interactions .................................................................................. 52  3.3.2  Moisture content, density, and IB strength ............................................................. 55  3.3.3  Face SWR ............................................................................................................... 57  3.3.4  Edge SWR ............................................................................................................... 58  3.3.5  Flexural strength (MOR and MOE) ........................................................................ 59  3.3.6  Vertical density profiles .......................................................................................... 61  3.4  Conclusions .................................................................................................................... 63  3.5  Literature cited ............................................................................................................... 66  Chapter 4 Empirical distribution models for slenderness and aspect ratios of the core particles of particulate wood composites3 ........................................................................................ 68 4.1  Introduction .................................................................................................................... 68  4.2  Experimental .................................................................................................................. 70  4.2.1  Particle preparation ................................................................................................. 70  4.2.2  Particle screening, classification, and dimensions .................................................. 71  4.2.3  Data analysis ........................................................................................................... 73  4.3  Results and discussion.................................................................................................... 75  4.3.1  Sieve screen weight data analysis ........................................................................... 75  4.3.2  Geometrical descriptors and their effect on panel strength .................................... 77  4.3.3  Relationship between properties (SWR and IB) and particle descriptors (AR and SR) of particleboard core ........................................................................................ 80  4.3.4  Particle geometrical descriptors and their distribution ........................................... 82  4.3.5  Goodness of fit tests ................................................................................................ 86  4.4  Conclusions .................................................................................................................... 88  4.5  Acknowledgement .......................................................................................................... 89  4.6  Literature cited ............................................................................................................... 90 v  Chapter 5 Improving core bond strength of particleboard through particle size redistribution4 .. 94 5.1  Introduction .................................................................................................................... 94  5.2  Materials and methods ................................................................................................... 97  5.2.1  Furnish preparation ................................................................................................. 97  5.2.2  Compilation of customized particle size mixes ...................................................... 97  5.2.3  Blending and mat formation ................................................................................... 98  5.2.4  Manufacture of uniform density boards .................................................................. 99  5.2.5  Manufacture of three layer boards ........................................................................ 100  5.2.6  Design of experiments .......................................................................................... 101  5.3  Results and discussion.................................................................................................. 103  5.3.1  Uniform density (UD) particleboards ................................................................... 103  5.3.1.1  Mean density and vertical density profile ......................................................... 103  5.3.1.2  IB strength and edge SWR of UD particleboard ............................................... 104  5.3.1.3  Effects of mass fraction of fine particles on IB strength and edge SWR of UD boards ................................................................................................................ 105  5.3.2  Three layer particleboards..................................................................................... 108  5.3.2.1  Summary of statistically significant effects of particle size mix on the properties of 3-layer particleboard ..................................................................................... 108  5.3.2.2  Mean density and VDP of the 3-layer particleboard ......................................... 108  5.3.2.3  Effect of mixture on flexural properties of the 3-layer boards .......................... 111  5.4  Conclusions .................................................................................................................. 112  5.5  Literature cited ............................................................................................................. 114  Chapter 6 Characterizing macro-voids in uncompressed mats and finished particleboard panels using Response Surface Methodology and X-ray CT5 ................................................ 118 6.1  Introduction .................................................................................................................. 118  6.2  Methodology ................................................................................................................ 121  6.2.1  Scaled-up (surrogate) and industrial particle preparation ..................................... 121  6.2.2  Shape determination for scaled-up (surrogate) particles ...................................... 123  6.2.3  RSM-mixture experiment ..................................................................................... 125  6.2.4  Simulation of surrogate and industrial particle packing ....................................... 127  6.2.5  Void determination of pressed panels with X-ray computer tomography ............ 128  6.3  Results and discussion.................................................................................................. 130  6.3.1  Surrogate particle packing .................................................................................... 130 vi  6.3.2  Industrial particle packing..................................................................................... 133  6.3.3  Relationship between core void volume in uncompressed mat and finished panel ............................................................................................................................... 136  6.3.4  Effects on IB strength and edge SWR of voids in RDP mat and finished panel .. 138  6.4  Conclusions .................................................................................................................. 140  6.5  Recommendation .......................................................................................................... 141  6.6  Acknowledgements ...................................................................................................... 141  6.7  Literature cited ............................................................................................................. 142  Chapter 7 Summary and conclusions .......................................................................................... 147 7.1  Project summary and conclusions ................................................................................ 147  7.2  Significance and limitations of the studies ................................................................... 152  7.3  Recommendations and further research ....................................................................... 154  7.4  Literature cited ............................................................................................................. 156  Appendices .................................................................................................................................. 158 Appendix A.  Questionnaire for particleboard manufacturers ............................................. 159  Appendix B.  Questionnaire for furniture manufacturers .................................................... 161  Appendix C.  Photographs and schematic drawing of screw used for screw withdrawal test ... ....................................................................................................................... 163  Appendix D.  Design of screw withdrawal testing apparatus .............................................. 164  Appendix E.  Photographs showing testing for screw withdrawal resistance ..................... 167  Appendix F.  Histograms of particle sizes of hydrolyzed sampled panels superimposed with the best distribution fit for particle length, AR, and SR ................................ 168  Appendix G.  Table of model parameters and maximum likelihood of lognormal and Weibull distributions fit to coarse, medium, and fine particle sizes............................ 174  Appendix H.  Table of AIC and -2loglikelihood values for testing lognormal, gamma, and Weibull distributions fit to coarse, medium, and fine particle sizes .............. 177  vii  List of tables Table 2.1  Mechanical properties, specimen ID and the number of samples for M2 particleboards. ......................................................................................................... 27  Table 2.2  Treatment structures for board physical and mechanical properties. ..................... 28  Table 2.3  Main effects on properties of M2 particleboards from six press lines.................... 31  Table 2.4  Means and COV (%) of physical and mechanical properties for boards from press lines A to F. ............................................................................................................. 32  Table 3.1  Minimum strength properties of medium (M) grades of industrial and shelving particleboard (from ANSI A208.1-1999). Note that no density ranges are specified in ANSI A208.1. ..................................................................................................... 47  Table 3.2  Mechanical properties, specimen ID and the number of samples for M2 and MS particleboards from two press lines. ....................................................................... 50  Table 3.3  Treatment structures for board physical and mechanical properties. ..................... 51  Table 3.4  Significant main effects and interactions for properties of MS and M2 grade particleboard from two press lines. ......................................................................... 53  Table 3.5  Means and COV of physical and mechanical properties of M2 and MS boards for two press lines. ........................................................................................................ 55  Table 4.1  Mesh sizes used for particle classification .............................................................. 72  Table 4.2  Particle mean mass on each mesh size.................................................................... 76  Table 4.3  Comparison of the means of geometrical descriptors of the core-fine, medium and coarse core particles ................................................................................................ 78  Table 4.4  Means of mechanical properties for panels from plants A to F. ............................. 80  Table 4.5  Model parameters of aspect and slenderness ratios for coarse particles ................. 84  Table 4.6  AICc values of aspect and slenderness ratios for coarse particles .......................... 84  Table 4.7  Goodness of fit test for the distribution models for length, aspect and slenderness ratios of core-fine, medium and coarse industrial particles. .................................. 86  Table 5.1  Particle size classes and amounts in furnish mixtures ............................................ 98  Table 5.2  Treatment structure of factors and response variables of 3-layer boards. ............ 102  Table 5.3  Mean values for the properties of the UD single layer boards. ............................ 104  Table 5.4  Effects of mixtures and target density on properties of 3-layer boards. ............... 108 viii  Table 5.5  Mean values for the properties of the 3 layer boards with novel furnish mix core. ... ............................................................................................................................... 110  Table 6.1  Mean dimensions, circularity, and aspect ratio (AR) of each particle size class. . 122  Table 6.2  Cylinder dimensions for packing simulation. ....................................................... 122  Table 6.3  Porosity of industrial furnish blends for RLP and RDP for industrial furnish and average responses for RLP and RDP. ................................................................... 130  Table 6.4  Porosity of surrogate particles for RLP and RDP. ................................................ 131  Table 6.5  Core void volume in pressed panels for various industrial particle mixtures. ...... 137  ix  List of figures Figure 2.1  Cutting pattern for the first two sub-panels showing randomized test specimen positions. Specimen numbers correspond to the specimen IDs given in Table 2.1. ................................................................................................................................ 26  Figure 2.2  The (a) oven dry density and (b) internal bond strength of M2 grade boards by press line. n = 80 for each mean. ANSI A280.1 (1999) recommended minimum IB of 0.45 MPa denoted by dashed line. ..................................................................... 34  Figure 2.3  Face SWR by press line and screw type, averaged across machine direction. n = 40 for each column. ANSI A280.1 (1999) recommended minimum of 1000 N denoted by dashed line. ........................................................................................................ 36  Figure 2.4  Edge SWR by press line, averaged across screw type and machine direction. n = 160 for each column. ANSI A280.1 (1999) recommended minimum of 900 N denoted by dashed line. .......................................................................................... 38  Figure 2.5  Average (a) MOR and (b) MOE by press line and machine direction. n = 80 for each column. The symbols within each column indicate machine direction (║parallel to and ┴ perpendicular to). ANSI A280.1 (1999) recommended minimum of 14.5 for MOR and 2.25 GPa for MOE denoted by dashed lines. ...... 41  Figure 3.1  Cutting patterns for the first two sub-panels showing randomized test specimen positions. Specimen numbers correspond to the specimen IDs given in Table 5.2. ................................................................................................................................ 49  Figure 3.2  The (a) oven dry density and (b) internal bond strength of MS and M2 grade panels by press line. ANSI A280.1 (1999) recommended minimum IB of 0.40 MPa for MS and 0.45 MPa for M2 are denoted by dashed lines. Each mean represents 80 replicates tested. ..................................................................................................... 54  Figure 3.3  Mean internal bond strength as a function of (a) mean oven dry density, and (b) mean core density, for MS and M2 boards from press lines A and B. Each mean represents 160 samples. .......................................................................................... 56  Figure 3.4  Mean face screw withdrawal strength by screw type for MS and M2 grade boards from press lines A and B. ANSI A280.1 (1999) recommended minimum of 1000 N  x  for M2 and 900 N for MS are denoted by dashed lines. Each column mean represents 40 replicates. ......................................................................................... 58 Figure 3.5  Edge SWR by press line and grade, averaged across screw type and machine direction. ANSI A280.1 (1999) recommended minimum of 900 N for M2 and 800 N for MS are denoted by dashed lines. Each column mean represents 160 replicates. ................................................................................................................ 59  Figure 3.6  Mean values for (a) MOR and (b) MOE for MS and M2 grade panels from press lines A and B; ║ = parallel to machine direction, ┴ = perpendicular. ANSI A280.1 (1999) recommended minimum MOR of 14.5 MPa (M2), 12.5 MPa (MS) and MOE of 2.25 GPa (M2) and 1.9 GPa (MS) are denoted by dashed lines. Each column mean represents 40 replicates. ................................................................... 60  Figure 3.7  Examples of typical vertical density profiles from M2 and MS specimens from (a) press line A, and (b) press line B............................................................................ 62  Figure 4.1  Distribution of mean particle mass as a percentage of total particle mass for each particle size class; LSD bars are given for comparison of the mean particles size from each plant for that size class. ......................................................................... 77  Figure 4.2  Relationship between mechanical properties (edge SWR and IB strength) and geometrical descriptors (AR and SR) for core-fine, medium, and coarse particles of particleboard core. The upper row, (a) and (b), is edge SWR and the lower row, (b) and (c), is IB strength. ............................................................................................ 81  Figure 4.3  Typical histograms of the observations overlaid with best fit distribution. Upper row shows distribution of particle length and aspect and slenderness ratios of the coarse particles from Plant A and the lower row shows distribution of particle length of fine, medium, and coarse particles from plant E..................................... 83  Figure 5.1  Cutting pattern of the 3-layer board (660 mm by 660 mm) for specimen sampling (m – furnish mixture; d – density; r – replicate; s – specimen). ........................... 101  Figure 5.2  Vertical density profile of the UD boards at a target density of 530 kg/m3.......... 103  Figure 5.3  Mean values from UD boards for IB strength (■), board density (●) and edge SWR (▲) loads for all particle size mixes. Note that the fines and dust content of boards, increases from M1 to M4 and the open symbols indicate mean values of CTL boards. .................................................................................................................. 106 xi  Figure 5.4  Vertical density profile of the 3-layer particleboard at a target density of (a) 650 kg/m3 and (b) 700 kg/m3. ..................................................................................... 109  Figure 5.5  Mean values of board densities for (a) IB strength and (b) edge SWR loads for each core particle size mix and target board density. Note that the fines and dust content of boards increases from M1 to M3. .................................................................... 111  Figure 6.1  Packing sequence of (a) surrogate blocks and (b) industrial particles. Rolling particles in cylinder (left), the resulting RLP (center) and the RDP after shaking the RLP (right)...................................................................................................... 123  Figure 6.2  Typical particle outlines obtained with image analysis software. Note: scale is the same for all particle sizes. .................................................................................... 125  Figure 6.3  Modified SCD for (a) surrogate blocks and (b) region of interest for industrial particles using the constraints............................................................................... 126  Figure 6.4  CT-imaging of particleboard core: (a) 2-D CT-image of sample, (b) Line profile across sample width, and (c) 3-D sample image showing core voids. ................. 129  Figure 6.5  Porosity of the RDP sample for surrogate blocks. Open dots indicate test mixtures .............................................................................................................................. 133  Figure 6.6  Porosity of industrial particles for (a) RLP and (b) RDP samples. Dots indicate test mixtures and the trapezium bounded by the dotted lines is the region of interest. .............................................................................................................................. 135  Figure 6.7  Correlations between porosity of particleboard panel for (a) fractions of core-fine, medium, and coarse particles used in the panel and (b) void fraction of uncompressed particle mat of the RDP and RLP samples. .................................. 138  Figure 6.8  Effect on IB strength and edge SWR of (a) percent void in RDP samples (b) percent void in pressed panel relative to density. Each point is an average of 48 samples for IB and 24 samples for edge SWR. .................................................... 139  xii  List of symbols k  Shape parameter for gamma distribution; Number of parameters in AIC  n  Number of samples  x1  Core-fine particle component  x1′, x2′, x3′  Pseudo components of core-fine, medium, and coarse particles  x1x2  Interaction between core-fine and medium particles  x2  Medium particle component  x2x3  Interaction between medium and coarse particles  x3  Coarse particle component  Ap  Projected area, mm2  Dr  Root diameter, mm  Dt  Shank diameter, mm  Fedge  Force required to pull a screw from the edge of a particleboard, N  Fface  Force required to pull a screw from the face of a particleboard, N  G (SG)  Specific gravity of sample at an MC of 10%  L  Embedment depth, mm ∧  L( θ )  Likelihood function  MC  Moisture content, %  P  Void fraction or porosity  Pd  Packing density  Pp  Particle perimeter, mm  X  Random variable  xiii  α  Scale parameter of Weibull distribution  σ  Standard deviation/scale parameter for lognormal distribution function  β  Shape parameter for Weibull distribution function  µ  Mean of sample population  Γ  Gamma function  λ -1  Scale parameter for gamma distribution  ∧  θ  Maximum likelihood estimator  βi, βij , βijk  Fractional solid volume coefficients  xiv  Acknowledgements An effort such as this over many years is never solely the result of a single individual. I would like to take the time here to appreciate and thank all those who helped me to make it possible. I would like to express my sincere gratitude and appreciation to my supervisor and chair of my supervisory committee Dr. Greg Smith for his guidance, advice, patience, and trust in me from the beginning to the end of this life-long journey. I wish to extend my sincere thanks to all the members of the supervisory committee. Dr. Stavros Avramidis, you were the one through whom I first came to UBC, I thank you for your tremendous advice and encouragement as a supervisor and committee member. My next thanks go to Dr. Chungping Dai as a committee member and for his astute advice and penetrating questions helped me to dig deeper. Many thanks go to Dr. Rizhi Wang for filling the vacant position on my advisory committee and for his tremendous and insightful advice.  Tremendous thanks go to Dr. Kate Semple, visiting Professors Drs He-Jun Park and Seung-Won Oh, Cheng Zhou, Hamid Fahkri, Graeme Dick, Chao Zhang, Solace Maame Araba Sam-Brew, and the entire Wood Composites Group for their comments, critic, assistance in material and sample preparations and measurements, and moral support. I would also like to extend my heartfelt gratitude to Drs Tony Kozak and Valerie Lemay in bringing sanity and direction to my statistical nightmare. My sincere appreciation also goes to Bob Myronuk and George Lee for their assistance in designing, fabricating, and mounting of my testing equipments. A tremendous thank you goes to Vincent Leung and Diana Hastings for their assistance in cutting my test samples and procurement of equipment parts. To Gabor in FP-Innovations, I say thank you for doing preliminary scans with your industrial X-ray CT and assisting me produce 3D images from xv  the scans. Special thank you to Peter Ens, Axel Anderson, and Martin Feng all of FPInnovations, Vancouver for their incredible help in using X-ray Densitometer and giving me insight into resin detection. The author is extremely grateful to NRCan, Value-To-Wood, and NSERC for funding the entire project. For allowing me use their many facilities like X-ray Densitometer and Industrial X-ray CT, I say thank you to FP-Innovations, Vancouver.  Now to Heinz and Traute Drenkberg, thank you and I appreciate your love, tenderness, and encouragement that you have always given to me since you adopted me as a son. Although you are far away in Germany, you seem to be always near to me and my family. I cannot end if I do not appreciate my spiritual supporters, who are my pastors Dr. Sam Owusu and Emmanuel Ayedzi, my CWC Vancouver Home Group, CWC Worship Team personified in Steve and Theo Bessem and the entire Calvary Worship Center.  Above all I would like to say, thank you God Almighty, I worship you for who You are, for the inner peace You give to me, for being always there whenever I need You, and leading me through to the end of this part of the journey. May the Name of Jehovah be exalted above all else.  xvi  Dedication I would like to dedicate this dissertation to my sweet heart Vida, and my children Queensther, Emmanuel, and Abigail for their relentless love, endless support, and countless sacrifices. Honey, without you I would have certainly given up. Your constant encouragement, prayer, and support kept me going: sometimes even coming to my laboratory to just cheer me up in the midnight hours. I can’t but say thank you and I will forever love you.  Queenie, your smile and your statements on the phone like “daddy I miss you, when are you coming back?” kept me going. Hey Junior, you have been a blessing to me when you call me at the office and say “Daddy I’m just checking on you to see how you are doing”. Your statement assured me that someone cares. And to you Abbie, I really appreciate you when you say “Daddy let’s take a ride” and we chat just only the two of us. I just want to say I love you all and I could not have achieved this without you. God richly bless you.  xvii  Co-authorship statement The five main chapters of this dissertation were co-written with the following co-authors: Kate E. Semple, He-Jung Park, Seug-Won Oh, and Gregory D. Smith.  Emmanuel K. Sackey, dissertation author, designed individual experiments, performed the research and data analysis, and in cooperation with the co-authors prepared the manuscripts. Kate Semple was involved in the design of benchmarking laboratory experiment measuring mechanical properties of industrial particleboards, sieving industrial particles, and collaborated in the preparation of the manuscripts in Chapters 2 and 3. The role of He-Jun Park was in sample preparation for testing and designing of jig for the measurement of SWR in Chapters 2 and 3. Seug-Won Oh assisted in preparation and sample measurements of laboratory made novel panels in Chapter 5. Gregory Smith was the research supervisor who identified the research topic, provided guidance throughout the project, commented, and edited all five manuscripts.  xviii  Chapter 1 Project introduction 1.1  Introduction  Particleboard is a generic term used for composite panel products manufactured from lignocellulosic materials, primarily in the form of discrete particles much larger than fibres; these particles are combined with a synthetic resin or other suitable binder and bonded together by heat and pressure (ASTM D1554 2005, Maloney 1993). The term particleboard (PB) has been used to denote flakeboard, waferboard and oriented strandboard, but for the purpose of this study, PB will refer to a panel manufactured from a mixture of wood particles smaller than wafers, flakes, and strands but larger than fibres. PB is primarily a non-structural wood composite used for furniture, door cores, and cabinets and for housing construction as floor underlayment, laminatefloorings, and stair treads. PB is the most used engineered wood product for furniture and furniture parts in North America (Wu and Vlosky 2000, Tabarsi et al. 2003). According their survey the respondents attributed PBs high usage to economic reasons, thickness uniformity, and good finishing characteristics. Another survey of RTA users by French (2009) also indicated that 61% of the respondent cited price as their major reason.  However, the North American PB and furniture plants are faced with fierce competition from lower cost furniture and furniture parts imports from the Asian-Pacific region (Furniture/Today 2003), as well as increased market share of South American and European producers. The beginning of the 21st century saw a sudden rise of Asian furniture imports to North America which liquidated almost all local wood production (Tourtellot and Dugan 2009). According to 1  RISI (2004), China’s furniture export value to the USA increased by 31% between 2002 and 2003, with Canada’s share increasing by only 1%. Compounded with this is the production curtailments of sawmills due the sub-prime mortgage crisis and low demand for lumber in the United States that has lead to a down turn in housing starts (Koncept Analytics 2009). This reduction means that there are fewer and fewer sources of wood residuals which thus necessitate the transport of these residuals over long distances to mills and results in higher wood and production cost (Wood Resource International 2007). Furthermore, producers of wood-based panels are confronted with high quality expectations from their customers. Research on flexural and internal bond (IB) strength of PB from some plants in North America has shown that only 38% of the plants met the ANSI minimum requirements for ready-to-assemble (RTA) panels (Cassens et al. 1994, Semple 2005). Telephone discussions and plant visits of PB and RTA furniture manufacturers indicated that PB has also limitations like edge particle pull-out, low edge fastener holding, and moldability compared to medium density fiberboard (MDF), its closest competitor. List of questions used during telephone discussions with PB and furniture manufactures can be found in Appendix A and Appendix B respectively. The issues have facilitated the closures of PB and furniture plants across North America. In order to mitigate the closures and increase board quality, research into efficient utilization of the available resources is required.  Wood constitutes about 90 to 95 % of the PB aggregate and is one of the major cost components in direct costs. Finding a more efficient means of utilizing the wood resource by improving board quality, reducing density, and creating new product lines can reduce production cost and enhance investment in new facilities. In the early years of comminuted wood composites, Mara (1969)  2  created a table of wood elements arranged from the largest (wood log) to the smallest (cellulose) (Maloney 1993). He indicated that with the application of basic processing concepts a myriad of other new homogenous products can be made. Since the first commercial PB in 1941, particle sizes in the PB have gone through transformations from flakes through particles and finally to strands in oriented strand board (OSB) (Chapman 2005). The development of PB mat formation has undergone significant refinement; mats were originally formed as a single layer, this was then replaced by a three layer mat where the fines were used in the face layers and coarse particles in the core, and most recently with the implementation of a graduated multilayered boards. These changes have resulted in a significant improvement in panel properties (Suchsland 1959, Maloney 1970, 1993).  One of the objectives of any wood composite is to reduce the variability in wood properties (mechanical and physical properties) by breaking the wood into smaller elements and then recombining them into a more homogenous assembly where large scale defects have been removed (English et al. 1997). However, the particles geometry of modern PB furnish are heterogeneous; this presents a challenge in controlling the variable behaviour of the particles in the panel structure, thus affecting the product conformity to performance standards. Notwithstanding the confounding effects of particle geometry and surface integrity, the strength properties of particulate wood composite panels are determined largely by the number and strength of the glue bonds between individual wood elements, (Walker 1993, Suzuki and Miyagawa 2003, Smith 2003). The earliest scientific papers on particle geometry in PB indicated that the relationships between the dimensions (length, width, thickness) of the particle not only determine the particle surface area available for gluing and bonding, but also affect the panel’s  3  strength and manufacturing costs (Pfohl 1937, Fahrni 1942, Klauditz 1949, Kollmann 1975). Suchsland (1959) indicated that the significant factor in determining flexural properties of a panel is the relative area of the compressed wood material and not its density. The relative compressed area is an indication of glue line between particle contact areas.  Among the earliest efforts to systematically optimize mechanical properties of particleboard through manipulation of particle geometry was a study by Turner (1954). Panels containing larger flakes measuring 76 x 25 x 0.46 mm were strongest with an MOR of approximately 65.5 MPa almost as strong as plywood (79.2 MPa). While smaller Douglas-fir (Pseudotsuga menziesii) planer shavings resulted in boards with MOR of around 17.2 MPa. Since particleboard is mostly used as non-structural panel and the required MOR for furniture grade M2 panels is 14.5 MPa, larger flakes with very high bending strength will be over engineering, hence the smaller particles were better suited as particleboard furnish. Numerous subsequent studies on the effects of particle geometry on particleboard strength properties indicate that flexural properties of particleboard increases with increasing flake length and reach a maximum at flake thicknesses of about 0.25 to 0.3 mm; although this may differ slightly according to flake length, panel density and resin content (Post 1958, 1961, Heebink and Hann 1959, Brumbaugh 1960, Gunn 1963, Gatchell and Heebink 1964, Heebink et al. 1964, Mottet 1954, 1967, Lehmann 1974, Shuler and Kelly 1976, Haselein et al. 2002).  Further studies have shown that flexural properties increase with slenderness ratio, SR, (ratio of particle length to thickness) (Brumbaugh 1960, Heebink and Hann 1959, Post 1961), but decrease with increasing aspect ratio, AR, (ratio of particle length to width). Intuitively this is  4  what one would expect as the narrower the particles, the more easily they can be aligned. A study on wood-plastic composites (WPC) shows that AR has the highest effect on tensile strength (Stark and Berger 1997, Stark and Rowlands 2003). As for studies examining particle width relative to panel strength, very few studies are found in literature with the exception of Kusian (1968) who observed that MOR increases with particle width, but decreases as AR approaches unity (Kelly 1977). Research on particle orientation and layer characteristics brought about developments of flakeboards, waferboards, and eventually strandboards (Elmendorf 1950, Clark 1953, Geimer et al. 1975).  Internal bond strength (i.e., the tensile strength perpendicular to the plane of the panel) is a measure of the panel integrity or how well the board is bonded together and is controlled by the weakest link in the product. The weakest link may be due to either a poorly bonded region (due to poorly distributed resin or uncured resin) and/or the presence of lowest density regions within the panel. Recently a mechanistic model to predict IB strength was developed (Dai et al 2008). The model predicts IB strength as a function of panel density, wood density, resin content, and particle dimensions, which is also a determining factor in macro-void creation in panels. Dai et al. found that IB strength increases with strand thickness monotonically due to the increasing area coverage of the resin on the strand surface.  Unlike MOR and MOE, IB strength generally increases in panels containing shorter, thicker flakes such as planer shavings or slivers (Brumbaugh 1960, Rackwitz 1963, Gatchell et al. 1966, Mottet 1967, Stewart and Lehmann 1973, Lehmann 1974, Schuler and Kelly 1976). Thicker and shorter particles have higher specific surface area, hence are better covered with resin. In  5  addition these particle types have better inter-particle contacts under compression which leads to higher bonding between adjacent particles, consequently leading to higher IB strength. Since strength requirements for furniture grade particleboard products are relatively low compared to structural wood panels, the size of wood elements can be significantly reduced to provide the required surface smoothness and uniformity of the internal structure.  Better bonding between thicker particles with high specific surface area (surface area to mass ratio) is not only an artefact of increased compaction ratio (typically 1.3 for PB), but also lends a greater resin spread per unit particle surface area (Maloney 1993). Properties critical to furniture grade particleboard such as screw withdrawal resistance (SWR) is closely tied to IB strength (Eckelman 1973, Fujimoto and Mori 1983, Semple 2005). This demonstrates that the size and shape of particles used in PB must be balanced to ensure the desired combination of flexural strength and internal bond properties for applications such as furniture components. However, most of the earlier investigations on mechanical strength concentrated on panels with large flakes on the surface with smaller flakes in the core. But three-layer configuration in particleboard with fine particles in the surface layers and short, thicker particles in the core served to balance both flexural and internal bond strength properties in particleboard (Suchsland 1959). Size and geometry of the surface particles also needs to be balanced between providing a smooth, closely packed surface for lamination but still enabling water vapour to travel through during pressing (Maloney 1993). The manipulation of particle size configuration in particle mat will strongly influence layer permeability and therefore the rate of heat and moisture flow from the surface layers and platens to the core layers (Maku et al. 1959, Hata 1993, Bolton and Humphrey 1994).  6  Various studies have sought to determine the effects of a range of factors such as resin type and content, particle size, density and wood species (either singly or in combination) used in the manufacture of particleboard on its mechanical properties. However, the influence of particle geometry on screw withdrawal resistance (SWR), which is one of the most important properties for furniture applications, was rarely investigated. An early work by Post (1961) examined the effect of varying flake thickness (from 0.15 to 1.27 mm) on screw withdrawal from particleboards. The resin coverage per unit flake surface area was held constant by manipulating flake size. His work suggested that increasing the length of thick flakes can increase SWR. However, Kimoto et al. (1964) found that SWR becomes constant above a slenderness ratio of 50. A patent by the Hiag Company (FP 2,161,199) for the manufacture of particleboard with high edge screw-holding strength is also cited in Meyer 1979. The patent recommends the use of 16 to 20% binder which is about twice the normal amount used in commercial manufacture of PB. Such a board would clearly be one solution to low screw-holding strength, but one with such a high resin content is not economically viable.  For PB bonded with 8% tannin formaldehyde and pressed to 9.5 mm thickness, SWR was significantly increased when particle thickness used in boards was increased from 0.5 to 1.0 mm (Haselein et al. 2002). This finding accords with the well known trend of IB strength in particleboard being enhanced by the use of thicker particles in the core through improved compaction ratio and resin coverage (Maloney 1993). In a study by Jamaludin et al. (2000) who worked on bamboo (Gigantochloa scortechinii) particleboard, particle size was also found to influence face and edge SWR; however, in this case SWR values increased for boards made from smaller particles. Altering the species mix used to manufacture panels has been found to have  7  little or no effect on SWR of boards (Menezzi et al. 1996, Iwakiri et al. 1996). This suggests that this property is influenced less by the nature and density of the parent wood but more by the density, inter-particle bonding, and particle characteristics of the finished particleboard.  The strong interaction of particle geometry with mat density affects springback and dimensional stability (thickness swell and water absorption) of particleboard. During hot pressing compression stress on particles leads to greater residual stress which in turn cause the particleboard to swell when exposed to moisture (Kollman et al. 1975, Kelly 1977). Thickness swell has been reported to increase for flake thicknesses greater than 0.3 mm (Klauditz 1955); thus the thicker the flakes or particles the greater the compression hence greater residual stress and thickness swell.  Early work by Maku et al. (1959) showed that particle size and configuration throughout the vertical cross section of the mat influences the rate of moisture transfer longitudinally from the central zone to the edges, with small, discrete granular particles retaining porosity of the mat better than flakes during press closure (Kamke 2004, Thoemen and Klueppel 2008). For low density layers, small, discrete particles provide more pathways through which moisture and other gases may flow than larger, flake-like particles (Maku 1959). Work by Hata (1993) showed that increasing particle length and width decreases both the horizontal (in-plane) and vertical (transverse) permeability of the mat, whereas increasing particle thickness (reduced slenderness ratio) increases mat permeability, i.e., it is more difficult to completely close the void spaces around blocky particles compared to thinner more compliant particles. This may be due to increased particle discontinuities which create more pathways. The in-plane permeability of the  8  mat is always higher than in the transverse (through the thickness) direction, and so adjusting particle size parameters has a more marked effect on in-plane permeability (Hata 1993).  The size and shape of particles determines the pore system in the mat during pressing, which in turn affects vapour and heat flow both vertically and horizontally through the mat (Bolton and Humphrey 1994); they demonstrated this by measuring the gas permeability of uniform density boards made from model particles (small cubes 0.5 to 0.5 mm for surface layers, splinters 0.5 x 13 mm for the core, and flakes 0.5 x 3 x 25 mm) to represent waferboards. The flakes created a heterogeneous, tortuous pore system while the cubes compacted together much more readily during pressing, obliterating many of the pores and greatly reducing mat permeability during pressing. Narrow splinters resulted in the greatest retention of the mat pore network with higher permeability during pressing than for the other two particle shapes. The need to increase the rate of heat and moisture transfer both into the mat for resin cure and out again for de-gassing to minimize total pressing time is therefore the justification for the use of small, discrete particles in modern particleboard.  Modern particle furnish has a range of particle sizes, which can be further exploited to improve panel properties or create new products from the combination of the different element sizes or with other non-lignocellulosic material. Research into particle sizes, distributions, and shapes has been modeled in areas like powder technology, pharmaceuticals, and metallurgy to improve product properties, but there is very little work reported on comminuted wood particles. The following research proposes the development of novel particle mixtures based on mixture design  9  using the response surface methodology (RSM) to improve core properties of PB at low density without increasing resin content.  1.2  Rationale  The geometry of the wood element plays a crucial role in the manufacturing process of wood composite. With the advent of the environmental movement, recycling of biomaterials, and nanotechnology, it is critical that a more in-depth understanding of the behaviour of wood particles and their intra- and inter-particle relationships with other processing parameters be acquired. Aside from the inherent properties of the wood species, particle geometry is the most important factor controlling the particle behaviour. Particle geometry relates to the shape and size of the particle. Particle geometry determines the drying and blending techniques employed in a plant and is very closely related to the resin consumption and spatial arrangement of particles in the mat which in turn affect the mechanical, fastener, and dimensional stability properties of the resulting panel.  Most of the research on flakeboard was for larger and wider (coarser) particles than used in modern PB and recent research on comminuted wood element geometry have concentrated on strand-based panels. Both flake and strand boards have coarse material on the face and finer in the core with strands oriented in modern oriented strand board (OSB), whereas PB mats are generally formed with fines on the face and coarse particles in the panel core. However, the definition of fines and coarse particles between plants and researchers varies substantially.  10  For furniture manufacturers the weight of their finished products is important. Not only does a lighter panel mean the manufactures pays less for the constituent materials, it will also help to address the current shortage of wood fibre. In addition, the amount of furniture that can be shipped on a truck is frequently limited by weight, rather than volume. If the weight of the furniture can be reduced, the amount of product that can be shipped on a truck is increased, and shipping costs reduced. The lower product weight will also be a competitive advantage for Canadian RTA furniture manufactures that are competing with offshore imports. The development of an attractive, marketable low density PB product with improved bonding and fastener holding properties will help to reduce production cost. It will also constitute a significant advancement for the modular furniture industry in North America.  With most PB being used as furniture and in non-structural applications, the most important panel properties are surface quality, SWR, IB strength, and thickness swell. There are ample studies into the effect of particle geometry on water relations of PB, but there is a gap in the public literature on the effect of wood particle geometry on SWR. The hammer-milled furnish used for PB has a wide variation in size and shape, but there is a dearth of literature on the size and shape distribution of particles in modern PB furnish and knowledge of their effect on a panel’s properties is lacking. Understanding the particle configurations of the larger core particles will help utilize wood efficiently, reduce wood consumption, consume less resin, and develop lightweight panels.  This research is therefore a contribution to the understanding of the configuration of wood particle in the spatial structure of the core of particleboard. It is focused on surveying and  11  benchmarking the mechanical properties of furniture grade particleboards from plants across Canada. The research is intended to characterize the particle size distribution in contemporary particleboard. A novel particle configuration for particleboard core is presented. Assessment of macro-voids in a simulated particle mat and pressed panel of the developed particle configuration is presented with predictive models that describe the macro-voids in these panels.  1.3  Objectives  The principal aim of this research is to develop a particle mixture for the core of particulate wood composites to improve mechanical properties that are most important to Ready-ToAssemble (RTA) furniture manufacturers. The main objectives are to:  1. Benchmark the properties of furniture grade M2 PB used to manufacture RTA furniture across Canada and determine any variation between those properties and the set standards. 2. Compare the properties of the furniture grade MS and M2 PB product manufacture in Canada. 3. Develop empirical distribution models for slenderness and aspect ratios for the core particles of particulate wood composites. 4. Fabricate a particle mixture to improve IB strength and edge SWR of PB through particle size redistribution. 5. Characterize macro-voids in an uncompressed mat made from a novel particle mixture and relate them to those in the finished panel using RSM with X-ray CT technology  12  1.4  Structure of dissertation  The dissertation is divided into seven chapters, including this chapter, the project introduction. Chapters 2 through 6 are written as standalone chapters each addressing the five objectives listed above. Chapters 2 to 5 have been published in peer reviewed journals. Chapter 6 has been accepted for publication by a refereed journal. Chapter 2 presents a survey and benchmark of the commonly used furniture grade M2 panel properties from particleboard companies in Canada. Chapter 3 compares and contrasts the properties of furniture grade MS and M2 panels from some companies in Canada. Chapter 4 examines the best particle geometrical distribution models for contemporary particleboard furnish. Chapter 5 presents the development and fabrication of a novel three particle size configuration for the use in the core of particleboard. Chapter 6 discusses the effects of the newly developed particle configuration on macro-voids and presents predictive models for macro-voids in the core of contemporary particleboards. Finally, Chapter 7 summarises the conclusions of Chapters 2 through 6 and the limitations of the research findings.  13  1.5  Literature cited  ASTM Standard D1037-06a. (2006) Standard Test Methods for Evaluating Properties of WoodBase Fiber and Particle Panel Materials. ASTM International, West Conshohocken, PA, 2006, DOI: 10.1520/C0033-03R06. ASTM Standard 1554 – 01 (2005) Standard Terminology Relating to Wood-Base Fiber and Particle Panel Materials. ASTM International, West Conshohocken, PA, 2005, DOI: 10.1520/D1554-01R05. www.astm.org. Bolton, A.J. and P. E. Humphrey. 1994. The permeability of wood-based composite materials Part I: A review of the literature and some unpublished work. Holzforschung 48 (Supplement): 95-100. Brumbaugh, J. 1960. Effect of flake dimensions on properties of particle boards. Forest Products Journal 243-246. Cassens, D.L., J. P. Bradtmueller, and F. Picado. 1994. Variation in selected properties of industrial grade particleboard. Forest Products Journal 44(10): 50-56 Chapman, K. M. 2005. Wood-based panels: Particleboard, fibreboards and oriented strand board (427-475 pp). In: ed. Walker J.C.F. Primary wood processing, principles, and practice. [Cited Sept 30, 2009]. Available from http://www.springerlink.com/content/w7480255686134u7/fulltext.pdf Clark, J. d’A. 1953. Production of fibrous elements from woody material. US. Patent No. 2 655 189. Dai, C., C. Yu, and J. Jin. 2008. Theoritical modeling of bonding characteristics and performance of wood composites. Part IV. Internal bond strength. Wood and Fiber Sc. 40(2): 146-160.  14  Eckelman, C.A. (1973). Holding strength of screws in wood and wood-based materials. Research Bulletin No. 895, Purdue University Agricultural Research Station, W. Layafette, Indiana. 15 pp. Elmendorf, A. 1950. Wood fibers from veneer waste. Paper Trade Journal (Feb. 9, 1950:29-31). English, B., P. Chow, and D. S. Bajwa. 1997. Processing into composites. In: Rowell, R. M., Young, R. A., Rowell, J. K., eds. Paper and composites from agro-based resources. Boca Raton, FL, CRC Lewis Publishers: 269-299. Chapter 8. Fahrni , F. 1942. Die Holzspanplatte. Holz als Roh-und Werkstoff 6:277-283. French, D. 2009. A consumer’s view of RTA. Furniture/Today and HGTV 2009 Consumer Views Survey. [Cited on October 2, 2009]. Available from http://www.furnituretoday.com/blog/Research_Says/23540A_consumer_s_view_of_RTA.php. Fujimoto, Y. and M. Mori. 1983. Performance of wood screw joints for particleboard. Science Bulletin of the Faculty of Agriculture, Kyushu University 30(1): 45-47. Furniture/Today. 2003. Wood bedroom shipments from China rise dramatically. Furniture/Today October, 03. 1pp. Gatchell, C.J. and B. G. Heebink. 1964. Effect of particle geometry on properties of molded wood-resin blends. For. Prod. J 15(11):501-506. Gatchell, C.J., B.G. Heebink, and F.V. Hefty. 1966. Influence of component variables on the properties of particleboard for exterior use. For. Prod. J.16 (4): 46-59. Geimer, R.L., H.M. Montrey and W.F. Lehmann. 1975. Effects of layer characteristics on the properties of three layer particleboards. For. Prod. J. 25(3): 19-29. 15  Gunn , J.M. 1963. Wafer dimension control: Number 1 design criterion for plant producing particleboard for building construction uses. For. Prod. J. 13(4):163-167. Haselein, C.R., L. Calegari, M. Y. Barros, C. Hack, E. Hillig, D. T. Pauleski, and F. Pozzera. (2002) Mechanical strength and dimensional stability of particleboard made with different particle sizes. Ciencia Florestal 12(2): 127-134. Hata, T. 1994. Heat flow in particle mat and properties of particleboard under steam injection pressing. Kyoto University Wood Research Bulletin No. 40, pp 1-47. Heebink, B.G. and R. A. Hann. 1959. How wax and particle shape affect stability and strength of oak particleboards. For. Prod.J. 9(7):197-203. Heebink, B.G., R. A. Hann, and H. H. Haskell. 1964. Particleboard quality as affected by planer shaving geometry. For. Prod. J. 14(10): 486-494. Iwakiri, S., J. V. Latorraca, D. A. Silva, J. L. Gabardo, R. J. Klitzke, A. Fofano, F. Fabrowski, and M. T. Interanmense. 1996. Particleboard manufacture from Pinus elliottii and Eucalyptus dunnii. Revista do Setor de Ciencias Agrarias 15(1):33-41. Jamaludin, K., A. Abd-Jalil, H. Jalaluddin, M. Abd-Latif, and M. Y. Mohd-Nor. 2000 Interior grade particleboard from bamboo (Gigantochloa scortechinii): influence of age, particle size, resin and wax content on board properties. Journal of Tropical For. Prod. 6 (2):142151. Kamke, F. A. 2004. Physics of hot pressing. Winandy, J. E.; Kamke, F. A., Eds. Fundamentals of composite processing. Proceedings of a workshop; November 5.6, 2003; Madison, WI. Gen. Tech. Rep. FPL-GTR-149. Madison, WI: U.S. Department of Agriculture, Forest Service, Forest Products Laboratory. 118 p.  16  Kelly, M.W. 1977. Critical literature review of relationships between processing parameters and physical properties of particleboard. Forest Products Laboratory Forest Service U.S. Dept. of Agriculture, Madison, Wis. Kimoto, K., E. Ishimori, H. Sasaki, and T. Maku. 1964. Studies on the particle boards Report 6:.Effects of resin content and particle dimension on the physical and mechanical properties of low-density particle boards. Departmental Bulletin Paper. KURENAI: Kyoto University Research Information Repository.15pp. Klauditz, W. 1949. Untersuchungen ueber die ergibigkeit von Phenol von Phenol- und Harnstoff-Formaldehyd-Kunstharz-Bindemittel  bei  der  Herstellung  von  Holzspannplatten. Inst. Fuer Holzforshung, Braunschweig, Ber. No 14/1949. Klauditz, W. 1955. Development, status, and economic significance of wood particlebord manufacturing. Holz als Roh-und Werkstoff 13(11):405-421. Kollmann, F. F. P., E. W. Kuenzi, and. A. J. Stamm. 1975. Principles of Wood Science and Technology II. Wood based materials. Berlin, Heidelberg, New York, Springer-Verlag. 703pp. Koncept Analytics. 2009. Wood-Based Panel Industry: An Analysis June 2009, Pages: 45. [Cited October 07, 2009]. Available from http://www.researchandmarkets.com/reportinfo.asp?report_id=998736. Kusian R (1968) Model investigations about the influence of particle size on structural and strength properties of particle materials. II. Experimental investigations. Holztechnologie 9(4):241-248. Lehmann, W.F. 1974. Properties of structural particleboard. Forest Products Journal 24(1):  17  Maku, T., R. Hamada and H. Sasaki. 1959. Studies on the particleboard (chipboard) IV. Temperature and moisture distribution in particleboard during hot pressing. Wood Research Bulletin Kyoto University 21: 34-46. Maloney, T.M. 1993. Modern Particleboard and Dry Process Fiberboard Manufacturing 2nd Edition. Miller Freeman, San Fransico. 681 pp. Maloney, T. M. 1970. Resin distribution in layered particleboard. Forest Prod. J. 20(1): 43-52. Marra, G. G. 1969. A vision of the future of particleboard. In: Proceedings of the 3rd Washington State Particleboard Symposium. Pullman, Washington. Menezzi, C.H., M. R. Souza, and J. C. Goncalez. 1996. Fabrication and technical evaluation of particleboard from a mixture of Eucalyptus urophylla T.S. Blake and Pinus oocarpa Schiede. Revista Avore 20(3): 371-379. Meyer, B. 1979. Urea Formaldehyde Resins. Addison-Wesley Publishing Company Inc., Massachusetts. Pp 176. Mottet, A.L. 1954. The manufacture of general-purpose and decorative pressed wood boards by a dry process. Journal of the For. Prod.Res.Soc. IV (5): 224-227. Mottet, A.L. 1967. The particle geometry factor in particleboard manufacturing. In Proceedings 1st Washington State University Symposium on Particleboard, Pullmann, Washington, ed T.M. Maloney. pp. 23-73. Pfohl, F. 1937. Panels for furniture and cabinetry and method for their production. Swiss Patent No. 193139  18  Post, P.W. 1958. Effect of particle geometry and resin content on bending strength of oak flakeboard. Forest Prod. J. 8(10): 317-322. Post, P.W. 1961. Relationship of flake size and resin content to mechanical and dimensional properties of flakeboard. Forest Prod. J. 11(1): 34-37. Rackwitz, G. 1963. Influence of chip dimensions on some properties of wood particleboard. Holz als Roh- und Werkstoff 21(6): 200-209. RISI. 2003. Particlebaord and MDF commentary. Resource Information Systems, Inc., Bedford, MA, USA. Semple, K.E., E. Sackey, H.J. Park, and G.D. Smith. 2005. Properties variation study of furniture grade M2 particleboard manufactured in Canada. Forest Prod. J. 55(12):117-124. Shuler, C.E. and R.A. Kelly. 1976. Effect of flake geometry on mechanical properties of Eastern spruce flake-type particleboard. Forest Prod. J 26(6): 24-28. Smith, G. 2003. A laboratory technique for coating strands with resin droplets of controlled size and spacing. Forest Prod. J 53(7/8): 70-76. Stark, N. M. and M. J. Berger. 1997. Effect of particle size on properties of wood-flour reinforced polypropylene composites. In: Proc. of the 4th International Conf. on Woodfiber-Plastic Composites. For. Prod. Soc., Madison, WI. Stark, N. M. and R. E. Roland. 2003. Effects of wood-fiber characteristics on mechanical properties of wood/PP composites. Wood and Fiber Sci. 35:167-174. Stewart, H.A. and W.F. Lehmann. 1973. High-quality particleboard from cross-grain knife planed hardwood flakes. Forest Prod. J. 23(8): 52-60.  19  Suchsland, O. 1959. An analysis of particleboard process. Quarterly Bulletin, Michigan Agricultural Experiment Station, Michigan State University, 42(2):350-372. Suchsland, O. 1960. An analysis of a two-specie three-layer wood flakeboard. Quarterly Bulletin, Michigan Agricultural Experiment Station, Michigan State University, 43(2):375-393. Suzuki, S. and H. Miyagawa. (2003) Effect of element type on the internal bond quality of woodbased panels determined by three methods. Journal of Wood Science 49: 513-518. Tabarsi, E., R. Kozak, D. Cohen, and C. Gaston. 2003. A market assessment of the potential for OSB products in the North American office furniture and door manufacturing industries. Forest Prod. J 53(7/8): 19-27. Thoemen, H. and A. Klueppel. 2008. An investigation on the permeability of different wood furnish materials. Holzforschung 62:215-222. Tourtellot, P. and M. K. Dugan. 2009. The American Furniture Industry: What will it take to survive? An ABTV Industry Watch Report Sept. 2009. Greensboro, NC 27402. [Cited September 25, 2009]. Available from http://abtv.com/furniturewatch.pdf Turner, H.D. 1954. Effect of particle size and shape on strenth and dimensional stability of resinbonded wood-particle panels. Forest Prod. J. 4(5): 210-223. Walker, J.C.F. 1993. Wood panels: particleboards and fibreboards. Pages 419-480 in Primary Wood Processing: Principles and Practice, ed J.C.F. Walker, Chapman & Hall, London. 595pp. Wood Resources International, LLC. 2007. North American Wood Fiber Market Update. North American Wood Fiber Review June 2007. 24pp.  20  Wu, Q. and R. P. Vlosky. 2000. Panel products: A perspective from furniture and cabinet manufacturers in the southern United States. Forest Prod. J. 50(9): 45-50.  21  Chapter 2 Properties variation study of furniture grade M2 particleboard manufactured in Canada1 2.1  Introduction  Particleboard is the most common reconstituted wood composite panel used in the furniture manufacturing industry (Wu and Vlosky 2000). It has much greater structural homogeneity and surface smoothness than oriented strandboard (OSB) and is cheaper to produce and lower in density than its closest competitor, medium density fiberboard (MDF). The suitability and acceptance of particleboard for use in furniture has greatly improved since a feasibility study by Suchsland and Good (1968) identified some specific areas for product improvement. The major problems were difficult edge moulding and banding, fastener holding problems, low strength to weight ratio, poor water resistance, and dimensional instability. Most of these problems have been rectified by the industry; the characteristic three-layer structure, i.e., densified surface layers containing fine particles and less densified core containing coarse particles, gives furniture-grade particleboard the required surface smoothness and flexural strength-to-weight ratio for laminated modular components.  Despite these improvements in particleboard quality, recent surveys in the United States have identified high variation in the mechanical properties of particleboards across different mills,  1  A version of this chapter has been published. Semple, K., E. Sackey, H.J. Park, and G.D. Smith. 2005. Properties variation study of furniture grade M2 particleboard manufactured in Canada. Forest Products Journal, 55(12):117125.  22  which in turn perpetuate the end-user perception of the product as low quality. A survey by Temple-Inland Panel Products (Bautista 1994, Zhang 1994) found wide variation in the mechanical properties of 40 particleboards from 25 mills representing 15 companies. The use of different wood species, particle geometries, and pressing strategies by mills were identified as the main causes of variation in properties. A survey by Cassens et al. (1994) across seven different U.S. suppliers of the same 3/8-inch M2 grade product also found significant variation in all properties except for internal bond (IB) strength. The products were supplied to a kitchen cabinet manufacturer who found that panels from two of the seven mills did not meet minimum standards. Wu and Vlosky (2000) point to a lack of information about specific customer requirements of particleboard and other furniture-grade panels hindering the development of new markets and making it difficult for existing suppliers to improve and tailor their products to customer requirements.  Most particleboard strength properties are improved by increasing panel density, which results in better inter-particle contact during pressing (Maloney 1993), but this is offset by the greater weight and dimensional instability of highly densified particleboard. The U-shaped density profile of modern particleboard is a compromise designed to provide sufficient bending strength and dense, smooth surfaces, but at the same time keeping overall panel weight and dimensional instability down. The lower density core, however, has the effect of reducing the edge fastener holding ability of particleboard (Kelly 1977, Wong et al. 1999). Over-driving of screws further compromises fastening strength (Carroll 1970), and a low density core can be easily damaged by the insertion of screws into pilot holes of an inappropriate size (Rajak and Eckelman 1993), further contributing to low screw holding ability. Because conventional screws require higher  23  strength embedment materials, a range of alternative ‘non-screw’ connectors such as joining plates, dowels, or plugs made from wood, plastic, or metal have evolved for assembling particleboard furniture components (Schmidt 1986). However in North America, screws and predrilled pilot holes are still the most common method of connecting particleboard furniture components.  In contrast to the United States, there is no publicly available comparative study of the variability in physical and mechanical properties of furniture-grade particleboards manufactured in Canada, which is the source of most raw particleboard used in the manufacture of ready-to-assemble (RTA) furniture in Canada. This study attempts to address this lack of information by sampling 5/8-inch M2 grade particleboards and measuring their strength and screw holding properties. The study compares panel properties across press lines and provides information on the within and between panel variation of properties, including face and edge screw withdrawal resistance (SWR).  The objectives of this study were to: 1. Compare the density, IB, face and edge SWR, and flexural strength and stiffness of furniture grade M2 particleboards produced on six press lines across Canada and 2. Determine the effects of screw thread pitch on SWR and of particle orientation (with respect to the machine direction of mat) on SWR and flexural properties.  24  2.2  Materials and methods  2.2.1 Sample sourcing particleboards and sample preparation A letter of request for five 15.87- mm- (5/8-in) thick furniture grade M2 particleboards was sent to companies known to produce M2 grade particleboard in Canada. Five out of nine mills contacted agreed to participate in the study, with these mills covering a wide geographical area. The five mills (hereafter referred to as press lines) represented four different companies and were randomly assigned a code A to F, with one mill producing panels from two different press lines. Of the six press lines from which panels were sampled, four were batch presses and two were continuous presses. Five replicate sheets of 4- by 8-foot particleboard wrapped in plastic were received from each mill, for a total of 30 panels. Each of these was cut into eight sub-panels each measuring 61 by 61 cm (2 by 2 ft) using a vertical panel ripping saw. Each sub-panel was labeled with the press line code (A to F), panel replicate (1 to 5), sub-panel position number (1 to 8), and machine direction of the mat (parallel to, ||, or perpendicular to, ⊥). These sub-panels were stacked between stickers in a conditioning room at 20 ± 1°C and 65 ± 5 percent relative humidity (RH) for a minimum of two weeks prior to cutting test samples in accordance with ASTM D 1037 (2000) testing procedures. They were then cut into test samples as shown schematically in Figure 2.1 and listed in Table 2.1; the numbers in the specimen ID column of the table correspond with the test specimen codes on Figure 2.1.  25  1  12  3  5  6  12 11  8,10  8,10 12 11  3 1  2  5  1 Figure 2.1  11  7,9  2  7,9  machine direction  4  11  4 12  6  2  Cutting pattern for the first two sub-panels showing randomized test specimen  positions. Specimen numbers correspond to the specimen IDs given in Table 2.1.  The two samples for screws A and B for edge SWR (machine direction), edge SWR (perpendicular direction), and face SWR were side-matched pairs but locations of the pairs and those of all of the other test pieces were randomized within each sub-panel to avoid any location bias. For those press lines with batch presses, the machine direction corresponded to the long edge of the panel. For the two press lines with continuous presses, the machine direction corresponded to the short edge of the panel; therefore the overlaid specimen location template for each sub-panel was rotated 90° clockwise. Test samples were conditioned again after cutting at 20 ± 1°C and 65 ± 5 percent RH for an additional two weeks prior to testing of mechanical properties. Density (ovendry), moisture content (MC), and strength properties were tested in accordance with ASTM D 1037 (2000).  26  Table 2.1  Mechanical properties, specimen ID and the number of samples for M2  particleboards. property  specimen ID  number of samples per press line  board  sub-panel  edge SWR (machine direction) Screw A  1  40  8  1  edge SWR (perpendicular) Screw A  2  40  8  1  edge SWR (machine direction) Screw B  3  40  8  1  edge SWR (perpendicular) Screw B  4  40  8  1  face SWR Screw A  5  40  8  1  face SWR Screw B  6  40  8  1  MOR (machine direction)  7  40  8  1  MOR (perpendicular  8  40  8  1  MOE (machine direction)  9  40  8  1  MOE (perpendicular  10  40  8  1  internal bond  11  80  16  2  board oven-dry density  12  80  16  2  2.2.2 Experimental design and statistical analysis The experimental design consisted of five full-sized (1.22 by 2.44 m or 4 by 8 ft) particleboard panels from each of six ‘press lines’. Due to lack of control over which mills were sampled, ‘press line’ as a main effect is considered a random factor. Each panel (replicate) contained eight test samples (one per sub-panel) for properties 1 to 10 listed in Table 2.1. In the case of density and IB, properties 11 and 12, respectively, there were two samples per subpanel, giving 16 samples per panel. The treatment structures (fixed and random factors) for each mechanical property are summarized in Table 2.2.  27  Table 2.2 property  edge SWR  face SWR  MOR, MOE  IB, density  Treatment structures for board physical and mechanical properties. specimen ID  1–4  5-6  7 – 10  11, 12  fixed factors  levels  random factors  levels  press line  A to F  machine direction  ║, ┴  board  1 to 5  screw type  A, B  sub panel  1 to 8  specimen  1 to 4  press line  A to F  board  1 to 5  sub panel  1 to 8  specimen  1, 2  press line  A to F  board  1 to 5  sub panel  1 to 8  specimen  1, 2  press line  A to F  board  1 to 5  sub panel  1 to 8  specimen  1, 2  screw type  machine direction  A, B  ║, ┴  none  —  Each treatment structure involved a hierarchical design with one or more fixed factors depending on the response variable tested. The responses — density, IB, modulus of rupture (MOR), modulus of elasticity (MOE), face and edge SWR data were modeled using a split-plot randomly complete block design (RCBD). Edge SWR, MOR, and MOE were tested in and perpendicular to machine direction. Presslines (6) were considered as blocks and within them, five experimental units (panels) were randomly assigned. The panels were then split into subpanels (8) from which specimens were sub-sampled.  28  The sample testing was blocked by replicate (panel), i.e., for each mechanical property the samples from all of the press lines for replicate 1 were tested in random order followed by all of those for replicate 2, and so on. The order in which the subpanels were tested was also randomized for each panel replicate and press line. The testing of mechanical properties took place over a period of approximately three months. The main effects of, and interactions between, each set of factors for a particular property were tested for significance by an appropriate analysis of variance (ANOVA) model at the 5% significance level using Genstat 5 release 4.21 (Lawes Agricultural Trust 2001). Before the final analyses, diagnostic checks on the normality of the errors and equality of variances were undertaken. Significant results are compared graphically and the Least Significant Difference (LSD) for p ≤0.05 is included on graphs to facilitate comparison of means.  2.2.3 Measurement of panel density Each IB sample (number 11 on Figure 2.1) had a matching ovendry density/MC sample (number 12) cut adjacent to it. The ovendry density of these samples was determined by the water displacement method using the ovendry weight and the undried specimen volume (at 10% equilibrium moisture content [EMC]). The minimum core density was determined from the vertical density profiles measured for each IB specimen prior to testing. An X-ray density profilometer (Quintex Measurement Systems; Model QDP-01X) was used to measure the density of IB specimens at intervals of 0.1 mm through the thickness of the specimen. A cross-sectional U-shaped density profile was compiled for each IB specimen. The minimum core density was taken as the minimum density of the middle 6-mm portion of each specimen.  29  2.2.4 Measurement of SWR Screw A was a 25.4 mm No. 10 Type AB sheet metal screw with 16 threads per inch (tpi) specified for screw pull tests in ASTM D 1037 (2000), and screw B was a 25.4 mm No. 10 T304 (18-8) sheet metal screw with 10 tpi. Screw B was selected since it was of the same length and diameter as screw A, but had the fewest threads per inch for this particular screw diameter. The exact screw specified for use in ASTM D 1037 (screw A) was difficult to obtain from fastener suppliers, and a closely matching screw was chosen. Root diameter, Dr, was 3.18 mm for screw A and 3.16 mm for screw B, and total shank diameter, Dt, including the threads was 4.73 mm for screw A and 4.86 mm for screw B. The thread height, Dt – Dr, for screw A was 1.55 mm and 1.70 mm for screw B. Photographs and schematic drawing of screws used for screw withdrawal test can be found in Appendix C  Pilot holes measuring 3.2 mm (just over 1/8 in) in diameter (specified by ASTM D 1037) (2000) were first drilled using a drill press. The pilot hole diameter was 68% of the total diameter of screw A and 66% of the total diameter of screw B. For face SWR the hole was drilled through the middle of the test piece. In the case of edge SWR samples, the hole was drilled to a distance of 24 mm mid-way through the specimen cross section and parallel to the surface. The screw was then inserted into the pilot hole and screwed into the sample using an electric hand-held drill until only the top 8 mm of screw shank protruded beyond the edge of the test specimen. Embedment depth for the screws was 17 mm, and measurement of SWR in N was undertaken according to ASTM D 1037 (2000). The design and photograph of the actual screw withdrawal testing apparatus used in the experiment are shown in Appendix D and Appendix E.  30  2.3  Results and discussion  2.3.1 Summary of effects The main effects of press line, machine direction, and screw type on particleboard physical and mechanical properties are summarized in Table 2.3. The effect of press line on panel density and strength properties was highly significant (p < 0.001). Face SWR was affected by screw type, whereas edge SWR was not. Interestingly, the machine direction of the mat had no discernible effect on edge SWR; however, it did influence flexural properties (MOR and MOE) as expected. There were no significant interactions between any of the main effects influencing properties, and so interaction terms are not shown in Table 2.3. Means and coefficients of variation (COV, %) for each measured property for the press lines are given in Table 2.4.  Table 2.3  Main effects on properties of M2 particleboards from six press lines.  effect  density  IB  face SWR  edge SWR  MOR  MOE  press line (P)  p < 0.001  p < 0.001  p < 0.001  p < 0.001  p < 0.001  p < 0.001  mach dir (M)  n.a.  n.a.  n.a.  n.s.  p < 0.001  p < 0.001  screw type (S)  n.a.  n.a.  p < 0.001  n.s.  n.a.  n.a.  n.a. = not applicable for that property. n.s. = not significant at the 5% confidence level.  31  Table 2.4  Means and COV (%) of physical and mechanical properties for boards from press  lines A to F. property density (kg/m3) IB (MPa) Face SWR-A (kN) face SWR-B (N) edge SWR (N) MOR ║ (MPa) MOR ┴ (MPa) MOE ║ (GPa) MOE ┴ (GPa)  unit  press line A  B  C  D  E  F  Mean  681.3  706.9  702.0  657.6  646.7  648.3  COV, %  1.6  4.9  3.0  3.7  3.1  3.0  Mean  0.713  0.427  0.653  0.588  0.576  0.613  COV, %  11.2  18.9  7.6  11.1  9.1  9.0  Mean  1.098  0.837  1.020  1.035  1.016  0.950  COV, %  12.3  11.5  8.8  10.8  7.0  10.6  Mean  1.166  0.883  1.031  1.076  1.026  0.992  COV, %  6.6  11.2  7.0  9.6  6.7  8.4  Mean  972.9  634.4  776.2  732.5  770.8  790.1  COV, %  12.4  17.4  11.9  14.0  10.9  11.8  Mean  16.03  13.50  14.71  16.61  12.31  13.38  COV, %  5.4  16.7  14.7  12.6  6.3  12.6  Mean  15.07  12.59  14.17  15.30  12.24  13.03  COV, %  3.7  14.3  6.6  13.7  11.3  13.8  Mean  3.086  2.336  2.599  2.998  2.553  2.811  COV, %  6.8  15.8  9.2  9.3  8.6  12.1  Mean  2.794  2.153  2.378  2.774  2.404  2.695  COV, %  2.5  11.8  5.5  10.2  8.9  10.9  2.3.2 Physical properties: density, core density, MC, and IB The variation in panel density and IB by press line is shown in Figure 2.2a and b. Mean density of panels from each press line are shown in Figure 2.2a, and ranged from 650 kg/m3 for press line E to 710 kg/m3 for press line B. The MC of samples ranged from 9.83 percent in press line D (COV = 2.6%) to 10.46 percent in press lines B and C (COV = 2.1%). Boards from press line A had the highest IB (averaging 0.73 MPa) and those from B the lowest (averaging 0.42 MPa) (Figure 2.2b). IB of panels from press lines C to F were all similar (0.58 to 0.65 MPa). Overall, 32  the IB of panels from all press lines met the voluntary ANSI A208.1 (1999) standard for M2 grade particleboard of 0.45 MPa, except those from press line B. This was despite the high density of samples from this press line, as can be seen from Figure 2.2a. Note from Table 2.4 the low variation within press lines for panel density. COV is mostly between 2 percent and 5 percent, but is higher in the case of IB; mostly above 9 percent. Variability in IB was also highest for press line B (COV = 18.9%).  The large variation in the IB of panels from different press lines occurred despite the fact that they were of similar density, especially core density. Core density had a narrow range of 530 to 545 kg/m3, except for press line D which had an average minimum core density of 500 kg/m3. These results suggest that the low IB of panels from press line B is not caused by insufficient compaction ratio, which is a common cause of low IB, but by other possible unknown factors such as variation in resin content between press lines, inadequate curing, and/or nonhomogeneous distribution of resin throughout the core furnish. According to Maloney (1993) process-related factors such as inappropriate pressing schedule or insufficient cooling of ureaformaldehyde-(UF) bonded panels can also adversely affect IB.  33  0.8 LSD = 7  oven dry density (kg/m 3)  700 680 660 640 620 600 580  internal bond strength (MPa)  720  LSD = 0.02  0.6 0.5 0.4 0.3 0.2 0.1  560 A (a)  ANSI A208.1 (M2)  0.7  B  C D press line  E  0  F (b)  A  B  C D press line  E  F  Figure 2.2 The (a) oven dry density and (b) internal bond strength of M2 grade boards by press line. n = 80 for each mean. ANSI A280.1 (1999) recommended minimum IB of 0.45 MPa denoted by dashed line.  Variation in IB in boards from the same manufacturer could reflect underlying variability in resin distribution within the furnish material used in the core of the panels, although this is speculative since resin content and formulation in the core furnish is not known. Another factor that could partially explain the variation in bond strength both within and between manufacturers is that environmental regulations of formaldehyde emissions have resulted in the widespread use of less efficient low mole ratio UF resins, making bond quality much more sensitive to physical and chemical variations in the wood feed stock (Gylseth and Maylor 1996, Graves 1999). Variation in particle physical and chemical characteristics arising from different wood species either within or across particleboard plants is a critical source of variation in bond strength and panel strength properties (Kelly 1977, Xu and Suchsland 1998). Inadequate monitoring and control of the species/wood density mix entering a particleboard plant in the form of chips or secondary processing wastes is a also major cause of unacceptably high property variation within panels 34  from the same plant (Cassens et al. 1994, Sjoblom et al. 2004). Wood species used in the particleboards sampled from across Canada in this study ranged from pure spruce to mixtures of hardwoods (birch- Betula alleghaniensis and maple- Acer macrophyllum) and/or softwoods (spruce - Picea sp. and fir - Abies grandis).  2.3.3 Face SWR Since face SWR was significantly affected by screw type (p< 0.001); the mean face SWR values are plotted as a function of press line and screw type in Figure 2.3. Similar to the trend for IB, face SWR was highest for panels from press line A and lowest for press line B. Values for press lines C to F were similar, ranging from 950 to 1050 N. Mean values for screw B (10 tpi) were, without exception, higher than those for screw A (16 tpi). The COV within press lines for face SWR with screw A ranged from 7 to about 12 percent, while the COV for screw B was slightly lower, ranging from 6.6 to 11.2 percent.  The ANSI A208.1 (1999) minimum requirement for face SWR of M2 particleboard is 1000 N. This is for tests made according to ASTM D 1037 (2000) using the No. 10 Type AB screw (25.4 mm long) with a pitch of 16 tpi (i.e., screw A used in our study). All press lines except for B and F met the standard with screw A. Boards from press lines A, C, D, and E had average face SWR over 1000 N with screw B. Face SWR was significantly lower for screw A in the cases of press lines A, B, D, and F; whereby the difference was greater than the LSD of 41 N. The difference between screws A and B was less than 41 N in the cases of press lines C and E. Differences in particle size, geometry, and bonding in particleboards from different press lines is likely to have influenced screw holding behaviour of screws with different thread configurations.  35  1400 LSD = 41  face SWR (N)  1200 ANSI A208.1 (M2)  1000 800 600 A B  A B  A B  A B  A B  A B  A  B  C  D  E  F  400 200 0  press line  Figure 2.3 Face SWR by press line and screw type, averaged across machine direction. n = 40 for each column. ANSI A280.1 (1999) recommended minimum of 1000 N denoted by dashed line.  The findings of this study are in agreement with those of a study by Superfesky (1974) who found that screws with fewer threads i.e., 12 tpi as opposed to 16 tpi, had slightly but consistently higher face SWR. While the embedment depth and screw dimensions in our study were similar to those used by Superfesky (1974), average face SWR loads in his study were higher (1375 N for the 12 tpi screws and 1299 N for the 16 tpi screws). Our values for face SWR correspond better with predicted values from a model based on specific gravity (SG) by Eckelman (1975) for face SWR from softwood particleboard. In that model, the force required to pull a screw from the particleboard face, Fface (lb), is given by:  F face  D  = 2655D  L −  3  0.5  1.25  G 2 (lb)  2.1  where: 36  D = the shank diameter, in. L = the embedment depth in inches, and G = the SG of the sample at a MC of 10%.  The grade of particleboard for which this model was derived was not specified. For example, for a panel with a SG of 0.6 and MC of 10 % at time of testing, Eckelman’s model predicts Fface values of 979 N for Type A (16 tpi) screws and 988 N for Type B (10 tpi) screws. The predicted difference between the two screw types is based solely on the difference in shank diameter, and is minimal at 9 N or 1 %, whereas average observed face SWR for screw B was 3.6 percent higher than screw A. This suggests that much of the difference between the screws in their effect on face SWR may be explained by factors other than shank diameter, such as thread pitch and thread height.  2.3.4 Edge SWR Since there was no significant effect of machine direction or screw type on the edge SWR from sampled particleboards; the results for edge SWR were pooled across machine direction and screw type and plotted as a function of press line only in Figure 2.4. Edge SWR values were approximately 25 % lower than face SWR, in agreement with the findings of Eckelman (2003). The ANSI A208.1 (1999) minimum requirement for SWR (edge) is 900 N. SWR from the edge of particleboard is also more variable than from the face due to the lower density of the core and greater structural heterogeneity from the presence of coarser particles in this layer (Superfesky 1974). Only panels from press line A exceeded the standard value (mean edge SWR = 973 N) while average edge SWR in panels from other press lines were significantly lower; between 630  37  N (press line B) and 790 N (press line F). Note from Table 2.4 the higher variability within press lines for edge SWR (COV values ranging from 11.0% to 17.4% in the case of press line B).  1200 ANSI A208.1 (M2)  edge SWR (N)  1000  LSD = 20.5  800 600 400 200 0 A  B  C  D  E  F  press line  Figure 2.4 Edge SWR by press line, averaged across screw type and machine direction. n = 160 for each column. ANSI A280.1 (1999) recommended minimum of 900 N denoted by dashed line.  The high edge SWR of panels from press line A corresponds with the high IB strength of its core, from Figure 2.2b. Other studies suggest that edge SWR is likely to be a function of both IB and density (Johnson 1967, Eckelman 1973, 1975, Rajak and Eckelman 1993, Wong et al. 1999). Relationships between panel density, IB, and SWR from the data collected in this study will be investigated in a subsequent publication.  Data from the National Particleboard Association (1968), Eckelman (1975), and Barnes and Lyon (1978) suggest that edge SWR in particleboard can be reliably predicted using only SG,  38  screw thickness, and depth of penetration. Eckelman (1975) gives a predictive model for the force required to pull a screw from the edge of a particleboard sample, Fedge (lb), based on its SG as:  Fedge  D  = 2055D  L −  3  0.5  1.25  G 2 (lb)  2.2  where: D = the shank diameter, in. L = the embedment depth in inches, and G = the SG of the sample at a MC of 10%.  Predicted Fedge values for panels of similar SG to our panels, i.e., 0.6, are 757 N for screw type A and 765N for screw type B. These values are below the 973 N for press line A and above the 630 N for press line B, but agree reasonably well with measured edge SWR of the other press lines (C to F).  By comparison, edge SWR values in particleboard measured by Superfesky (1974) were higher: 1107 N for 10 tpi screws (Screw A) and 1119 N for 12 tpi screws. The grade of particleboard used by Superfesky was unspecified, but that work may have used a higher grade of particleboard than the panels used in this study. In agreement with our findings, Superfesky (1974) found no consistent difference between the two screw types for edge SWR, a finding that was attributed to the possible masking effect of greater heterogeneity of structure and bonding within the core region of particleboard. The screw holding ability of particleboard is lower than 39  that of solid wood, plywood, OSB, or MDF. For example, the SWR of particleboard is around 40 percent of that for hardwood MDF (Eckelman 2003). Results from Erdil et al. (2002) for Douglas-fir plywood of the same thickness, screw type, and embedment depth as used in this study gave edge SWR values of approximately 1802 N. Values for No. 10 AB screws in the edge of 19-mm-thick OSB were on average 1686 N (Erdil et al. 2002), although they used an embedment depth of 25.4 mm, compared with only 17 mm used here.  2.3.5 Flexural strength (MOR and MOE) As expected, machine direction had a significant effect (p < 0.001) on MOR and MOE (Table 2.3). The means for each press line and machine direction are compared in Figure 2.5a for MOR and Figure 2.5b for MOE. The ANSI A208.1 standard minimum values for MOE and MOR of M2 grade particleboard are 2250 MPa and 14.5 MPa, respectively. Most panels exceeded the requirement for MOE, but not MOR (except for those from press lines A and D). The variability in MOR and MOE differed by press line (Table 2.4). Variation was smallest for press line A, COV ranged from 2.5 to 6.8 % for MOR and MOE, whereas for press line B the COV for the same properties ranged from 11.8 to 16.7 %. This higher variability in flexural strength for press line B may be due to the greater within press line variation of density and IB observed for press line B.  40  ANSI A208.1 (M2)  16  4.0 LSD = 0.69  14 12 10 8 6 4  modulus of elasticity (GPa)  modulus of rupture (MPa)  18  3.5  ANSI A208.1 (M2)  LSD = 0.11  3.0 2.5 2.0 1.5 1.0 0.5  2  0.0  0 A (a)  Figure 2.5  B  C D press line  E  F  A (b)  B  C D press line  E  F  Average (a) MOR and (b) MOE by press line and machine direction. n = 80 for  each column. The symbols within each column indicate machine direction (║parallel to and ┴ perpendicular to). ANSI A280.1 (1999) recommended minimum of 14.5 for MOR and 2.25 GPa for MOE denoted by dashed lines.  It was observed that the mean MOR of panels from press line B was not very different to other press lines despite their lower IB. It is speculated that a faster press closure rate may have been used in the manufacture of the panels sampled from press line B. It is well known that faster press closure rate improves bending strength but is at the expense of IB strength and screw holding ability (Bismarck 1974, Rice et al. 1967). This is because press closure rate, in combination with surface and core furnish MC, determines the shape of the vertical density profile (Kelly 1977, Wang and Winistorfer 2000, Dai and Wang 2004). Vertical density profiles will be presented and discussed in a subsequent paper in which MS and M2 grade particleboards from two press lines are compared.  41  2.4  Conclusions  M2 furniture grade particleboards produced on six different press lines varied significantly in physical and mechanical properties; the main findings from the survey were:  1.  All press lines except for one exceeded ANSI 208.1 (1999) requirements for IB of 0.45 MPa or greater.  2.  The face SWR of most press lines was just equal to or below the ANSI minimum of 1000 N, except for one which exceeded this value by 10%. Screw type (10 or 16 tpi) had a significant effect on withdrawal resistance. SWR was inversely proportional to thread pitch with the 10 tpi screws having higher face SWR than those with 16 tpi.  3.  Five of the six press lines failed to meet the ANSI minimum of 900 N for edge SWR. Edge SWR was not affected by thread pitch.  4.  Only two press lines exceeded the ANSI minimum of 14.5 MPa for MOR, but five of the six press lines exceeded the minimum of 2.25 GPa for MOE.  5.  Panels from one press line were consistently lower in all strength properties despite having high average density. The variation in properties for this press line was also higher than for other press lines.  42  2.5  Literature cited  American National Standards Institute (ANSI). 1999. ANSI A208.1 Particleboard. 11 pp. American Society for Testing and Materials (ASTM). 2000. Annual book of ASTM standards, Section 4. ASTM D-1037 Construction (Wood). Vol. 04.01. Barnes, H.M. and D.E. Lyon. 1978. Fastener withdrawal loads for weathered and unweathered particleboard decking. Forest Prod. J. 28(4): 33-36. Bautista, C.Q. 1994. Particleboard industry properties survey. Part I. Survey results. In: Proc. of the 48th Forest Prod. Soc. Annual Meeting, Portland, ME. Panel Products Technology Centre, Temple-Inland Forest Prod. Corp., Diboll, TX. 22 pp. Bismarck, C. 1974. Optimizing the pressing of particleboards. The manufacture of particleboards with urea-formaldehyde binders using special automated regulation systems for the pressing process. Holz- Zentralblatt. 100(80):1247-1249. Carroll, M.N. 1970. Relationship between driving torque and screw holding strength in particleboard and plywood. Forest Prod. J. 20(3):24-29. Cassens, D.L., J.P. Bradtmeuller, and F. Picado. 1994. Variation in selected properties of industrial grade particleboard. Forest Prod. J. 44(10):50-56. Eckelman, C.A. 1973. Holding strength of screws in wood and wood based materials. Agr. Res. Bulletin 85. Purdue Univ. 15 pp. Eckelman, C.A. 1975. Screwholding performance in hardwoods and particleboard. Forest Prod. J. 25(6):30-35.  43  Eckelman, C.A. 2003. Chapter 6: Strength of screws in wood composites. In: Textbook of Product Engineering and Strength Design of Furniture. Purdue Univ., School of Agriculture. pp. 56-67. Erdil, Y.Z., J. Zhang, and C.A. Eckelman. 2002. Holding strength of screws in plywood and oriented strand board. Forest Prod. J. 52(6): 55-62. Graves, G. 1999. Urea formaldehyde resins: Yesterday, today and tomorrow. In: Proc. of the 1998 Resin & Blending Seminar, 29-30 Oct., Portland OR. J. Bradfield, Ed. Composite Panel Assoc., Gaithersburg, MD. pp. 3-10. Gylseth, B. and R. Maylor. 1996. New melamine modified binders for low formaldehyde emission wood panels. Asian Timber, 9:52-55. Johnson, J.W. 1967. Screw-holding ability of particleboard and plywood. Forest Research Lab. Rept. No. T-22, School of Forestry, Oregon State Univ., Corvallis, OR. Kelly, M.W. 1977. Critical literature review of relationships between processing parameters and physical properties of particleboard. Gen. Tech. Rep. FPL 10.  USDA Forest Serv.,  Forest Prod. Lab., Madison, WI. 64 pp. Lawes Agricultural Trust. 2001. Genstat –A General Statistical Program 5 Release 4.21. Lawes Rothhamsted Experimental Station, UK. Maloney, T.M. 1993. Modern Particleboard and Dry Process Fiberboard Manufacturing. 2nd ed. Forest Prod. Soc., Madison, WI. 681 pp. National Particleboard Association (NPA). 1968. Screw holding of particleboard. Tech. Bulletin No. 3. NPA, Washington DC. Rajak, Z.I. and C.A. Eckelman. 1993. Edge and face withdrawal of large screws in particleboard and medium density fiberboard. Forest Prod. J. 43(4):25-30.  44  Rice, J.T., J.L. Snyder, and C.A. Hart. 1973. Influence of selected resin and bonding factors on flakeboard properties. Forest Prod. J. 17(8):49-57. Schmidt, H. 1986. Industrial particleboard: Application requirements and special manufacturing features. In: Proc. of the 20th WSU International Symp. T.M. Maloney, Ed. Washing State  Particleboard/Composite  Materials  Univ., Pullman, WA. pp. 343-354.  Sjoblom, E., B. Johnson, and H. Sundstrom. 2004. Optimization of particleboard production using NIR spectroscopy and multivariate techniques. Forest Prod. J. 54(6):71-75. Suchsland, O. and W.S. Good. 1968. The selection of panel materials by furniture and cabinet manufacturers. Michigan Agr. Expt. Sta. J. Article No. 4378. Michigan State Univ., East Lansing, MI. 11 pp. Superfesky, M.J. 1974. Screw withdrawal resistance of types A and AB sheet metal (Tapping) screws in particleboard and medium density fiberboard. Res. Paper No. 239. USDA Forest Serv., Forest Prod. Lab.,Madison, WI. 8 pp. Wong, E.D., M. Zhang, Q. Wang, and S. Kawai. 1999. Formation of the density profile and its effects on the properties of particleboard. Wood Sci. Tech. 33(4):327-340. Wu, Q.L. and R.P. Vlosky. 2000. Panel products: A perspective from furniture and cabinet manufacturers in the southern United States. Forest Prod. J. 50(9):45-50. Xu, W. and O. Suchsland. 1998. Variability of particleboard properties from single and mixedspecies processes. Forest Prod. J. 48(9): 68-74. Zhang, J. 1994. Particleboard industry physical properties survey. Part II. Analytical technique. In: Proc. of the 48th Forest Prod. Soc. Annual Meeting, Portland, ME. Panel Products Technology Center, Temple- Inland Forest Prod. Corp., Diboll, TX. 27 pp.  45  Chapter 3 Properties comparison of furniture grade MS and M2 particleboard products manufactured in Canada2 3.1  Introduction  There are four ‘industrial and shelving’ grades of particleboard manufactured in a range of dimensions and densities tailored for use in furniture applications including cabinets and readyto-assemble (RTA) furniture components (CPA 2003). These panels typically have smooth, densified surfaces for lamination and a lower density core to reduce overall weight. The MS and M2 grades comprise the bulk of laminated particleboard used in RTA furniture components. The lowest grade (M1) is not widely used since there are no established minimum requirements for face or edge screw holding ability, while the highest grade (M3) exceeds the strength and density requirements of the majority of RTA furniture applications.  MS grade product was developed by the particleboard industry for the RTA furniture industry in response to the need for a lower cost, lighter weight panel with slightly reduced strength properties. The minimum required properties of M1, MS, and M2 grade industrial particleboards (ANSI A208.1-1999) are given in Table 3.1. A reduction in panel density at the expense of flexural strength is acceptable to the RTA industry to a certain extent; however, wider adoption of MS, or even M1, grade particleboard is limited by the accompanying reduction in fastener holding strength. This may be a generalization since, as Wu and Vlosky (2000) point out; there 2  A version of this chapter has been published. Semple, K., E. Sackey, H.J. Park, and G.D. Smith. 2005. Properties comparison of furniture grade MS and M2 particleboard products manufactured in Canada. Forest Products Journal 55(12):125–131.  46  has been poor communication of specific customer requirements of particleboard to the industry which has hampered the development of new products specifically tailored to different end-use applications.  Table 3.1  Minimum strength properties of medium (M) grades of industrial and shelving  particleboard (from ANSI A208.1-1999). Note that no density ranges are specified in ANSI A208.1. MOR  MOE  IB  SWR-face  SWR-edge  MPa (psi)  MPa (ksi)  MPa (psi)  N (lb)  N (lb)  M1  11.0 (1595)  1725 (250.2)  0.40 (58)  –  –  MS  12.5 (1813)  1900 (275.6)  0.40 (58)  900 (202)  800 (180)  M2  14.5 (2103)  2250 (326.3)  0.45 (65)  1000 (225)  900 (202)  M3  16.5 (2393)  2750 (398.9)  0.55 (80)  1100 (247)  1000 (225)  grade  While both MS and M2 grade particleboard have been on the market for some years, there is no published data to enable a direct comparison of properties of particleboards marketed as MS or M2 in Canada. Key unresolved questions include how distinct are these two products in terms of density and strength properties? Do they conform to North American standards for their grade, and is the quality of panels marketed as a particular grade consistent between different manufacturers?  In a previous paper, Semple et al. (2005), M2 grade particleboard produced by six different Canadian manufacturing facilities or ‘press lines’ were tested and their physical and mechanical properties were compared. The M2 grade products from five of the six press lines exceeded the voluntary American National Standards Institute (ANSI) A208.1 (1999) minimum requirements for internal bond (IB) strength, but there was greater variation among press lines and a lower 47  compliance rate in the case of flexural strength (modulus of rupture [MOR]), and face and edge screw withdrawal resistance (SWR). Similarly, a survey of U.S.-made 3/8- inch-thick M2 grade particleboards by Cassens et al. (1994) found that while IB was relatively consistent across suppliers, there was significant variation among suppliers in other mechanical properties. Two of the seven suppliers surveyed by Cassens et al. (1994) did not meet minimum ANSI (1999) standards.  For this study, MS grade particleboard was obtained from only two of the six manufacturers that supplied M2 grade panels for the properties survey in the previous paper (Semple et al. 2005). This paper reports on the comparison of properties between two furniture grades (MS and M2) of 5/8-inch- (15.87- mm-) thick particleboards produced by the same two press lines. Differences between the two grades in vertical density distribution are also evaluated and discussed.  The objectives of this study were to: 1. Compare density, flexural strength, IB strength, and face and edge SWR of MS and M2 grade particleboard products, both of which were manufactured by two different mills or press lines’ and 2. Test whether the properties of products marketed as MS or M2 conform to ANSI A208.1 standards and whether properties of each grade are similar regardless of where they were manufactured.  48  3.2  Materials and methods  3.2.1 Sample procurement and preparation Five particleboard panels measuring 4 by 8 feet (1.22 by 2.44 m) of two furniture grades, MS and M2, were randomly sampled from two Canadian companies that manufacture both grades of particleboard. Each of the 20 panels was divided into eight sub-panels measuring 2 by 2 feet (610 by 610 mm) from which the same set of 12 test specimens were cut according to randomized templates illustrated in Figure 3.1 (two of which are shown). To avoid any location bias, the arrangement of the specimens was randomized differently on each sub-panel. Each physical and mechanical property and the number of specimens tested are given in Table 3.2;  2  11  1  12  3  5  6  12 11  8,10  8,10 12 11  3 1  11  7,9  4  7,9  machine direction  note that the specimen ID corresponds to the ID numbers on Figure 3.1.  2  5  1  4 12  6  2  Figure 3.1 Cutting patterns for the first two sub-panels showing randomized test specimen positions. Specimen numbers correspond to the specimen IDs given in Table 3.2.  49  Table 3.2  Mechanical properties, specimen ID and the number of samples for M2 and MS  particleboards from two press lines. property  specimen ID  number of samples per press line/grade  panel  sub-panel  edge SWR (machine direction) Screw A  1  40  8  1  edge SWR (perpendicular) Screw A  2  40  8  1  edge SWR (machine direction) Screw B  3  40  8  1  edge SWR (perpendicular) Screw B  4  40  8  1  face SWR Screw A  5  40  8  1  face SWR Screw B  6  40  8  1  MOR (machine direction)  7  40  8  1  MOR (perpendicular)  8  40  8  1  MOE (machine direction)  9  40  8  1  MOE (perpendicular)  10  40  8  1  internal bond  11  80  16  2  oven-dry density  12  80  16  2  3.2.2 Experimental design and statistical analysis The experimental design consisted of five replicate panels of two grades (MS and M2) and two press lines (A and B) for a total of 20 panels. Due to lack of control over which press lines were sampled, ‘press line’ as a main effect is considered to be a random factor. The other main effects including grade (two levels), machine direction (two levels), and screw type (two levels) are considered fixed factors. The treatment structures indicating the fixed and random factors for each physical and mechanical property are given in Table 3.3.  50  Table 3.3  Treatment structures for board physical and mechanical properties.  property  specimen ID  edge SWR  1-4  fixed factors  levels  grade  M2, MS  machine direction ║, ┴ screw type  face SWR  MOR, MOE  IB, density  5-6  7 – 10  11, 12  A, B  random factors  levels  press line  A to F  panel  1 to 5  sub panel  1 to 8  specimen  1 to 4  press line  A to F  grade  M2, MS  panel  1 to 5  screw type  A, B  sub panel  1 to 8  specimen  1, 2  press line  A to F  panel  1 to 5  sub panel  1 to 8  specimen  1, 2  press line  A to F  panel  1 to 5  sub panel  1 to 8  specimen  1, 2  grade  M2, MS  machine direction ║, ┴  grade  M2, MS  Each treatment structure involved a hierarchical design with one or more fixed factors depending on the response variable tested. In the case of density, IB, and face SWR, data was modeled as a split-plot design with panel replicate as the treatment block, i.e., specimens (8 for face SWR, 16 for density and IB) within each of eight sub-panels within each of five replicate panels for each press line. Edge SWR and MOR/MOE had the extra effect of machine direction and were modeled using a split-split plot design, i.e., eight specimens for each of two machine directions within each of eight sub-panels within each of the five replicate panels. These two treatment structures remained the same for each press line and grade. The order of sample testing was blocked by replicate (panel), i.e., for each mechanical property the samples from all the press lines for replicate 1 were tested in random order followed by all those for replicate 2, and so on. 51  The order in which the sub-panels were tested was also randomized for each panel replicate and press line.  Statistical analysis for each property was undertaken using an appropriate analysis of variance (ANOVA) model at the 5 percent significance level using Genstat 5 release 4.21 (Lawes Agricultural Trust 2001). The data sets were first checked to ensure normal distributions and similar variances. The main effects of press line, grade, screw type, and machine direction were tested for significance (p ≤ 0.05) and any significant interactions between these factors were examined. Significant effects or interactions for press line, grade, screw type, and machine direction (where applicable) are compared graphically and the Least Significant Difference (LSD) for p ≤ 0.05 is included on graphs to facilitate comparison of means.  3.2.3 Specimen testing The testing of density, IB, face and edge SWR, MOR, and MOE was carried out in accordance with ASTM D 1037 (2000) and has been described in the previous paper (Semple et al. 2005) which compared properties of M2 grade particleboard across six different press lines.  3.3  Results and discussion  3.3.1 Main effects and interactions The significant main effects of grade (G), press line (P), screw type/machine direction (where applicable), and the G × P interaction on properties are given in Table 3.4. None of the other two-way or higher order interactions between factors was significant and are therefore not shown in Table 3.4. 52  Overall, the strength properties of M2 grade panels were significantly higher (p < 0.001) than those of MS grade from both press lines. The G × P interaction indicates that there was less distinction between the two grades of particleboard manufactured on one of the press lines. This effect can be clearly seen in Figure 3.2a whereby the difference between the density of MS and M2 panels is less for press line B than press line A. Means and coefficients of variation (COV, %) for each measured property for the press lines are given in Table 3.5  Table 3.4  Significant main effects and interactions for properties of MS and M2 grade  particleboard from two press lines. Effects or  density  IB  face SWR  edge SWR  MOR  MOE  grade (G)  p < 0.001  p < 0.001  p < 0.001  p < 0.001  p < 0.001  p < 0.001  press line (P)  p < 0.001  p < 0.001  p < 0.001  p < 0.001  p < 0.001  p < 0.001  screw type (S)  n.a.  n.a.  p < 0.001  n.s.  n.a.  n.a.  machine dir.(M)  n.a.  n.a.  n.a.  n.s.  p < 0.001  p < 0.001  GxP  p < 0.001  p = 0.004  p = 0.04  p < 0.001  p < 0.001  p = 0.012  interactions  n.a. = not applicable for that mechanical property. n.s. = not significant at the 5% confidence level; p < 0.001 = significant at the 0.1% level.  53  0.8  700  press line  680  B  0.7 internal bond strength (MPa)  oven dry density (kg/m 3 )  720  660 640 620  A  ANSI A208.1 (M2)  0.4 0.3  ANSI A208.1 (MS)  B  0.2 LSD = 0.023 0.1  580  0 MS 1  Figure 3.2  0.5  A  LSD = 8.5  600  (a)  0.6  press line  1 MS  M2 2 grade  (b)  2 M2 grade  The (a) oven dry density and (b) internal bond strength of MS and M2 grade  panels by press line. ANSI A280.1 (1999) recommended minimum IB of 0.40 MPa for MS and 0.45 MPa for M2 are denoted by dashed lines. Each mean represents 80 replicates tested.  54  Table 3.5  Means and COV of physical and mechanical properties of M2 and MS boards for  two press lines. press line and  M2 A  MS A  M2 B  MS B  Mean (kg/m3)  681.0  633.5  706.7  684.4  COV (%)  1.3  3.4  2.7  3.8  Mean (MPa)  0.70  0.58  0.43  0.34  COV (%)  11.2  14.5  18.9  14.0  Mean (N)  1098.2  956.6  837.2  729.1  COV (%)  12.3  9.6  11.5  12.8  Mean (N)  1166.1  987.4  883.1  748.2  COV (%)  6.6  8.7  11.2  13.5  Mean (N)  972.9  763.8  634.4  555.6  COV (%)  12.4  12.9  17.4  15.9  Mean (MPa)  16.0  13.3  13.5  12.3  COV (%)  5.4  8.3  16.7  18.5  Mean (MPa)  15.0  12.9  12.6  10.7  COV (%)  3.7  7.7  14.3  15.3  Mean (GPa)  3.1  2.7  2.3  2.3  COV (%)  6.8  8.0  15.8  14.0  Mean (GPa)  2.8  2.4  2.2  2.0  COV (%)  2.5  6.8  11.8  13.6  grade density IB face SWR Screw A face SWR Screw B edge SWR MOR ║ MOR ┴ MOE ║ MOE ┴  COV = coefficient of variation (ratio of the standard deviation to the mean in percent).  3.3.2 Moisture content, density, and IB strength The conditioned moisture contents (MC) of samples at time of testing ranged from an average of 10.0 percent (COV = 2.5%) for MS grade panels from press line A to 10.5 percent (COV = 2.2% for M2 grade from press line B). The mean values for ovendry density and IB strength of M2 and MS grade panels are compared in Figure 3.2a and Figure 3.2b, respectively. As expected, the MS grade product was significantly lower (p < 0.001) in density and IB than its M2 grade 55  counterpart. The most important result to note from Figure 3.2 is that both grades of panel from press line B were higher in overall density (Figure 3.2a), but lower in IB strength (Figure 3.2b) than those from press line A. The average density of MS grade panels from press line A was 633 kg/m3 and for press line B, 685 kg/m3, whereas the corresponding average IB of MS grade was 0.58 MPa for press line A and only 0.34 MPa for press line B. Note from Table 3.5 that within press line variation in IB was greater than for density. COV for density among press line and grade groups ranged from 1.3 percent for M2 A to 3.8 percent for MS B. COV for IB ranged from 11.2 percent for M2 A to 18.9 percent for M2 B. The discrepancy between panel density and IB is further highlighted when IB is plotted as a function of density (Figure 3.3a and b).  LSD IB = 0.023 LSD ODD = 6.02  0.75 0.70  0.80 internal bond strength (MPa)  internal bond strength (MPa)  0.80  M2 A  0.65 0.60  MS A  0.55 0.50  ANSI A208.1 (M2)  0.45  M2 B  ANSI A208.1 (MS)  0.40  0.75 M2 A  0.70 0.65  LSD CD = 5.13  0.60  MS A  0.55 0.50 0.45  M2 B  0.40  0.35  MS B  0.35  0.30  MS B  0.30 500  550  600  (a)  650 oven dry density (kg/m 3 )  700  Figure 3.3  Mean internal bond strength as a function of (a) mean oven dry density, and (b)  500 (b)  550  600  650  700  core density (kg/m 3 )  mean core density, for MS and M2 boards from press lines A and B. Each mean represents 160 samples.  These figures show that IB is proportional to density only for MS and M2 products made on the same press line. The minimum IB strength required for MS grade particleboard specified by ANSI A208.1 (1999) is 0.40 MPa and for M2 grade panels, 0.45 MPa. The MS grade product 56  from press line A greatly exceeded this standard (mean IB = 0.58 MPa), while that from press line B did not meet it (mean IB = 0.33 MPa). The core density of the MS and M2 grade panels, Figure 3.3b, is much lower than the whole board density shown in Figure 3.2a and Figure 3.3a and ranges from 510 to 540 kg/m3.  3.3.3 Face SWR As indicated in Table 3.5, face SWR was significantly affected (p < 0.001) by screw type, with screw B requiring more force to withdraw the screw from the sample than for screw A. The effects of screw thread pitch on face SWR in particleboard was discussed in the previous paper dealing with M2 grade particleboard; our results are in accord with those of previous studies (Superfesky 1974) where screws with a coarser thread pitch (fewer threads per unit length) tend to have higher face withdrawal loads.  Mean face SWR for MS and M2 grade panels from press lines A and B for each screw type, are shown in Figure 3.4. Overall, the results for face SWR were similar to the IB results. Both grades of panel from press line B were significantly lower in face SWR than those from A, and note also from Figure 3.4 that the difference between grades for press line B was less distinct than for press line A. This caused a small, but significant G × P interaction (p = 0.04) for face SWR as indicated in Table 3.4. Minimum face SWR recommended in ANSI A208.1 (1999) for MS grade particleboard is 900 N. MS grade panels from press line A met the standard (mean face SWR values were 957 N, COV= 9.6% for screw A and 987 N, COV= 8.7% for screw B), whereas MS grade from press line B did not (mean face SWR was 729 N, COV = 12.8% for screw A and 748 N, COV = 13.5% for screw B).  57  1200  LSD = 40.9 ANSI A208.1 (M2) ANSI A208.1 (MS)  Screw B  Screw A  Screw B  Screw A  Screw B  400  Screw A  600  Screw B  800 Screw A  face SWR (N)  1000  200 0 MS A  M2 A  MS B  M2 B  grade and press line  Figure 3.4  Mean face screw withdrawal strength by screw type for MS and M2 grade boards  from press lines A and B. ANSI A280.1 (1999) recommended minimum of 1000 N for M2 and 900 N for MS are denoted by dashed lines. Each column mean represents 40 replicates.  3.3.4 Edge SWR Edge SWR was not affected by either screw type or machine direction, and mean values for edge SWR were obtained from the pooled results for screw type and machine direction for both grades and press lines. Mean edge SWR for grades and press lines are shown in Figure 3.5. In keeping with the trend for other properties, the MS grade product was significantly lower (p < 0.001) in edge SWR than the M2 grade from both press lines, with the difference between the two grades less distinct in the case of press line B. Nevertheless, the M2 grade product from press line B was still significantly higher (p < 0.001) than the MS grade. The minimum ANSI A208.1 (1999) standard for edge SWR for MS grade particleboard is 800 N. Mean edge SWR values for MS grade panels from both press lines were below 800 N (i.e., 764 N, COV = 12.9% for press line A  58  and 555 N, COV = 15.9% for press line B). Low screw holding capacity of MS grade particleboard may be one of the key issues hampering the wider adoption of this lighter weight grade of particleboard by the RTA furniture industry.  1200 LSD = 22.0  edge SWR (N)  1000  ANSI A208.1 (M2) ANSI A208.1 (MS)  800 600 400 200 0 MS A  M2 A  MS B  M2 B  grade and press line  Figure 3.5  Edge SWR by press line and grade, averaged across screw type and machine  direction. ANSI A280.1 (1999) recommended minimum of 900 N for M2 and 800 N for MS are denoted by dashed lines. Each column mean represents 160 replicates.  3.3.5 Flexural strength (MOR and MOE) Since machine direction of the panel has a significant effect on its flexural strength properties, mean MOR (Figure 3.6a) and MOE (Figure 3.6b) values are plotted for MS and M2 panels from press lines A and B for both machine directions. Samples tested parallel to the machine direction of the mat were significantly higher (p < 0.001) for MOR and MOE than those tested perpendicular, with the values parallel to the machine direction approximately 15 percent higher 59  than those in the perpendicular direction. The ANSI A208.1 standard minimum requirement of MS grade particleboard is 12.5 MPa for MOR and 1.9 GPa for MOE. MS grade panels mostly met the standard requirements; however, mean MOR of MS panels from press line B was below 12.5 MPa.  18  ANSI A208.1 (M2)  modulus of elasticity (GPa)  modulus of rupture (MPa)  16  3.5  LSD = 0.65 ANSI A208.1 (MS)  14 12 10 8 6 4  ANSI A208.1 (M2)  LSD = 0.10  3.0 ANSI A208.1 (MS)  2.5 2.0 1.5 1.0 0.5  2  0.0  0 MS A (a)  Figure 3.6  M2 A  MS B  grade and press line  M2 B  MS A (b)  M2 A  MS B  M2 B  grade and press line  Mean values for (a) MOR and (b) MOE for MS and M2 grade panels from press  lines A and B; ║ = parallel to machine direction, ┴ = perpendicular. ANSI A280.1 (1999) recommended minimum MOR of 14.5 MPa (M2), 12.5 MPa (MS) and MOE of 2.25 GPa (M2) and 1.9 GPa (MS) are denoted by dashed lines. Each column mean represents 40 replicates. Note from Figure 3.6b that M2 and MS grade panels from press line B were not markedly different in MOE. The difference between the means for the two grades in || MOE was not significant, i.e., less than the LSD of 0.1 GPa. As can be seen from Table 3.5, the COVs for MOR and MOE within press line and grade groups were higher for the panels from press line B.  60  3.3.6 Vertical density profiles The establishment of a density gradient between surface and core or vertical density profile of particleboard during pressing is critical to most properties, particularly bending strength and face SWR (Plath and Schnitzler 1974, Geimer et al. 1975, Kelly 1977, Wong et al. 1999).Variation in the density gradient through the cross-section of particleboard strongly affects flexural properties; MOR can vary by up to 80% in panels of the same mean density but containing different density profiles (Shen and Carroll 1970).  Typical vertical density profiles from M2 and MS grade panels are shown Figure 3.7a for press line A and Figure 1.7b for press line B. The core density, as shown in Figure 1.3b, was more consistent across grades than surface density, especially in the case of panels from press line B, where average core density ranged from about 530 to 545 kg/m3. There was a greater reduction in surface density of MS grade panels, which was approximately 6 percent lower than the M2 grade for both press lines. In the case of panels from press line A (Figure 3.7a), the density profile of the MS grade generally did not differ greatly from that of the M2 grade; except that for the MS grade panels there was a downward shift in density of approximately 100 kg/m3 at all points through the thickness. Surface density was around 1000 kg/m3 for M2 grade and 900 kg/m3 for MS, while core densities for the M2 and MS grade panels were around 600 kg/m3 and 500 kg/m3, respectively. Panels from press line A decreased in density from the surfaces through to about 4 mm deep whereupon core density remained fairly consistent throughout the middle 8 mm of the cross-section.  61  1200  1200  press line B  press line A  1000 density (kg/m 3)  density (kg/m 3)  1000 800 M2  600  MS  400  M2  600  MS  400 200  200 0  800  0  (a)  Figure 3.7  2  4  6 8 10 12 depth (mm)  0  14 16 (b)  0  2  4  6  8  10 12  14 16  depth (mm)  Examples of typical vertical density profiles from M2 and MS specimens from (a)  press line A, and (b) press line B.  In contrast, panels from press line B (especially the M2) had more V-shaped density profiles as shown in Figure 3.7b, with core density reaching a minimum only near the center of the crosssection. MS grade panels from press line B were lower in surface density (average = 881 kg/m3) than press line A (average surface density = 933 kg/m3), whereas core density was similar across panels from both press lines, which can be seen in Figure 3.3b and Figure 3.7. Reduced surface density is often caused by resin pre-cure and results in excessive springback with press opening (Dai and Wang 2004). Excessive springback requires more of the surface material to be removed by sanding to bring the final product to within required thickness tolerance limits.  The vertical density profile established through the thickness of reconstituted wood panels is strongly dependent on the complex interaction of furnish MC and the press closure rate (Geimer et al. 1975, Kelly 1977, Andrews et al. 2001). Differences in furnish and/or pressing schedules may explain the observed differences in vertical density profile between plants A and B. Greater moisture in the furnish results in faster densification than drier furnish and produces a board with 62  higher density faces relative to the core (Kelly 1977). Faster press closure rates have the effect of increasing the bending strength of particleboards by changing the vertical density gradient to more of a V-shape, i.e., continuously decreasing density from surface to core (Rice et al. 1967, Bismarck 1974, Geimer et al. 1975). However, if the density gradient in particleboard becomes too steep then shear failure between high and low density layers may occur during loading which will reduce the flexural strength (Kawai and Sasaki 1986).  A steeper vertical density profile in particleboard often results in lower IB and screw holding in the lower density core (Geimer 1975, Kelly 1977). It was observed in the assessment of M2 grade panels from six press lines in Part I of the properties survey (Semple et al. 2005) that panels from press line B had lower IB and SWR, despite being similar in density and flexural strength to panels from other press lines. This in combination with their more V-shaped density profiles seem to suggest that a faster press closure rate on press line B may have been used in the manufacture of these panels.  3.4  Conclusions  There was significant variation in the properties of particleboard marketed under the same grade designation, either MS or M2 depending on the manufacturer.  1.  Panels from one press line were higher in density but lower in all strength properties than those from the other press line. Within each press line, however, the MS grade panels were lower in both density and strength properties than the M2 grade.  63  2.  There was a small but significant effect of screw thread pitch on face SWR, whereby screws of 10 threads per inch had higher withdrawal loads. Both MS and M2  grades  from press line A exceeded their respective ANSI A208.1 minimum recommended face SWR of 1000 and 900 N, respectively; whereas panels from press line B did not.  3.  Only M2 grade panels from press line A complied with the ANSI A208.1 standard  for  edge SWR in furniture-grade particleboard. MS grade from both press lines were below the 800 N in edge SWR recommended for this grade, which could affect its suitability for use in furniture components that are screwed together.  4.  Both grades of panels from press line A met their respective ANSI A208.1 recommendations for MOR and MOE, whereas those from press line B were borderline or below. There was little difference in MOE between the MS and M2 grade products from press line B.  5.  MS and M2 grade panels from press line A had similar shallow U-shaped vertical density profiles, except the density of the MS grade was shifted downward by about 100 kg/m3. M2 grade from press line B had a steeper, more V-shaped vertical density profile, while the surface density in the MS grade was markedly reduced to leave a flatter profile. These differences may have been the result of different pressing strategies between the press lines.  64  6.  Despite the fact that MS grade particleboard is formulated and distributed as a more cost effective, lighter weight product with lower minimum standard properties than the M2 grade, the results from this survey suggest that M2 and MS grade particleboards from different manufacturers are not standardized products on the market. MS panels from one supplier may be equivalent in strength properties to the M2 grade of another.  65  3.5  Literature cited  Andrews, C.K., P.M. Winistorfer, and R.M. Bennett. 2001. The influence of furnish moisture content and press closure rate on the formation of the vertical density profile in oriented strand board. Forest Prod. J. 51(5):32-39. American National Standards Institute (ANSI). 1999. ANSI A208.1 Particleboard. 11 pp. American Society for Testing and Materials (ASTM). 2000. Annual book of ASTM standards, Section 4. ASTM D 1037 Construction (Wood). Vol. 04.01. Bismarck, C. 1974. Optimizing the pressing of particleboards: The manufacture of particleboards with urea-formaldehyde binders using special automated regulation systems for the pressing process. Holz-Zentralblatt. 100(80):1247-1249. Cassens, D.L., J.P. Bradtmeuller, and F. Picado. 1994. Variation in selected properties of industrial grade particleboard. Forest Prod. J. 44(10):50-56. Composite Panel Association (CPA). 2003. Buyers and Specifiers Guide to North American Particleboard, MDF and Hardboard/Fiberboard Manufacturers and Products.  CPA,  Gaithersburg, MD. 15 pp. Dai, C.P. and S. Wang. 2004. Press control for optimized wood composite processing and properties. Part I. Pressing variables and sensors. In: Proc. of the Fundamentals of Composite Processing Conf., Nov. 5-6, 2003, Madison, WI. J.E. Windandy and  F.A.  Kamke, Eds. pp. 54-64. Geimer, R.L., H.M. Montrey, and W.F. Lehmann. 1975. Effects of layer characteristics  on the  properties of three layer particleboards. Forest Prod. J. 25(3):19-29. Kawai, S. and H. Sasaki. 1986. Production technology for low-density particleboard I: Forming a density gradient and its effect on board properties. Mokuzai Gakkaishi. 32(5):324-330. 66  Kelly, M.W. 1977. Critical literature review of relationships between processing parameters and physical properties of particleboard. Gen. Tech. Rep. FPL-10. USDA Forest Serv., Forest Prod. Lab., Madison, WI. 64 pp. Lawes Agricultural Trust. 2001. Genstat –A General Statistical Program 5 Release 4.21. Lawes Rothhamsted Experimental Station, UK. Plath, L. and E. Schnitzler. 1974. The density profile: A criterion for evaluating  particleboard.  Holz als Roh- und Werkstoff. 32(11): 443-449. Rice, J.T., J.L. Snyder, and C.A Hart. 1973. Influence of selected resin and bonding factors on flakeboard properties. Forest Prod. J. 17(8):49-57. Semple, K.E., E. Sackey,  H.J. Park, and G.D. Smith. 2005. Properties variation study of  furniture grade M2 particleboard manufactured in Canada. Forest Prod. J. 55(12):117124. Shen, K.C. and M.N. Carroll. 1970. Measurement of layer strength distribution in particleboard. Forest Prod. J. 20(6):53-55. Superfesky, M.J. 1974. Screw withdrawal resistance of types A and AB sheet metal (Tapping) screws in particleboard and medium density fiberboard. Res. Paper No. FPL 239. USDA Forest Serv., Forest Prod. Lab., Madison, WI. 8 pp. Wong, E.D., M. Zhang, Q. Wang, and S. Kawai. 1999. Formation of the density profile  and its  effects on the properties of particleboard. Wood Sci. Tech. 33(4):327-340. Wu, Q.L. and R.P. Vlosky. 2000. Panel products: A perspective from furniture and manufacturers in the southern United States. Forest Prod. J. 50(9):45-50.  67  cabinet  Chapter 4 Empirical distribution models for slenderness and aspect ratios of the core particles of particulate wood composites3  4.1  Introduction  Physical and mechanical properties of wood composite products are dependent on the properties of the individual constituents, the manufacturing process, and the resulting structure from those constituents. Earlier studies on particleboard have suggested that length to thickness ratio, or slenderness ratio (SR), is a better indicator of the effect of particle shape on modulus of rupture (MOR) than the individual dimension. Increasing length to width ratio, or Aspect Ratio (AR), reduces MOR, but increases screw withdrawal resistance (SWR) (Post 1961, Kusian 1968, Lehmann 1974, Lin et al. 2002). However, the inter-relationship between particle size distribution (PSD), particle descriptors (AR and SR), compaction ratio, and mat compression has not been fully investigated.  Particleboard and flakeboard furnish have traditionally been characterized by screen fractionation and manual caliper measurement (Geimer et al. 1999). Although caliper measurement is a direct method, it is a tedious and labour intensive process for a large particle numbers. However, it is very useful for smaller samples in a laboratory setting for analysis of particle dimensions and distributions. Geimer and Link (1988) used a sonic digitizer and 3  A version of this chapter has been published. Sackey, E.K. and G.D. Smith .2009. Empirical distribution models for slenderness and aspect ratios of core particles of particulate wood composites. Wood and Fiber Science, 41(3) 255-266.  68  micrometer to characterize flakeboard furnish. An image analysis (IA) instrument, the Cambridge Quantinet 970, was also used by Geimer et al. (1999) to measure the dimensions of individual flakes. Currently, IA systems consist of a microscope, a CCD camera, a computer, and IA software. These systems have the advantage of shorter processing times and online automated measurements for relatively larger amounts of particle furnish. Nevertheless, the IA technique only quantifies the shape of the projected area of particles on a two dimensional plane (Allen 1981). Particles tend to fall onto a horizontal surface with their widest face parallel to that surface and as a result IA measurements overestimate the true particle size. As such, automated measurement of SRs requires a different measuring technique. In the recent past particle analysis systems based on laser diffraction (LD) technology has become available and popular (Khalili et al. 2002, Li et al. 2005). However, LD does not measure individual particles and one must have prior morphological information of the particles such as shape or size to correctly interpret LD results (Kelly et al. 2006).  Studies of distribution models in the wood products sector have used normal, lognormal, and the Erlang family of distributions to characterize fiber length distributions (Yan 1975, Kropholler and Sampson 2001). In an earlier study Geimer et al. (1999) systematically characterized flakeboard furnish using geometrical descriptors and cumulative distribution curves and panel properties were modeled using these geometrical descriptors. Lu et al. (2007) recently used the Weibull and lognormal models for fibers of medium density fiberboard (MDF) and core particleboard furnish. They found that the lognormal was the best fit for short fibers while the Weibull was the best fit for long fibers (α = 0.05). In a recent study, Cao and Wu (2007) used mixture and segmented distributions to describe wood fiber and particle length. It must be noted  69  that the number of particle samples for the study by Lu et al. (2007) was limited to two 100g bag of particles from a single source and those of Cao and Wu (2007) were from two different sources. Since in hammer milled particleboard furnish, there is a naturally wide distribution, the fitted distribution models in those studies may not describe particles from a larger number of sources. Most of these models have focused on characterizing fiber length with the exception of Geimer et al. (1999), who also examined flakeboard furnish. Since fibres are more cylindrical and needle-like compared with irregular, chunky, 3D particles, these models may not be applicable to particleboard furnish particles. Thus, there is a knowledge gap in the literature concerning the most appropriate distribution models for describing the SRs and ARs of particleboard furnishes. This study proposes to develop empirical distribution models to describe the three most important particle geometry descriptors, i.e., length, SR, and AR and correlate these to the mechanical properties of particulate wood composite panels.  The main objectives are to: 1. Characterize the particle geometry of the core furnish in particleboard in terms of length, width, thickness, slenderness ratio and aspect ratio and relate them to panel properties. 2. Develop distribution models for length, slenderness ratio, and aspect ratio of core particles for commercial particleboard furnish.  4.2  Experimental  4.2.1 Particle preparation The source for the particles used in the study was the thickness swell (TS) samples from an earlier study by Semple et al. (2005). Three TS samples from three different panels were 70  procured from each of six different particleboard plants across Canada, for a total of 54 samples. The particles in each sample were separated by cooking each sample in water in a pressure vessel for 30 minutes to hydrolyze and dissolve the urea resin from the furnish particles. The separated particles were then filtered using a 150 microns (0.0059 in) mesh opening sieve to remove water. The small mesh size sieve was chosen to minimize the number of small particles from being washed out of the wet furnish; however, the loss of the very fine wood flour was unavoidable. The wet particles were dried at 70oC for 1.5 hours with interval stirring of the particles every 15 minutes to avoid agglomeration. Dried samples were then conditioned at 65% RH and 20oC for two weeks and bagged according to plant. An issue with this process is that the original particle dimensions prior to their incorporation into the panel may be affected by the hysteresis that occurs in desorption process when the furnish was rewetted and then re-dried. Some particles might also have collapsed cell walls during hot pressing and cannot reach their original size (Mahoney 1980, Wolcott et al. 1994). Since this is a comparative study of panels from different plants, this will be present in all samples regardless of plant.  4.2.2 Particle screening, classification, and dimensions The moisture content of 200 g samples taken from each bag was measured and recorded. Particles were screened with a Ro-Tap (Model RX 29) sieve shaker into seven different size classes as shown in Table 4.1. The mesh size range was selected in order to separate the face from core furnish and to obtain more size classes within the core particles. The mesh sizes used here were similar to those used in the production of multilayer particleboard (Maloney 1970, Eusebio and Generalla 1983).  71  Table 4.1 ASTM sieve number  Mesh sizes used for particle classification Tyler mesh  Sieve opening  type  mm  in  5  5 mesh  4.00  0.1570  10  9 mesh  2.00  0.0787  18  16 mesh  1.00  0.0394  35  32 mesh  0.50  0.0197  60  60 mesh  0.25  0.0098  120  115 mesh  0.125  0.0049  Although particle size classification differs from study to study, industry usually divides particleboard furnish into surface fines for panel faces and coarse particles for the core. For this study, particles that pass through a 0.5 mm mesh were considered to be surface fines and those remaining on the screen as core particles. The core particles were further divided into core-fine (>0.5 mm), medium (> 1 mm) and coarse (> 2 mm) particles (Sackey et al. 2008). Particle dimensions were measured manually using digital calipers.  The dimensions (length and width) of 200 and 300 randomly selected particles for medium and coarse size classes, respectively, were scanned with flatbed scanner and analyzed with image analysis software, while the thickness was measured with calliper and recorded for each panel from each plant. For the core-fine particles the sample size from a replicate panel was reduced to 40 particles as preliminary particle size measurements indicated negligible variation between particles from various plants. Sample size for particle dimensions from each source was, therefore, 120, 600 and 900 for core-fine, medium, and coarse particles respectively, making a total of 1620 particles per particleboard plant.  72  4.2.3 Data analysis The mass of the different particle size classes was measured and their mass percentage computed. Particle mass fraction in each size class was treated separately with a one-way ANOVA and the plants means were compared with Tukey multiple comparison test to identify significant differences between the means. Wide variation in the size and shape of particles, especially for the medium and coarse core particles, was observed. Because all the histograms were right skewed and zero bound, they were fit using 2-parameter lognormal and Weibull distributions. These distributions can take a wide array of shapes and have been used to fit naturally occurring observations and wood fibres (Lu et al. 2007). The lognormal with heavy right tail and Weibull with heavy left tail can be considered complementary and convenient for this work as they can fit the natural variability of the particles with extremely low and high values (Law and Kelton 2000, Meeker and Escobar 1998).  The random variable X has a lognormal distribution, if Y = ln X follows a normal distribution with a standard deviation σ, also called the scale parameter, greater than zero. The lognormal distribution has a probability density function (PDF) as follows:  f(x) =  (  1 xσ  =0  2π for x  )   ln x - µ 2  exp  −  2σ 2    for x  ≥ 0  4.1  ≤ 0  The lognormal cumulative distribution function (CDF) can be expressed in terms of the normal CDF Φ (z), where Z = (X- µ)/ σ. For x ≥ 0, the CDF is  73   ln x − µ  F ( x; µ , σ ) = Φ  for x > 0  σ  = 0 for x ≤ 0  4.2  The CDF of Weibull distribution has the form  F ( x; β , α ) = 1- exp [- (x / α) β ]  for x ≥ 0  4.3  = 0 for x ≤ 0  where α and β are the scale and shape parameters respectively and are positive numbers (Stanford and Vardeman 1994). The PDF can be obtained as a derivative of F(x; β, α) which can be expressed as:  f ( x; β , α ) = βα - β x β -1 exp [- (x / α) β ]  for x ≥ 0  = 0 for x ≤ 0  4.4  The observations were also fit to the gamma distribution, which can accommodate variables that are highly skewed. The form of the PDF, which can be expressed in terms of the gamma function (Γ) parameterized in terms of a shape parameter k and scale parameter λ -1, is as follows:  f ( x; k , λ ) =  λ ( λ x ) k −1 e − λx Γ( k )  for x > 0 and k , λ > 0  The CDF for the incomplete gamma function is expressed as:  74  4.5  x  1 F (k , x) = u k −1e − u du ∫ Γ (k ) 0  4.6  where Γ( x) = ∫ t x−1e −t dt ( Lawless 2003). All of these distributions have two parameters with the location or threshold parameter of the gamma distribution, set to zero. Goodness-of-fit test for significance of the distribution fits was performed using Anderson-Darling (AD) and Kolgomorov-Smirnov (KS) tests. The AD test procedure compares the observed CDF to a theoretical CDF and is dependent on the actual distribution being tested. The KS test is based on the vertical difference between a theoretical and empirical CDF’s (EasyFit Software 2009). The AD test gives more weight to the tails of a distribution compared to the KS test. Since the differences in the observations between plants may be more pronounced at the tails, the interpretation of the results will be based on AD test results.  4.3  Results and discussion  4.3.1 Sieve screen weight data analysis The mean weight percentage of each particle size class for each plant is given in Table 4.2. Overall, 80% of particles screened remained on the 0.5 mm, 1 mm, and 2 mm meshes with most particles, 30 to 35% of the total particle mass, retained on the 1 mm screen.  75  Table 4.2  Particle mean mass on each mesh size Mass percentage of each mesh size by Plant  Mesh size (mm)*  A  B  C  D  E  F  - 0.125  0.45  1.41  2.51  2.10  2.06  2.12  +0.125  1.57  2.65  5.05  4.32  5.04  4.02  +0.250  10.67  11.00  14.93  14.57  15.63  12.54  +0.500  29.06  32.60  30.15  29.12  25.02  21.83  +1.000  37.46  36.44  30.94  32.51  32.10  31.19  +2.000  18.23  14.68  15.05  16.93  19.69  22.81  +4.000  2.64  1.20  1.40  0.45  0.47  5.50  *mesh size opening, + indicates particles retained on the mesh size and – indicates particle passing through the mesh size. Each mean is the result of n = 9 sample bags with each bag containing 200g furnish.  According to discussions with plant personnel, typical industrial surface furnish passing through a 0.75 mm mesh and makes up about 48% of the total furnish mass (Besselt 2005). These addition rates were similar to those used in this study, i.e., particles passing through 1 mm mesh were between 40 and 53% for all plants. The means of particle mass percentages are plotted in Figure 4.1. Although the shapes of the distributions were similar, there were wide differences in the screen size masses between plants for the 0.5, 1, and 2 mm mesh sizes. There were statistically significant differences in mass percentages between plants for all particle sizes. However, the differences between plants were lower for particles passing through the 0.5 mm mesh and those remaining on the 4 mm meshes. The Tukey multiple comparison test at 0.05 αlevel indicates a significantly higher mass of the 1 mm particle size class in panels from plants A and B, while plant F had the highest amount of particle sizes above the 2 mm mesh.  76  Figure 4.1 Distribution of mean particle mass as a percentage of total particle mass for each particle size class; LSD bars are given for comparison of the mean particles size from each plant for that size class.  4.3.2 Geometrical descriptors and their effect on panel strength Means of particle geometry with their calculated aspect and slenderness ratios, and projected surface area are listed on Table 4.3. There were significant differences (p<0.0001) between plants for all geometrical descriptors, with the exception of width for the medium particles. It was also observed that the finer particles had smaller size differences between plants, which was similar to the observation for mass percentages. The mean length and width of the core-fine particles from plant F were higher than those of plant A, consequently, the AR, SR, and surface area followed a similar trend.  77  Table 4.3  Comparison of the means of geometrical descriptors of the core-fine, medium and  coarse core particles Furnish source (plant) Particle Class  A  B  C  D  E  F  Length (mm) Core-fine  3.96 a  1.98 d  3.66 a,b  2.88 c  3.14 c  4.05 a  Medium  9.44 a  5.97 b  5.78 b  7.03 a  7.57 a  8.03 a  Coarse  12.26 a  8.67 c  9.16 b  9.64 b  9.27 b  10.53 a  Width (mm) Core-fine  0.90 b,c  0.97 a,b  1.00 a  0.97 a,b  0.85 c  0.92 a,b,c  Medium  1.84 a  1.88 a  1.85 a  1.91 a  1.86 a  1.91 a  Coarse  3.79 c  4.26 a,b  4.17 b  3.78 c  3.62 d  4.38 a  Thickness (mm) Core-fine  0.28 b  0.27 b  0.36 a  0.28 b  0.23 b  0.25 b  Medium  0.74 a  0.59 b  0.70 a  0.74 a  0.74 a  0.67 a  Coarse  1.12 b  1.11 b  1.22 a  1.41 a  1.29 a  1.15 b  Aspect Ratio (-) Core-fine  4.63 a,b  2.21 c  4.28 a  3.26 b  4.09 a,b  4.84 a  Medium  5.64 a  3.43 e  3.62 d,e  4.00 c,d  4.34 c  4.72 b  Coarse  3.80 a  2.48 e  2.71 d  3.04 b  2.95 b  2.84 c  Slenderness (-) Core-fine  16.45 a,b  10.09 c  12.14 c  13.22 b,c  19.81 a  20.42 a  Medium  15.02 a  12.55 b  14.71 b,c  11.82 c  12.05 b,c  14.23 a  Coarse  12.13 a  8.58 b  8.25 b  7.69 b  8.53 c  10.38 a  Surface area (mm2) Core-fine  3.48 a  1.92 c  3.40 a  2.71 b  2.57 b  3.65 a  Medium  17.42 a  11.16 d  11.59 d  13.30 c  14.07 b,c  14.95 b  Coarse  47.30 a  37.02 b,c  38.62 b  38.10 b,c  35.23 c  47.86 a  Note: Means are not significantly different, if the letter (a,b,c) beside them is the same. n = 120 (core-fine); n = 600 (medium); and n = 900 (coarse) for each mean.  78  However, core-fine particles were less in the core and hence have minimal effect on the mechanical properties compared with the medium and coarse particles. For the medium and coarse particles, Plant A had significantly longer, but relatively narrow particles compared with all other plants, resulting in higher AR, SR, and surface area for the medium and coarse particles. Since particle length determines the number of inter-particle contact points of a particle with other adjacent ones, the longer the particle the better the inherent particle strength contributes to the overall consolidated panel strength (Marra 1954) as demonstrated by the particles from plant A ( see Table 4.4).  The mechanical properties of the panels from which the particles were derived has been reported by Semple et al. (2005), hence only the likely effect of particle geometry on the mechanical properties will be discussed. According to Heebink and Hann (1959) and Post (1961) increasing SR increases flexural properties and these were shown in plant A for MOE and MOR and in plant F for MOE. The larger surface area of the particles from plant A and F were likely to have resulted in more bonds to adjacent particles, thus increasing their ability to transfer stress. Although increasing aspect ratio decreases flexural properties, this was not seen in panels from plant A, because the high aspect ratio was likely offset by the very high slenderness ratio of those particles. The higher IB and edge SWR from plant A may be due to the lower thickness and width of particles producing a better packing efficiency. The high surface area also increases the area available for resin coverage. Coarse and medium particles from plant B, which were relatively short, wide, and thin, had lower AR, SR, and surface area. With the smaller surface area, the particles have less opportunity for bonding with adjacent particles and likely led to the lower mechanical properties of those boards (Semple et al. 2005).  79  Table 4.4  Means of mechanical properties for panels from plants A to F.  property  Plants A  B  C  D  E  F  density (kg/m3)  681.3  706.9  702.0  657.6  646.7  648.3  IB (MPa)  0.713  0.427  0.653  0.588  0.576  0.613  Face SWR-A (kN)  1.098  0.837  1.020  1.035  1.016  0.950  Face SWR-B (kN)  1.166  0.883  1.031  1.076  1.026  0.992  edge SWR (N)  972.9  634.4  776.2  732.5  770.8  790.1  MOR ║ (MPa)  16.03  13.50  14.71  16.61  12.31  13.38  MOR ┴ (MPa)  15.07  12.59  14.17  15.30  12.24  13.03  MOE ║ (GPa)  3.086  2.336  2.599  2.998  2.553  2.811  MOE ┴ (GPa)  2.794  2.153  2.378  2.774  2.404  2.695  Adapted and modified from Semple et al. 2005a  4.3.3 Relationship between properties (SWR and IB) and particle descriptors (AR and SR) of particleboard core Figure 4.2 shows the relationship between the mechanical properties of the core board structure and the AR and SR of particleboard furnish. There is a general positive trend between the properties and the descriptors. Figure 4.2a shows a relatively high correlation between SWR and AR of the medium and coarse particles with R2 = 0.82 and 0.84 respectively, confirming the results of Lehmann (1974) and Lin et al. (2002). Particles with higher aspect ratio tend to be chunky having more wood substance and form particle stacks from overlaps of adjacent particles and present a better grip for screws and higher resistance during withdrawal. Higher AR also offers a higher surface area for resin coverage which assists in bonding.  80  1000  1000  950  950 R² = 0.8382  Edge SWR(N)  900  900  850 R² = 0.5838  800 750  700  Core-fine Medium Coarse  650 2.5  3  3.5  (a)  4  4.5  5  5.5  6  600  Medium Coarse  2  7  12  17  22  27  Slenderness ratio  (b)  0.75  0.75  0.7  0.7  R² = 0.393  Internal bond strength (MPa)  Core-fine  650  Aspect ratio  0.65  R² = 0.6671  0.65  R² = 0.2925  0.6  0.6  0.55  0.55  0.5  0.5  0.45  0.45  0.4  Core-fine  0.35 0.3  R² = 0.1981  750  R² = 0.8244  2  R² = 0.6598  800  700  600  R² = 0.4533  850  Medium Coarse  2  (c)  Figure 4.2  2.5  3  3.5  4  Aspect ratio  4.5  5  5.5  6  R² = 0.1237 R² = 0.0914  R² = 0.3548  0.4  Core-fine  0.35  Medium Coarse  0.3  2  (d)  7  12  17  22  27  Slenderness ratio  Relationship between mechanical properties (edge SWR and IB strength) and  geometrical descriptors (AR and SR) for core-fine, medium, and coarse particles of particleboard core. The upper row, (a) and (b), is edge SWR and the lower row, (b) and (c), is IB strength.  Although higher SR leads to more particle overlaps and stacking, the particles are too thin with less material available for the screw to grip, consequently leading to lower R2 of 0.45 and 0.66 with SWR for medium and coarse particles. Relative to SWR, the correlation between IB and AR was very low with the exception of AR of core-fine particles, which had an R2 of 0.67. The core-fines fill the voids within and between the coarse and medium particles and increase the bonds formed, i.e., the higher the fines in the core the higher the IB strength (Nemli 2003, 81  Kakaras and Papadopoulos 2004, Sackey et al. 2008). As in earlier studies, IB strength did not show a good correlation with slenderness ratio of any of the particle sizes. (Note that the flexural properties of the panel were not considered because they are influenced mainly by the face structure of the panel).  4.3.4 Particle geometrical descriptors and their distribution The histograms for particle length, aspect and slenderness ratios are shown in Figure 4.3 and are overlaid with the best fitting distribution models from the lognormal, gamma and Weibull family of distributions. Rest of the histograms for all plants can be found in Appendix F. Maximum likelihood method of the Reliability procedure in SAS®9.1 (SAS Institute Inc. 2002) and the Akaike’s Information Criterion (AIC) were used to determine the best fitting distribution for each plant’s particle data set. The AIC, which is an extension of the maximum likelihood principle, provides a quantitative measure for selecting the best distribution model for the data set and can be expressed as:  ∧  AIC = - 2log (L ( θ │data)) +2k  4.7  ∧  ∧  where L( θ ) is the likelihood function evaluated at the maximum likelihood estimator θ and k is the number of parameters (Akaike 1973, Bozdogan 2000, Burnham and Anderson 2004). The shape and scale parameter estimates and their maximum log (likelihood) or ML values for the lognormal and Weibull models for aspect and slenderness ratios of coarse particles are listed in Table 4.5, while that of fine and medium particles can be found in Appendix G. The distribution with the highest ML value is the best fit. Comparing the ML values from the lognormal and 82  Weibull distributions, the lognormal was consistently a better fit for the three descriptors of particles from most plants than the Weibull distribution. The Weibull distribution was a better model only for the length and aspect ratio of the medium particles from plant D as shown in Table 4.7.  140  Plant A size : coarse mean:12.26 mm n : 900  200 160  Plant A size : coarse mean: 3.50 mm n : 900  120  160  100  Lognormal  Lognormal  Lognormal  Gamma  Count  Plant A size : coarse mean: 12.13 mm n : 900  200  Gamma  80  120  120  60  80  80 40  40 0  40  20 0  0  5  10 15 20 25 Particle length, mm  30  35  0 1  2  3  4  5 6 7 Aspect ratio  250  20  Plant E size : fine mean: 3.15mm n : 40  16  9  10 11  Plant E size : medium mean: 7.57 mm n : 600  200  0  250  200  Lognormal Gamma  Lognormal Weibull Gamma  12  8  8  100  100  4  50  50  Count  150  0  1  2  3 4 5 6 Particle length, mm  7  8  0  2  4  6  8 10 12 14 16 18 20 Particle length, mm  5  10  15 20 25 30 Slenderness ratio  35  40  Plant E size : coarse mean: 9.27 mm n : 900 Lognormal  150  0  0  0  Gamma  0 2 4 6 8 10 12 14 16 18 20 22 24 26 Particle length, mm  Figure 4.3 Typical histograms of the observations overlaid with best fit distribution. Upper row shows distribution of particle length and aspect and slenderness ratios of the coarse particles from Plant A and the lower row shows distribution of particle length of fine, medium, and coarse particles from plant E.  83  Table 4.5  Lognormal  Aspect ratio  Plant  Slenderness ratio  Model parameters of aspect and slenderness ratios for coarse particles Weibull  ML  Scale (σ)  Shape (µ)  ML  Scale (α)  Shape (β)  A  -598.440  1.1468  0.4707  -629.397  3.9647  2.2910  B  -519.653  0.6882  0.4313  -662.568  2.4956  2.0887  C  -547.326  0.7690  0.4448  -634.361  2.7063  2.2234  D  -499.077  0.9868  0.4215  -537.753  3.3027  2.5320  E  -523.653  0.9711  0.4332  -569.468  3.2714  2.4289  F  -564.131  0.8503  0.4531  -644.940  2.9443  2.1999  A  -596.391  2.3849  0.4697  -674.213  13.7351  2.1103  B  -466.085  2.0627  0.4064  -592.133  9.6986  2.2584  C  -492.370  2.0197  0.4184  -612.834  9.3291  2.2011  D  -400.301  1.9987  0.3777  -591.334  8.9863  2.1543  E  -391.377  2.0258  0.4229  -532.979  9.4672  1.9702  F  -621.188  2.2166  0.4828  -746.684  11.7654  1.9025  ML = maximum log (likelihood)  Table 4.6  AICc values of aspect and slenderness ratios for coarse particles Aspect ratio(AR)  Plant  Lognormal  Gamma  Weibull  Slenderness ratio (SR) Lognormal  Gamma  Weibull  A  3265.43  3263.27 3327.26  5489.73  5527.02 5645.34  B  2282.16  2385.63 2568.03  4649.07  4719.43 4901.20  C  2482.80  2536.36 2656.86  4624.03  4687.04 4865.06  D  2716.01  2723.55 2798.75  4535.70  4645.24 4708.72  E  2798.63  2819.90 2916.56  4761.80  4910.30 5149.88  F  2662.93  2824.62 2709.93  5236.39  5324.98 5487.34  Note: Bold italics denote best fit distributions, italics are next best and the least fit are the distributions in regular font.  Log-likelihood values for lognormal, gamma, and 2-Weibull distributions and their parameter estimates were performed using JMP 8 (2008) software. However, to allow for comparison with 84  more than 2-parameter distributions in the future and other studies, AICc (corrected AIC) listed in Table 4.6 was computed and used in ranking the distributions; the lower an AIC value the better the distribution fit. Table of AIC and loglikelihood values for testing lognormal, gamma, and Weibull distributions fit to coarse, medium, and fine particle sizes can be found in Appendix H. How well a distribution fits a descriptor of a particle class from a plant was ranked 1, 2, or 3 for best, better, or good fit respectively. For example, for the AR of plant A, the gamma distribution was assigned a value of 1, the lognormal a value of 2, and the Weibull a value of 3. This was repeated for all plants and the total score was computed for each distribution; the scores range from 6 to 18 for each descriptor. The distribution with the lowest ranking score for all plants indicates the best model for a particular descriptor. Generally, 2-parameter lognormal distribution provided the best fit model for SR in all particle size classes. The AIC scores also showed that the 2-parameter Weibull was the least fit for all three descriptors. This is contrary to the results from Lu et al. (2007) who found the Weibull distribution to be the best fit for the core particles of particleboard. This discrepancy may due to lower variability in their samples, which were drawn from only one source. Care must be taken in comparing the two studies, however, since there was a further partitioning of the core particles in this study, which was not the case with the study of Lu et al. (2007). It must be noted that the whole furnish characterization followed similar distribution patterns as the partitioned furnish. Length and AR of the core-fine particles are best fit by a lognormal distribution, while the same descriptors of the medium particles are best fit with the gamma distribution model.  85  4.3.5 Goodness of fit tests In the goodness of fit test, the null hypothesis Ho is that the data follows the lognormal (or Weibull or gamma) distribution and the alternate hypothesis HA is that the data does not follow the specified distribution at α= 0.05. In Table 4.7, L, G, and W represent cases where the distribution of lognormal, gamma, or Weibull, respectively, fits length, AR, and SR from a particular plant. Although, both AD (Anderson-Darling) and KS (Kolmogorov-Smirnov) tests were used in all instances, there were few cases where only one of these tests was significant and it is indicated by an AD or KS superscript.  Table 4.7  Goodness of fit test for the distribution models for length, aspect and slenderness  ratios of core-fine, medium and coarse industrial particles. Length Plant Corefine  Aspect ratio  Medium  Coarse  Corefine  Slenderness ratio  Medium  Coarse  Corefine  Medium. Coarse  A  L  L,G  L,G  L,G,W  L  L,G,  L,G  L  L  B  nf  L,G  nf  L  L  nf  L  L  L  C  nf  L,G,WKS  L  nf  L,G,WKS  LKS  nf  nf  L  D  L,G  G,W  L,G  L  G,W  L,G  LKS  nf  L  AD  E  L,G  L,G  L,G  F  L,G  G,W  L  L  L,G  L,G  L,G  nf  L  L  L,G,WKS  AD  L  L  LKS L  L, G, W = fail to reject HO, data from a lognormal, gamma, or Weibull distribution respectively, while nf indicates no fit. KS = Kolmogorov Smirnov test and AD = Anderson-Darling.  Generally the lognormal distribution was the best fit for most of the descriptors, particularly for the medium particles, because most of the particle distributions have a heavier right tail. Particle length and AR of the medium particles from Plant D was best fit with the gamma and Weibull models, whereas the lognormal model was not a good fit. These particles were relatively short, 86  see Table 4.3, and hence weighted towards zero (Meeker and Escobar 1998, Lu et al. 2007). Unlike Lu et al. (2007), all three 2-parameter distribution models were rejected for some particles from plants B, C, D, and E, with the greater percentage from plants B and C, which are inadvertently sister plants in two different locations. It must be noted that the gamma family of distributions, which has been the least used distribution for modelling wood particles and fibres, was a better fit for most particle size classes than the 2-parameter Weibull distribution. The lack of fit for most of the distributions for SR for medium and coarse particles may be due to the lower variability found in particle thicknesses, making the distribution pattern of SR to some extent similar to that of the length distribution.  The results of this investigation suggest that properties of conventional particleboard panels on the market can be correlated with the length, AR, SR of particles from hydrolyzed panels. Knowledge of the continuous particle distribution models obtained opens avenues for further work in obtaining optimal particle packing efficiencies in particulate wood composites. The distribution models should make possible the manipulation of particle geometries to improve strength properties. Distribution models will also assist in formulating particle distributions that will increase or decrease porosity of a panel mat, which affects permeability during hot pressing and may lead to particle mixtures that reduce degassing times. Mat compression behaviour can also be influenced with the knowledge of particle size distribution and their variability. For instance from a lognormal distribution model, the thickest particles in the upper 5% in the core of the particle mat, which leads to higher mat compression and consequent compressive failure, can be removed by lowering the heavy right tail of the distribution.  87  4.4  Conclusions  Within the confines of this study the following conclusions can be drawn: 1. Overall, about 80% of the total mass of particles was retained on the 0.5 mm, 1 mm and the 2 mm meshes, with about 35% of the total mass of particles retained on the 1 mm mesh alone.  2. There was a significant difference in furnish mass percentage between all six plants for all particle sizes.  3. Less variation was found between percentage mass of fine particles (< 0.5 mm mesh) and between percentage mass of the largest (> 4 mm) particles than the midrange sizes.  4. Particle aspect ratio is a better material predictor for screw withdrawal resistance of particulate wood composites than absolute geometrical dimensions and increase in corefine particles increase internal bond strength.  5. The lognormal distribution provided the best fit for length, slenderness ratio, and aspect ratio for all particle types (core-fine, medium, and coarse) for the core furnish.  6. Gamma distribution also provide a good model fit for length, aspect ratio, and slenderness ratio for the core-fine, medium, and coarse particles compared with the Weibull.  88  7. From this study, it can be concluded that lognormal and gamma distributions are good distribution models, for characterizing the slenderness ratio and aspect ratio of irregular particulate wood material.  4.5  Acknowledgement  The authors wish to gratefully acknowledge the funding support of Natural Resources Canada and extend our thanks to Dr. Kate Semple for her invaluable proofreading and input.  89  4.6  Literature cited  Akaike, H. 1973. Information theory and an extension of the maximum likelihood principle. In B.N. Petrov and F. Csaki (eds.), Second International Symposium on Information Theory, Acaderniai Kiado, Budapest, 267-281. Allen, T. 1981. Particle size measurement 3rd ed. Chapman and Hall Ltd. NY 10017 USA. Besselt, N. 2005. Technical discussion at NewPro particleboard plant in Wahnam AB with the Technical Director on May 9th 2005. Bozdogan, H. 2000 Akaike's information criterion and recent developments in information complexity. Journal of Mathematical Psychology, 44(1), 62-91. Burnham, K.P. and D. R. Anderson. 2004. Multimodel Inference: Understanding AIC and BIC in Model Selection. Colorado Cooperative Fish and Wildlife Research Unit (USGS-BRD). Sociological Methods and Research 33(2): 261-304. Cao, Q. V. and Q. Wu. 2007. Characterizing wood fiber and particle length with a mixture distribution and segmented distribution. Holzforshung 61:124-130. EasyFit Software. 2009. EasyFit Help, Goodness of fit. Published by MathWave Technologies. Eusebio, G. A. and N. C. Generalla. 1983. Effect of particle resin adhesive distribution in particleboard manufacture of Kaatoan Bangkal [Anthocephalus chinensis (Lam.) Rich. Ex Walp.]. FPRDI Journal Vol. 12(3-4):12-19. Geimer, R. L. and C. L. Link. 1988. Flake classification by image analysis Res. Pap. FPL-RP486. Madison, WI: U.S. Department of Agriculture, Forest Service, Forest Prod. Lab. 25p.  90  Geimer, R. L., J. W. Evans, and D. Setiabudi. 1999. Flake furnish characterization - Modeling board properties with geometric descriptors. Res. Pap. FPL-RP-577. Madison, WI: U.S. Department of Agriculture, Forest Service, Forest Prod. Lab. 36 p. Heebink, B. G. and R. A. Hann. 1959. How wax and particle shape affect stability and strength of oak particleboards. Forest Prod. J. 9(7):197-203. JMP 8 Statistical Discovery Software. 2008. SAS Institute Inc. SAS Campus Drive, Building S, Cary, NC, 27513, Cary, NC, USA. Kakaras, I.A. and A. N. Papadopoulos. 2004. The effects of drying temperature of wood chips upon the internal bond strength of particleboard. Journal of the Institute of Wood science. 16 (5):277-279. Kelly, R. N., J. K. DiSante, E. Stranzl, J. A. Kazanjian, P. Bowen, T. Matsuyama, and N. Gabas. 2006. Graphical Comparison of Image Analysis and Laser Diffraction Particle Size Analysis Data Obtained From the Measurements of Nonspherical Particle Systems. AAPS PharmSciTech.7 (3): E1-E14. Khalili, M.A, W. L. Roricht, and L. S. Y. Luke. 2002. An investigation to determine the precision for measuring particle size distribution by laser diffraction. World Congress on Particle Technology 4, July 21-25, 2002, Sydney, Australia. Paper no. 111. Kropholler, H. W. and W. W. Sampson. 2001. The effect of fiber length distribution on suspension crowding. J. Pulp Paper Sci. 27(9):301-305. Kusian, R. 1968. Model investigations about the influence of particle size on structural and strength properties of particle materials. II. Experimental investigations. Holztechnologie 9(4):241-248.  91  Law, A. M. and W. D. Kelton. 2000. Simulation modeling and analysis. 3rd ed. McGraw-Hill Publishers. 760 p. Lawless, J. F. 2003. Statistical models and methods for lifetime data. John Wiley and Sons, Inc. Hoboken, New Jersey, USA. Lehmann, W.F. 1974. Properties of Structural Particle boards. For. Prod. J. 24(1): 19-26. Li, M., D. Wilkinson. and K. Patchigolla. 2005. Comparison of particle size distributions measured using different techniques. Particulate Science and Technology. 23:265-284. Lin, H. C., Y. Fujimoto, Y. Murase, and Y. Mataki. 2002. Behaviour of acoustic emission generation during tensile tests perpendicular to the plane of particleboard II: effects of particle sizes and moisture content of boards. Journal of Wood Science 48(5): 374-379. Lu, J. Z., C. J. Monlezun, Q. Wu, and Q. V. Cao. 2007. Fitting weibull and lognormal distributions to medium-density fiberboard fiber and wood particle length. Wood and Fiber Science, 39(1), 82-94. Mahoney, R. J. 1980. Physical changes in wood particles induced by the particleboard hot pressing operation. In. Proceedings 14th International Symposium on Particleboard, April 1980, ed T.M. Maloney, Pullman Washington. pp. 213-223. Maloney, T. M. 1970. Resin distribution in layered particleboard. Forest Prod. J. 20(1): 43-52. Marra, G. G. 1954. Discussion following article by Turner HD – Effect of particle size and shape on strength and dimensional stability of resin-bonded wood-particle panels. Forest Prod. J. 4(5):210-223. Meeker, W.Q. and L. A. Escobar. 1998. Statistical method for reliability data. John Wiley & Sons, Inc., New York, NY. 92  Nemli, G. 2003. Effects of some manufacturing factors on the properties of particleboard manufactured from Alder (Alnus glutinosa subs. Barbata). Turkish Journal of. Agriculture and Forestry. 27: 99-104. Post, P.W. 1961. Relationship of flake size and resin content to mechanical and dimensional properties of flake board. Forest Prod. J. 11(9):34-37. Sackey, E., K. Semple., S-W. Oh, and G. D. Smith. 2008. Improving core bond strength of particleboard through particle size redistribution. Wood and Fiber Sci., 40(2), 214-224. Semple, K., E. Sackey, H. J. Park, and G. D. Smith. 2005. Properties variation study of furniture grade M2 particleboard manufactured in Canada. Forest Prod. J. 55(12): 117-124. Shuler, C.E. and R. A. Kelly. 1976. Effect of flake geometry on mechanical properties of eastern spruce flake-type particleboard. For Prod J 26(6):24–31. Stanford, J.L. and S. B. Vardeman. 1994. Statistical methods for physical science. Vol. 28. Method of experimental physics. Academic Press, San Diego, CA 92101. SAS®9.1. 2002. The Reliability Procedure. SAS Institute Inc., SAS/QC® User’s Guide, SAS Institute, Inc. Cary, NC. Wolcott, M. P., F. A. Kamke, and D. A. Dillard. 1994. Fundamentals aspects of wood deformation pertaining to manufacture of wood-based composites. Wood and Fiber Sci. 26:496-511. Yan, J. F. 1975. A method for the interpretation of fiber length classification data. Tappi 58(8): 191-192.  93  Chapter 5 Improving core bond strength of particleboard through particle size redistribution4  5.1  Introduction  The size and shape of particles influence the mechanical properties, appearance, and machinability of particleboard (PB). Since its inception the strength-to-weight ratio of particleboard has been greatly improved through the adoption of a three layer structure (smooth high density surface and lower density core containing coarse particles) and advances in resin and press technology, however, low edge screw withdrawal resistance (SWR) is still an issue for producers and users of particleboard today. A recent investigation (Semple et al. 2005) of the mechanical properties of Canadian-made particleboards showed that of the 6 plants surveyed, only boards from one plant met the minimum ANSI A208.1 (1999) standard for edge SWR. This was attributed to the possible variability in press schedules, resin formulation and content between plants, and in part to the varied and complex particle shapes and sizes in the core that created numerous large voids within the core. These inter-particle voids are a source of discontinuities in structure and of bondlines in the board core that act as flaws and facilitate delamination of inter-particle bonds (Conrad et al 2004, Lei and Wilson 1980, River 1994).  4  A version of this chapter has been published. Sackey, E., K. Semple, S-W. Oh, and G.D. Smith. 2008. Improving core bond strength of particleboard through particle size redistribution. Wood and Fiber Science, 40(2), 214-224.  94  Particle size and shape strongly influences the availability of particle surface for gluing and the number and area of inter-particle contact points between adjacent particles, which then determines the degree to which the inherent particle strength can contribute to the overall final panel strength (Kollmann et al. 1975, Lynam 1959, Marra 1954). The relationships between particle length, width, and thickness are known to influence different panel properties. For instance, flexural properties generally increase with increasing slenderness ratio (ratio of length to thickness) (Brumbaugh 1960, Heebink and Hann 1959, Post 1958, 1961), but decrease with increasing aspect ratio (ratio of length to width). In contrast, internal bond (IB) strength increases for thicker and shorter flakes (i.e. lower slenderness ratio and higher aspect ratio). Post (1961) suggested that SWR could be improved as slenderness ratio approaches unity, while Kimoto et al. (1964 – as cited in Moslemi 1974), recorded no increase in SWR for slenderness ratios above 50.  In contemporary three layer particleboard, the surface consists of finer and more uniform particles, whereas the core is composed of larger and coarser particles of variable sizes and shapes. Efficient combination of particles of various sizes in multilayer and three layer boards with the aim of efficient resin usage and mechanical property improvement have been pursued. Comparing the IB strength of two particleboard formations, Maloney (1970) constructed a 3layer board with about 50% coarse core content comprising of particles of varied sizes and a multilayer board with a graduated core, whose center consisted of 20% of particles > 2 mm Tyler mesh size. He found the IB-strength of the 3-layer boards to be greater. The inclusion of fines and wood dust in the core of boards has been shown to increase IB strength (Kakaras and Papadopoulos 2004, Talbott and Maloney 1957, Nemli 2003), but reduce static bending and  95  modulus of elasticity (Mottet 1967, Nemli 2003). The increase in IB strength has been attributed to the filling of void spaces between the larger particles with fines to produce a higher degree of inter-particle contact (Nemli 2003).  Although the effect of core fines content on flexural properties and IB strength has been the focus of several investigations, the effect on edge SWR and the threshold at which addition of fines would be unfavorable to IB strength has not been studied. The hypothesis of this study is that the particle size mixture in most Canadian-made industrial particleboard is sub-optimal for IB strength and edge SWR and that a better core mixture can be produced by transferring some fines to the core. This is based on the idea that increasing the fines content in the core of particleboard will fill the void spaces, leading to increased bonding and edge SWR. Since the key properties for non-structural particleboard are SWR and IB strength, this study aims to find a core particle size mixture which will increase these without increasing board density and resin requirement or adversely affecting the flexural properties. Specific objectives of the study are:  1.  Determine the effect on edge SWR and IB strength by varying the ratios of coarse and fine particles in uniform vertical density profile (UD) particleboard.  2.  Determine the effect on vertical density profile (VDP), edge SWR, IB and flexural properties (modulus of rupture -MOR and modulus of elasticity-MOE) of different particle size mixes in the core of 3 layer particleboards.  3.  Compare the properties of boards manufactured with customized particle size with those containing unscreened industrial furnish.  96  mixes  5.2  Materials and methods  5.2.1 Furnish preparation Dried face and core industrial PB furnish were collected from a particleboard mill in Alberta, Canada and screened with a mechanical shaker table using 3 different screens with square openings of 2.0, 1.0, and 0.5 mm (Tyler mesh sizes: 9, 16, and 32, respectively). Based on telephone conversations with personnel from different particleboard plants (Semple et al. 2005), these screen sizes are representative of those used in particleboard plants across Canada. Screening the furnish through this set of screens resulted in four particle size classes, with the smallest particles (P1) passing through the 32 mesh screen. The next smallest particles (P2) passed through the 16 mesh screen, but not the 32 mesh; the next size (P3) passed through the 9 mesh screen, but not the 16 mesh screen; and the largest particles (P4) were those retained on the 9 mesh screen.  5.2.2 Compilation of customized particle size mixes Four artificial particle size mixtures (M1 to M4) were compiled from the screened size fractions, with the proportions of the particle size fractions in each mix shown in Table 5.1 The contents of control mix (CTL) were determined by screening the industrial core furnish as received, and found to contain equal proportions of P3 and P4 particle sizes with comparatively little dust. The custom mixtures were therefore formulated around the CTL, ranging from 100% coarse (M1) to 100% fine (M4). Two intermediate size mixtures were formulated with the objective to fill void spaces between coarsest particles first with the P2 particles and then smaller voids with the dust (P1).  97  Table 5.1  Particle size classes and amounts in furnish mixtures  Particle size  Classification  (based on Tyler mesh )  Mesh  Percentage of particle size in mixture  opening  (%)  size (mm)  M1  M2  M3  CTL  M4  P4 retained on 9 mesh  coarse  2.0  100  60  40  42  0  P3 retained on 16 mesh  medium  1.0  0  20  20  42  0  P2 retained on 32 mesh  fines  0.5  0  20  20  11  0  P1 retained in pan  fines  <0.5  0  0  20  5  100  Total :  100  100  100  100  100  Note: CTL = control  5.2.3 Blending and mat formation The mill that supplied the furnish also supplied surface and core urea formaldehyde (UF) resin for blending of boards. In order to determine the volumetric resin distribution in the different blended batches, hydrated CuSO4 solution was added to the industrial resin following a method used by Feng and Anderson (2004), which reduced the solids content from 65 to 60%. Doped resin was applied to the furnish in a Drais particleboard batch blender equipped with an air atomizing nozzle. In the blender the resin is sprayed on the particles as they are stirred by rotating paddles within the blender cavity. Face and core furnish were blended separately and in each batch sufficient furnish was blended to make 3 boards. All furnish was conditioned to about 6% moisture content before blending.  From discussions with particleboard producers, the total resin content of commercial particleboard varies between 9 and 13% of the oven dry furnish mass.  Because resin  consumption increases with increased specific surface area of furnish, i.e. the ratio of mass to total surface area (Maloney 1970, Moslemi 1974), it is expected that the properties of a board  98  made with furnish containing a large amount of fines will be lower for the same resin content compared with a board made from larger furnish particles. Since the objective is to increase bond strength with reduced resin requirement, resin content at the lower end of the industrial range, i.e., 9% was used. The UD boards containing only core furnish mixtures were blended with core UF-resin at a resin content of 6% of oven-dry weight of furnish. For the 3 layer boards, the face layers were blended with surface UF-resin at 9% (oven-dry weight) resin content, while the core layers were blended with 6% resin. All furnish mats were hand-formed.  5.2.4 Manufacture of uniform density boards Particle mixtures were first tested in uniform density profile (UD) boards and then in the standard 3-layer board configuration found in industrial particleboard. A set of 20 boards (4 replicates per mixture) were made from the 4 custom furnish mixtures M1, M2, M3, and M4, plus the control as shown in Table 5.1. Boards were pressed to thickness of 11 mm and target density of 530 kg/m3. This density was based on the average core density of commercial boards in our previous investigation of Canadian-made particleboards (Semple et al. 2005). In order to obtain a flat, uniform VDP board, a pressing procedure described by Wong et al. (1999) was used. The mats were first cold pressed in an electrically heated Pathex press to target thickness, the temperature was then increased from room temperature to the target temperature of 160oC and maintained for 3 min, at which point the press was opened. The outer edges (50 mm wide) of boards were trimmed leaving a 250 by 250 mm square from which 3 edge SWR specimens and 4 IB strength specimens were cut. Restricted board thickness required the use of 25 mm No. 6 wood screws 1.6 mm per thread to test edge SWR. Sample dimensions and physical and mechanical properties testing procedures were done in accordance with ASTM D 1037 (2000).  99  All specimens were conditioned for a minimum of 2 wk at 20oC and 65% RH before testing. Mechanical property tests were conducted on a Sintech 30D universal test machine.  5.2.5 Manufacture of three layer boards Larger 3-layer boards were made after analyzing the results of the UD boards. The analysis showed that the properties of boards containing 100% dust P1 particles were substantially lower than that of the others. As a result the experimental design was modified to exclude the M4 mixture. The 3-layer boards were made with the core comprising 54% of the total furnish mass and each face comprising 23% of the total furnish mass. This ratio was chosen based on work by Maloney (1970) and is similar to that of commercial particleboard. The top and bottom surface layers furnish of all boards was as received from the particleboard company. With the exception of the core of the CTL boards, board cores contained customized furnish mix (M1, M2, or M3) of screened particle sizes.  Blended furnish for the 3-layer boards was distributed evenly by hand into a 710 by 710 mm forming box. Unlike the UD boards, the 3-layer boards were hot pressed without cold prepressing using an electrically heated Pathex press to a target thickness of 16 mm at a maximum platen temperature of 160oC. Due to the higher moisture content of the mats caused by the excess water used to dissolve the hydrated CuSO4, the press cycle was lengthened to a total of 13.75 min, instead of the usual 3 to 7 min to permit the additional water vapor to escape. The press cycle consisted of 15s closing time, 590s cooking time, and 220s degassing time to avoid delamination. Each board was then cooled at room temperature, trimmed to 660 by 660 mm, and cut as shown in Figure 5.1. The VDP of the IB samples were measured prior to gluing-up the IB  100  samples. The VDP was measured with an X-ray density profilometer (Quintex Measurement Systems; Model QDP-01X). All testing procedures were done in accordance with ASTM D 1037 (2000).  MOR/MOE m, d, r, s SWR m,d,r,s  IB m,d,r,s  IB m,d,r,s  IB m,d,r,s  IB m,d,r,s MOR/MOE m, d, r, s  MOR/MOE m, d, r, s  IB m,d,r,s  IB m,d,r,s  SWR m,d,r,s  IB m,d,r,s  IB m,d,r,s  SWR m,d,r,s  IB m,d,r,s  SWR m,d,r,s  MOR/MOE m, d, r, s IB m,d,r,s  Figure 5.1  IB m,d,r,s IB m,d,r,s  IB m,d,r,s  SWR m,d,r,s  SWR m,d,r,s  IB m,d,r,s  IB m,d,r,s SWR m,d,r,s  SWR m,d,r,s  IB m,d,r,s  Cutting pattern of the 3-layer board (660 mm by 660 mm) for specimen sampling  (m – furnish mixture; d – density; r – replicate; s – specimen).  5.2.6 Design of experiments A factorial experimental design was used to determine the effect of different mixtures of particle size fractions on PB mechanical properties. The construction of UD board was necessary to test the effect of the mixtures on bond strength in core furnish without any possible confounding 101  effects of highly compressed, densified surface layers. The UD boards study was designed and analyzed as a single factor experiment with 5 furnish types (M1 to M4, plus CTL furnish), and four board replicates. The response variables were edge SWR (three specimens per board) and IB strength (four specimens per board). For the 3-layer boards, a 2-factor experimental design and analysis was used with four furnish types (M1, M2, M3, and CTL) for the core and two densities, 650 and 700 kg/m3, giving a total of eight combinations. M1, M2, M3, and CTL furnish were of the same composition as used in the UD boards. Three replicates of each mixdensity combination were made for a total of 24 boards. The following properties were measured: edge SWR, IB strength, MOR and MOE as shown in Table 5.2. The factors, response variables, and specimen numbers are listed in Table 5.2. A 2-factor ANOVA was used to assess the effect of and interaction between particle size mix and board density. The significance level for results in both experiments was p ≤ 0.05.  Table 5.2  Treatment structure of factors and response variables of 3-layer boards.  Factor  Levels  Response  Number of  Units  samples/board Particle size  M1  edge SWR  8  N  mixture  M2  IB strength  16  MPa  MOR/MOE  4  MPa/GPa  core density  8  kg/m3  M3 CTL Board density  650 700  Replicate  3  Note: CTL = control  102  5.3  Results and discussion  5.3.1 Uniform density (UD) particleboards 5.3.1.1 Mean density and vertical density profile The cold pressing technique adopted from Wong et al. (1999) resulted in the desired uniform VDPs in the UD boards, as shown in Figure 5.2. Although all VDPs were roughly flat, their mean densities were considerably higher than the target density of 530 kg/m3, which could have been caused by over compression of mats during hot pressing in pressure mode.  700  Density (kg/m3 )  600  500  400  M1  M2  M4  CTL  M3  300 top  bottom  200 0  2  4 6 8 Panel thickness (mm)  10  12  Figure 5.2 Vertical density profile of the UD boards at a target density of 530 kg/m3.  Mean board density was not significantly different between boards, however, M1 and M2 boards with 80% or more medium to coarse particles had lower mean densities compared with M3 and 103  M4 boards having 60% or fewer coarse to medium particles. Boards made from control furnish (CTL) having a relatively high proportion of medium to coarse particles (84%) also had higher mean density. The relative proportion of particle sizes in the mat appeared to affect board density. In the M1 and M2 mixes, higher compaction ratio experienced by the thick particles likely influenced board density, whereas particle packing efficiency may also have influenced board density in the mixes containing more fine particles. With thicker particles more and larger voids are present after reaching maximum compaction compared with the more homogenous packing achieved when finer particles are present. 5.3.1.2 IB strength and edge SWR of UD particleboard Table 5.3 shows the average values for IB strength, edge SWR, and mean density of the UD boards. Mean IB strength of boards made from the M1 and M2 mixtures were significantly higher than the control at the same resin content and higher than the minimum requirements of 0.45 MPa in the ANSI A208.1 (1999) standard. The IB strength from M2 boards was 34% higher than the control, while M1 and M3 boards were 22 and 17% higher than boards made with normal industrial core furnish respectively. The all dust, M4, boards were significantly lower in IB strength than the rest. Failure mode in the M4 IB samples was similar to MDF, occurring near the surface, leaving a thin, 1 to 2 mm surface layer detached from the bulk of the sample.  Table 5.3  Mean values for the properties of the UD single layer boards. M1  M2  M3  M4  CTL  MD (kg/ m3)  570 (3.0)  568 (2.2)  580 (3.1)  588 (3.8)  580 (1.8)  IB (MPa)  0.50 (11.3)  0.55 (8.8)  0.48 (12.5)  0.18 (23.8)  0.41 (8.3)  SWR (N)  579 (11.3)  582 (10.8)  523 (19.9)  235 (28.9)  492 (31.6)  MD = mean density; coefficient of variation (COV) is given in parenthesis.  104  The pattern of mean edge SWR of the UD laboratory boards made with different particle size mixtures is very similar to that of mean IB strength, which follows the close correlation between edge SWR and IB strength found in our previous studies on industrial particleboard (Semple et al. 2005). The all dust M4 boards were significantly lower than the rest of the UD board types for edge SWR. Unlike the IB results, no statistically significant differences were found between M1, M2, M3, and CTL boards. However, mean values of edge SWR of M1 and M2 boards were 18% higher than the control, whereas M3 (containing more fines) was only 6% higher. The edge SWR values were not directly comparable with the ANSI A208.1 (1999) standard minimum values, because smaller screw (No. 6) was used for the test.  5.3.1.3 Effects of mass fraction of fine particles on IB strength and edge SWR of UD boards Figure 5.3 shows the relationship between the mean values of board density, IB strength, and edge SWR and mass fraction of fine particles in the board. Interestingly, the higher densities of the control and M4 boards did not translate into higher IB strength or edge SWR, which indicates that changes in these properties were not due to density but were strongly influenced by particle size mix. Previous research has found that adding fines to core PB increases IB strength (Kakaras and Papadopoulos 2004, Nemli 2003, Talbott and Maloney 1957), but our findings indicate that the increase is limited. It can be seen from Figure 5.3 that increasing fines from 0 to 20% in the customized mixtures lead to an initial increase in IB strength which then decreases as fines content increases above 20%. However, boards from the control mixture with a lower proportion of fines than the M2 boards did not follow this trend. This indicates that changes in the property 105  values are not caused purely by changes in fines content, but also by the proportions of the various particle sizes in the furnish.  700  M2 M1 M3  (■) IB strength (MPa)  0.5  M4  density  600  0.4 500 IB  0.3  400  CTL  0.2  SWR SWR:  LSD = 104.9  300  0.1  (●) Density (kg/m3 ) and (▲) edge SWR (N)  0.6  IB: LSD = 0.075  0.0  200 0.0  Figure 5.3  0.2  0.4 0.6 0.8 Mass fraction of fines  1.0  1.2  Mean values from UD boards for IB strength (■), board density (●) and edge  SWR (▲) loads for all particle size mixes. Note that the fines and dust content of boards, increases from M1 to M4 and the open symbols indicate mean values of CTL boards.  For the laboratory made uniform density profile boards, the findings partially refute our hypothesis that increasing fines will increase IB strength and edge SWR. It is rather the combination of compaction ratio and packing efficiency that cause an increase in the properties. Having the right proportions of different particles might have contributed to higher packing efficiency, whereby the large voids created in the packed P4 (coarse) particles were filled by medium P3 particles and the resulting smaller voids then filled by finer particles. This could 106  partially help explain the lower values obtained from boards made from the control furnish, where the proportion of coarse to medium to fine was 42/42/16, while M2 mix had a proportion of 60/20/20.  For edge SWR of the uniform density profile boards, our findings support the suggestions from Post (1961) and Maloney (1993) that SWR could be increased by using thicker and longer particles, which are present in greater quantities in the M1 and M2 mixes. It also confirms the findings of Haselein et al. (2002) who observed an improvement in screw holding strength, when average core particle thickness increased from 0.5 to 1 mm. This is because fine particles are less effective in transferring stress from particle to particle compared with coarse or medium particles. The cumulative stress concentration locations in the furnish containing more finer particles is higher and lead to critical discontinuity in the stress field inducing more microfracture and ultimately failure (River 1994, Smith et al. 2003). The improvement in edge SWR with increasing coarse particles can also be attributed to the higher resin spread rate on coarse particles resulting from lower relative surface area of coarse particles as compared to fines (Duncan 1974, Moslemi 1974, Hill and Wilson 1978). Furthermore, higher compaction ratio leads to more inter-particle contact and better bonding as was the case here for the M2 boards with fewer fines. The effect of replacing coarse particles with fines is more pronounced in the case of IB strength.  107  5.3.2 Three layer particleboards 5.3.2.1 Summary of statistically significant effects of particle size mix on the properties of 3-layer particleboard The main effects of and interactions between density and particle size mix on the properties of 3layer particleboards are shown in As expected, board target density significantly affected all mechanical properties. There was a small but statistically significant increase in board density as the proportion of fine particles in the core mix increased. Particle size mix significantly affected board core density and IB. There was a borderline significant effect (p = 0.05) of core particle mix on edge SWR. MOR and MOE were not affected by particle size mix in the core.  Table 5.4  Effects of mixtures and target density on properties of 3-layer boards.  Effect  Core density  IB strength  Edge SWR  MOR  MOE  Density  p = 0.004  p < 0.0001  p = 0.0001  p < 0.0001  p < 0.0001  Mixture  p = 0.03  p = 0.0004  p = 0.05  n.s  n.s  Density* mixture  p = 0.04  p = 0.0009  p = 0.02  n.s  ns  n.s. = not significant at the 5% confidence level; * = interaction between density and mixture.  5.3.2.2 Mean density and VDP of the 3-layer particleboard The U-shaped VDP of the 3 layer boards are shown in Figure 5.4. As expected the core densities of the higher density (HD) boards were slightly higher than that of the lower density (LD) boards. As shown in Figure 5.4a, the core and face densities of the LD boards containing more fines (M3) and CTL were higher than the rest in the same density category. This was also the case for the UD boards, and could be attributed to the fact that fines are effectively compacted leaving very little void space (Nemli 2003), resulting in higher density. From Figure 5.4b HD 108  boards containing higher fines content in the core (M3) and CTL were thinner than those containing M1 and M2 mixes. Mats containing a lot of fines offer less resistance to compaction from more effective filling of void space than those containing thicker core particles (Maloney 1993). The greater thickness in M1 and M2 boards may also be attributed to higher residual stress existing in boards containing thicker particles leading to a greater spring back (Wong et al. 1999).  1200 LD = 650 kg/m 3  HD = 700 kg/m3  Density (kg/m3 )  1000  800  600 M1 M2 M3 CTL  400 top  bottom  top  200 0 (a)  3  6 9 12 Panel thickness (mm)  15  M1 M2 M3 CTL  18  0 (b)  3  6 9 12 Panel thickness (mm)  bottom 15  18  Figure 5.4 Vertical density profile of the 3-layer particleboard at a target density of (a) 650 kg/m3 and (b) 700 kg/m3.  Average values for properties of the 3-layer boards of the 2 target densities and 4 different core mixtures are shown in Table 5.5; IB strength and edge SWR are shown graphically in Figure 5.5. The LD boards containing the M2 mix (i.e. 60% coarse, 20% medium, and 20% fines) had the highest IB strength (0.66 MPa) and edge SWR (779 N), representing a 40% increase in IB strength and 18% increase in edge SWR compared with LD boards containing control mix. For  109  the HD boards, the M1 boards were 23% higher in IB and 18% higher in edge SWR than the CTL boards.  Table 5.5  Mean values for the properties of the 3 layer boards with novel furnish mix core. Target density: LD– 650 kg/ m3 IB  CD 3  M1 M2 M3 CTL  SWR  MOR  Target density: HD – 700 kg/ m3 MOE  CD 3  IB  SWR  MOR  MOE  (kg/ m )  (MPa)  (N)  (MPa)  (GPa)  (kg/ m )  (MPa)  (N)  (MPa)  (GPa)  552  0.55  731  18.3  2.87  590  0.75  923  22.2  3.34  (7.0)  (10.6)  (20.0)  (9.7)  (6.8)  (3.6)  (10.1)  (16.6)  (7.6)  (4.2)  548  0.66  779  19.6  2.94  588  0.65  764  20.9  3.26  (8.7)  (11.9)  (33.2)  (12.5)  (10.6)  (2.3)  (11.5)  (17.3)  (6.7)  (6.3)  538  0.41  585  18.3  2.77  574  0.67  857  20.8  3.24  (4.2)  (11.8)  (19.8)  (10.2)  (8.6)  (4.0)  (13.5)  (16.7)  (6.33)  (7.9)  604  0.47  660  17.8  2.75  586  0.61  783  21.6  3.41  (4.6)  (16.0)  (28.5)  (12.4)  (8.2)  (3.68)  (15.5)  (20.4)  (7.3)  (8.1)  CD = core density; coefficient of variation (COV) is given in parenthesis under each mean.  As can be seen in Figure 5.5, there is an interactive effect between particle mixtures and board density for IB and SWR. The most obvious feature is that the particle size effect on bonding and screw-holding is masked to a large extent in the HD boards. The greater compaction of the mat likely offset some of the adverse effects that high quantities of fines and dust had on less compacted boards by closing up void spaces. The interaction between board density and particle size mix is comparable to a study done by Fakhri et al. (2006) who found a similar interaction between density and core fines content affecting the transverse permeability of OSB boards.  110  0.80  1000 900  0.70  800  0.60  Edge SWR (N)  IB strength (MPa)  700  0.50 0.40 LSD = 0.077 0.30 0.20 0.10  400  LSD = 112.5  300 200  650 kg/m3  700 kg/m3  100  700 kg/m3  0  M1  Figure 5.5  500  650 kg/m3  0.00 (a)  600  M2 M3 Core particle mix  M1  control (b)  M2 M3 Core particle mix  control  Mean values of board densities for (a) IB strength and (b) edge SWR loads for  each core particle size mix and target board density. Note that the fines and dust content of boards increases from M1 to M3.  When compacted to higher density, the M1 mix (all coarse) produced higher IB strength and edge SWR than boards containing some fines. This is attributed to the fact that under the same ram pressure a mat containing 100% thick particles will undergo a higher compaction and therefore have better inter-particle contact (Heinemann et al. 2002). The reduced bond strength in the LD boards containing the M1 mix may have been caused by void space that were closed up in the HD boards resulting in a higher bond strength.  5.3.2.3 Effect of mixture on flexural properties of the 3-layer boards The flexural properties (MOR and MOE) of 3 layer boards as listed on Table 5.5 were not significantly affected by particle size mix in the core, and their values were all above the minimum required levels of ANSI A208.1 (1999). The controlled addition of fine particles to the 111  core of particleboard did not result in reduced bending strength and stiffness, because the flexural properties of particleboard are largely influenced by the composition and consolidation of face furnish material, pressing conditions, and density gradient (Kelly 1977), all of which were similar for the boards in this study.  5.4  Conclusions  Our initial hypothesis was that adding fine particles to the core of particleboard would improve consolidation and bonding in the core by filling void spaces, and that this would improve screw holding properties of particleboard. However, our results from laboratory made uniform density single layer boards indicate only a slight increase in bond strength and SWR after replacing 40% of the coarse particles with medium and fine particles, and decreased with further increase in fines content. Although the trend of results of the 3-layer did not follow that of the single-layer boards, the effect of fines in the particle mix of 3-layer boards was particularly perceptible in boards compressed to low density. Boards made with cores containing a customized mix of particle sizes were up to 40% higher in IB strength and 18% higher in edge SWR than boards made from industrial furnish. Results from uniform and low density 3-layer boards suggest that the core of commercial furniture grade particleboard contains too many fine particulates and dust, and this may be responsible for reducing edge SWR to below the suggested minimum levels. The ratios of particle sizes present in commercial particleboard core furnish may also not be optimized for the best particle packing efficiency. Increasing the coarse particle (> 2 mm) content of the core of contemporary particleboard and optimizing the particle size mixture of the core could improve edge SWR and IB strength of particleboard, especially for lower density boards. Flexural properties of the 3-layer boards with differential VDP were unaffected by core 112  fines content. For better evaluation of the effect of mixtures on 3-layer boards, thickness swell, and water absorption, further investigation using mixture design is being undertaken.  Since most plants partition particleboard furnish into 2 size classes (fine surface and coarse core material), our findings suggest that significant benefit in IB strength and edge SWR could be obtained by adopting three size classes of particles: coarse, medium, and fine.  113  5.5  Literature cited  ANSI A208.1-1999. American National Standard. Particleboard. Composite Panel Association, Gaithersburg, MD, USA. 11pp. American Society for Testing Materials. 2000. Annual book of ASTM standards 2000; Section four, Construction (wood). Vol. 04.01 West Conshohocken, PA, ASTM. Brumbaugh, J. 1960. Effect of flake dimensions on properties of particle boards. Forest Prod. J. 5:243-246. Conrad, M.P.C., G.D. Smith, and G. Fernlund. 2004. Fracture of wood composites and woodadhesive joints: A comparative review. Wood Fiber Sci. 36 (1):26-39. Duncan, T. F. 1974. Normal resin distribution in particleboard manufacture. Forest Prod. J. 24(6): 36-44. Fakhri H.R., K.E. Semple, and G.D. Smith. 2006. Permeability of OSB. Part I. The effects of core fines content and mat density on transverse permeability. Wood Fiber Sci. 38 (3): 450-462. Feng, M.W. and A.W. Andersen. 2004. Uniformity of UF resin distribution in MDF - in a mill study using the glue marker method. In Proceedings of 7th Pacific Rim Bio-Based Composites Symposium, eds, X. Zhou, C. Mei, J. Jin and X. Xu, Nanjing, China October 31-November 2, 2004. Haselein, C.R., Calegari, L., Barros, M.V., Hack, C., Hillig, E., Pauleski, D.T. and Pozzera, F. 2002. Mechanical strength and dimensional stability of particleboard made with different particle sizes. Ciencia Florestal 12(2): 127-134.  114  Heebink, B.G., and R. A. Hann. 1959. How wax and particle shape affect stability and strength of oak particleboards. Forest Prod. J. 9(7):197-203. Heinemann, C., P.E. Humphrey, and A. Frühwald. 2002. Evaluation of adhesive cure during hot pressing of wood-based composites. In: Proceedings of the International Symposium on Wood-based Materials, Wood Composites and Chemistry (WOOD Kplus) and COST E13 Final Meeting, Universität für Bodenkultur, Wien, Österreich, 19. – 20. September 2002: S. 163-170. Hill, M.D. and J. B. Wilson. 1978. Particleboard strength as affected by unequal resin distribution on different particle. Forest Prod. J. 28(11): 44-48. Kakaras, I.A. and A.N. Papadopoulos. 2004. The effects of drying temperature of wood chips upon the internal bond strength of particleboard. J. Inst. Wood Sci. 16 (5):277-279. Kelly, M.W. 1977. Critical literature review of relationships between processing parameters and physical properties of particleboard. Forest Prod. Lab. Gen. Tech. Rep. FPL 10, 64 pp. Kollmann, F.F.P., E.W. Kuenzi, and A.J. Stamm. 1975. Principles of wood science and technology II. Wood based materials Springer-Verlag, Berlin, Heidelberg, New York. Lei, Y.-K. and J. B. Wilson. 1980. Fracture toughness of oriented flakeboard. Wood Sci.12 (3): 154-161. Lynam, F.C. 1959. Factors influencing the properties of chipboard. J. Inst. Wood Sci. 2(4):14-27. Maloney, T. M. 1970. Resin distribution in layered particleboard. Forest Prod. J. 20(1): 43-52. Maloney, T. M. 1993. Modern Particleboard and Dry Process Fiberboard Manufacturing. 2nd ed. Miller Freeman, San Francisco, 681 pp.  115  Marra, G.A. 1954. Discussion following article by Turner. Forest Prod. J. 4(5):210-223. Moslemi, A. A. 1974. Particleboard. Volume 1: Materials. Southern Illinois Press. 244pp. Mottet, A.L. 1967. The particle geometry factor in particleboard manufacturing. In Proceedings 1st Washington State University Symposium on Particleboard, Pullmann, Washington, ed T.M. Maloney. pp. 23-73. Nemli, G. 2003. Effects of some manufacturing factors on the properties of particleboard manufactured from Alder (Alnus glutinosa subs. Barbata). Turk J. Agric For. 27: 99-104. Post, P.W. 1958. Effect of particle geometry and resin content on bending strength of oak flake board. Forest Prod. J. 8:317-322. Post, P.W. 1961. Relationship of flake size and resin content to mechanical and dimensional properties of flake board. Forest Prod. J. 11(9):34-37. River, B. H. 1994. Fracture of adhesive bonded wood joints. In: Handbook of adhesive technology. eds. Pizzi, A., Mittal, K.L. New York: Marcel Dekker Inc. Chapter 9. Smith I., E. Landis, and M. Gong. 2003. Fracture and fatigue in wood. West Sussex, John Wiley & Sons Ltd. Semple, K., E. Sackey, H. J. Park, and G. D. Smith 2005. Properties variation study of furniture grade M2 particleboard manufactured in Canada. Forest Prod. J. 55(12): 117-124. Talbott, J.W. and T.M. Maloney. 1957. Effect of several production variables on modulus of rupture and internal bond strength of boards made from green Douglas fir planer shavings. Forest Prod. J. 7(10):395-398.  116  Talbott, J.W. and M.D. Hill. 1978. Resin efficiency of commercial blenders for particleboard manufacture. Forest Prod. J. 28(2): 49-54. Wong, E.D., M. Zhang, Q. Wang and S. Kawai 1999. Formation of the density profile and its effects on the properties of particleboard. Wood Sci. Tech. 33(4): 327-340.  117  Chapter 6 Characterizing macro-voids in uncompressed mats and finished particleboard panels using Response Surface Methodology and X-ray CT5  6.1  Introduction  Wood composites consist of three major components: wood elements, adhesive, and voids. Voids can be classified into micro-voids (intra-particle voids found within wood elements) and macro-voids (inter-particle voids found between wood elements). Dai et al (2005) modeled void formation in strand-based composites and subdivided macro-voids into non-contact voids (voids between strands) and edge voids (voids created by strand overlaps). The core structure of a particulate wood composite encompasses a myriad of complex particle shapes and geometries that create numerous macro-voids. Large size disparity in wood element size has a profound influence on the internal mat structure and hence on the macro-voids formed during processing and those in the finished panels (Geimer et al. 1999). Dai et al. (2005, 2007) and Li et al. (2009) reported that macro-voids increased with increasing strand thickness, but decreased with increasing strand length and width. Intuitively, the effect of element thickness on void volume may be even greater in a particle mat since particles are chunkier in nature than strands.  5  A version of this chapter has been accepted for publication in Holzforschung, January, 2010. Sackey, E.K. and G.D. Smith. Characterizing macro-voids in uncompressed mats and finished particleboard panels using Response Surface Methodology and X-ray CT.  118  Macro-voids influence mat permeability, heat and moisture transfer, and viscoelastic compression behaviour during hot-pressing (Humphery and Bolton 1989, Kamke and Wolcott 1991, Wolcott et al. 1994, Dai et al. 2005, Lee et al. 2006). They also create flaws where delaminations can initiate (River, 1994) leading to lower panel mechanical properties (Wu and Lee 2002. During consolidation it is essential to minimize voids and maximize inter-particle contact in order to achieve the highest bond strength possible (Dai et al. 2007). Increasing the number and size of macro-voids in a wood composite mat can increase permeability during processing, whereas fewer and smaller macro-voids should increase inter-particle bonds. However, excessive densification of a composite mat leads to heavy panels, more wood usage (higher cost), increased thickness swell, and springback (Kelly 1977, Dai 2001). Therefore, there is the need to identify the best particle mixture that gives the optimum void fraction and distribution of voids to improve composite mat permeability and inter-particle contact points while minimizing density.  Various models have been developed to describe the packing efficiency (PE) (defined as the fraction of the volume of a unit cell actually occupied by particles) for crystal structures and powder packing, but these are mostly based on spherical particles (Westmann and Hugill 1930, Scott 1960, German 1989, Zou and Yu 1996, Yang et al. 2000, Zou et al. 2001). Mileweski (1978) investigated the combined packing of spheres and rods in reinforced plastics and demonstrated the effect of the ratio of sphere to rod diameter on void volume. Theoretical models for evaluating packing characteristics of particles for mono-sized, non-spherical fibrous particles (Yu et al. 1992, 1996, Zou et al. 1999) and for estimating porosity in granular materials have also been proposed (Yu and Standish 1991, Yu et al. 1993), most of which used response  119  surface methodology (RSM) and mixture designs. In a mixture experiment, the responses vary as a function of the proportion of the components, which must sum up to unity (Cornell 1981, Khuri and Cornell 1987, Zou et al. 1999, Stat-Ease, Inc. 2001).  Other experimental techniques like image analysis (Ellis et al. 1994) and X-ray CT-scanning (Sugimori and Lam 1999, Zhang et al. 2005, Wu et al. 2006) have also been employed to evaluate and visualize macro-voids in pressed wood composite panels. Most of these investigations were done on Parallam and OSB panels. Senden and Morrison (2004) used X-ray CT to investigate resin distribution in particleboard by tagging UF-resin with copper sulphate. Standfest et al. (2009) adopted the CT-scanning technology to compare the density profiles of particleboard, MDF, and OSB with those obtained using the conventional X-ray technique.  In our earlier work on particle mixtures (Sackey et al. 2008), it was observed that bonding in the core of particleboard was influenced by the compaction ratio during consolidation and void fraction. Void fraction is affected by size and shape of particles in the mixture and their distribution (German 1989, Yu et al. 1992). Ideal particle packing where void volume approaches that of veneer-based products is not possible with the typical technologies available for particle preparation and panel processing. However, Sackey et al. (2008) showed that it is possible to reduce void fraction in the core of particleboard through appropriate mixtures of particle sizes. This study aims to develop a model for estimating the macro-void volume of mats made from mixtures of different particle size classes. It was also designed to evaluate the macrovoid volume in finished particleboard panels made from these mixtures and establish the relationship between macro-voids in uncompressed and compressed mats of particulate wood  120  composites. The hypothesis is that minimizing macro-voids will increase panel mechanical properties. The main objectives were to:  1. Measure the void fraction of (i) surrogate particles (made of wood blocks) of known sizes and (ii) novel industrial particle mixtures, 2. Develop a predictive model for industrial particleboard core furnish that relates the proportions of the particle size classes in the mixture to the void fraction of the uncompressed mat and to the mechanical properties of the pressed panels, 3. Determine the void volume of panels produced with novel furnish mixtures using X-ray CT, and to 4. Characterize the changes in macro-voids between uncompressed mat and pressed particleboard.  6.2  Methodology  6.2.1 Scaled-up (surrogate) and industrial particle preparation The shape of industrial particles is irregular making it difficult to calculate the actual particle volume, so surrogate blocks of controlled dimensions were used as model particles. Several sizes of wooden blocks with dimensions 10 times larger than the mean dimensions of the particles, listed in Table 6.1, were cut and planed from Lodgepole pine (Pinus contorta var. latifolia). Thirty randomly selected wood blocks were conditioned at 20oC and 60% relative humidity until constant mass was obtained and their density and moisture content (MC) computed.  121  Table 6.1  Mean dimensions, circularity, and aspect ratio (AR) of each particle size class.  Particle  mean dimensions (mm)  type  Circularity  AR  (% of total furnish mass)  (%)  length  width  thickness  ≤ 0.5  > 0.5  > 0.6  > 0.8  ≤ 1.2  core-fine  3.94  1.01  0.35  44.6  55.4  37.3  11.4  27.5  medium  7.28  1.86  0.69  44.9  55.1  32.3  5.1  23.3  coarse  10.03  4.14  1.19  43.1  56.9  37.9  5.2  27.7  Three different sized graduated cylindrical containers were made from clear plastic sheets; their dimensions are listed in Table 6.2. To mitigate edge effects the diameter of the largest container was approximately four times the length of the longest surrogate block (Scott and Kilgour 1969, Zou and Yu 1996). Photographs of the containers and packing types used for particle packing are Figure 6.1.  Table 6.2  Cylinder dimensions for packing simulation.  Size  diameter (cm)  height (cm)  small  15.5  28.0  medium  25.0  43.0  large  35.5  49.0  Furnish for face and core industrial particleboard was obtained from a particleboard mill in Alberta, Canada. The particles were screened with a mechanical shaker table using square screen openings of 2.0 mm, 1.0 mm, and 0.5 mm (Tyler mesh sizes: 9, 16, and 32, respectively) as detailed in Sackey et al. (2008). The particles sizes were classified as core-fine (>0.5 mm), medium (>1 mm), and coarse (>2 mm).  122  (a) Surrogate particles  Rolling process  fine: coarse = 1:4  fine: coarse = 1:4  packing density = 42%  packing density = 51%  Random loose packing  Random dense packing  only coarse particles  only coarse particles  packing density = 28%  packing density = 37%  Random loose packing  Random dense packing  (b) Industrial particles  Rolling process  Figure 6.1  Packing sequence of (a) surrogate blocks and (b) industrial particles. Rolling  particles in cylinder (left), the resulting RLP (center) and the RDP after shaking the RLP (right).  6.2.2 Shape determination for scaled-up (surrogate) particles In order to decide on the shapes and dimensions of the surrogate blocks, the circularity and aspect ratio (AR) of samples of industrial particles were measured (Table 6.1). The circularity, C, of particles, which is defined as the ratio of its projected area, Ap (mm2), to the square of its perimeter, Pp (mm), were described by 3-D calliper measurements (Cox 1927, Huck 2005, Nakamura el al. 2005); i.e,  123  C=  4πAp 2 Pp  6.1  The factor 4π is included so that a circle has a circularity of one. Since a circle has the minimum perimeter possible to enclose an area, the circularity of actual particles is always less than one. In the limit, a very, very long thin particle would have a circularity approaching zero.  Images of the coarse, medium, and fine particle size classes were obtained by a flatbed scanner (Epson Perfection 4870, Pro Model J131A) set at 300 dpi and analyzed using Image J Version 1.38 software. After analyzing the outlines of the particle images, the circularity and aspect ratio of the particles were calculated. Typical particle outlines tend to be irregular trapezoids and parallelograms as shown in Figure 6.2. Particles with circularity greater than 0.6 looked essentially square. Hence particles with circularity greater than 0.6 and AR less or equal to 1.2 as listed in Table 6.1 were designated as square in shape, the remainder designated as rectangular. With respect to AR, about 30% of fine and coarse and 20% of medium particles were considered to be square and the rest rectangular.  124  10mm Figure 6.2  Typical particle outlines obtained with image analysis software. Note: Note scale is the  same for all particle sizes.  6.2.3 RSM-mixture ixture experiment Particles from the fine, medium and coarse particle size classes were mixed together such that their mass fractions added up to unity; the mixtures used aare re shown in the ternary diagram in Figure 6.3.. The two standard designs used in mixture experiments are the simplex lattice and simplex centroid designs (SCD), which evaluate the triangular response surface at the triangular vertices and centroids (Cornell 1981). Previous work (Sackey et al. 2008) showed that only one of the vertices, i.e. the coarse pure particles produced better er panel properties. Internal bond (IB) strength and edge screw withdrawal ithdrawal resistance (SWR) R) significantly decreased when the furnish contained more than 40% fine content or less than 40% coarse particles. Consequently, upper bounds were placed on the fine and medium particles and a lower bound on the coarse component proportions of the industr industrial ial furnish. An SCD with the constraints that fine ≤ 0.4,  125  medium ≤ 0.6, and coarse ≥ 0.4, as shown in Figure 6.3b, was therefore prepared using JMP 8 software (2008).  fine 1  fine  9  9  10 6 2 medium  11 13  12 5  8  4  6  3 coarse  7  10  medium  3  5 8  2  11 7  4 1 coarse  (a) Modified simplex centriod design (b) Constrained design for industry for surrogate particles. Figure 6.3  particles  Modified SCD for (a) surrogate blocks and (b) region of interest for industrial  particles using the constraints.  For packing of the surrogate blocks all three pure blends at the vertices were tested. In both designs, the modeling of response surface was calculated by Equation 6.2 (Cornell 1981), so that the response to any mixture over the entire simplex can be empirically predicted. Void fraction or porosity (P), defined as the ratio of void volume to total volume occupied by particles, was modeled using a special cubic model as follows:  3  2  P = ∑ βi X i + ∑ i =1  3  ∑  i =1 j = i +1  1  βij X i X j + ∑  2  3  ∑∑  βijk X i X j X k  i =1 j = i +1k = i + 2  126  6.2  where βi , βij , and βijk are fractional solid volume coefficients. Further details of this model can be found in Cornell (1981). The equation was also used to predict the particle packing density, (Pd) and porosity (P), which are expressed as follows:  Pd =  volume of particles total volume occupied by particles  6.3  The porosity described here is inter-particle void volume. The ternary plot of the constrained design in Figure 6.3 shows an augmented simplex design with interior test points and the elimination of some edge points. Thirteen test points for the surrogate blocks and eleven for industrial furnish were tested with three replicates each making a total of 39 and 33 runs respectively. The replicate blends were used to evaluate lack-of-fit (LOF).  6.2.4 Simulation of surrogate and industrial particle packing Particles were thoroughly mixed in a large container and then poured into the graduated cylinder to minimize particle segregation. Two packing techniques, random loose packing (RLP) and random dense packing (RDP) as described by German (1989) and Scott (1960), were adopted for the particle packing in this study. RLP was obtained by placing the lid on the cylinder top, gradually laying it on its side, and rolling it for three complete revolutions as shown in Figure 6.1. This was done three times. The cylinder was then gradually turned upright and the lid was lowered until it touched the surface of particles. RDP was obtained by pouring the particles into the cylinder, placing it on a shaker table and agitating it until the position of the lid became constant. The fine particles were placed on top of the packing system before shaking to mitigate particle segregation at the end state, which was similar to the methods applied in other 127  investigations (Scott 1960, Scott and Kilgour 1969, German 1989, Yu 1992). Actual wood volume of the surrogate blocks was computed from the dimensions of the individual particles, while that of the industrial particles was estimated based on the mass of particles and the average density of the Logdepole pine (Pinus contorta var. latifolia) parent material at 12% MC (430 kg m-3). Logdepole pine (Pinus contorta var. latifolia) was chosen as this is the dominant species in the region around the particleboard plant from which the furnish was obtained.  6.2.5 Void determination of pressed panels with X-ray computer tomography Twelve 3-layer particleboards were manufactured including control boards from the industrial particle mixtures produced from the blend matrix given in Table 6.3. Each panel measured 305 mm by 305 mm by 15.9 mm and had an average density of 673 kg m-3. Three replicates per mixture were produced. Samples measuring 43 mm by 125 mm were cut from the center of two panel replicates and from the edge of the third replicate for the CT-scan. The samples were scanned with a Somatom Senastion 64 CT scanner from Siemens at the Vancouver General Hospital (UBC Branch). It is equipped with a 0 MHU Straton X-ray tube with a generator peak power of 80 kW. Routine isotropic resolution of the system is 0.33 mm at a fastest rotation time of 0.33 s for 64 slices per rotation. The scanner was set at 80 kV, 200 mA, and 0.5 mm slice thickness. SyngoFastView software was used to view the 2-D image slices as shown in Figure 6.4a.  128  250 200 150 100 Gray level  threshold  50  core void  0 1 cm (a) 0.5 mm 2-D sliced image sample with a line across the width  Figure 6.4  0  50  100 150 200 250 300 Distance (Pixels)  (b) Line profile across image in (a) showing threshold value of 100  0.5 cm (c) 3-D image of sample showing core voids and faces  CT-imaging of particleboard core: (a) 2-D CT-image of sample, (b) Line profile  across sample width, and (c) 3-D sample image showing core voids. In order to differentiate voids from wood material, a macro was developed to determine the appropriate threshold value. Figure 6.4b shows a typical gray intensity graph along the sample width from 2-D sliced images of the panel. The minimum intensity was recorded in the dark regions of the images, which corresponded to the voids within the panel. Ten images from each sample were used to determine the average value that distinguished between wood and void; the numerical value for this threshold was found to be 100, similar to the threshold value used by Zhang et al. (2005). Binary images were created using this threshold value and the panel void fraction was determined. A typical 3-D visualization as shown in Figure 6.4c was reconstructed (VG Studio Max 1.21 software).  129  Table 6.3  Porosity of industrial furnish blends for RLP and RDP for industrial furnish and  average responses for RLP and RDP. Originalcomponents (-) x1 x2 x3 0 0 1 0 0.3 0.7 0 0.6 0.4 0.2 0 0.8 0.2 0.2 0.6 0.2 0.4 0.4 0 0.2 0.8 0 0.4 0.6 0.4 0.2 0.4 0.4 0 0.6 0.15 0.25 0.6  6.3  Pseudo components (-) x1′ x2′ x3′ 0 0 1 0 0.5 0.5 0 1 0 0. 0 0.67 33 0. 0.33 0.33 33 0.67 0 0. 33 0.33 0.67 0 0 0.67 0.33 0. 0.33 0 67 0 0. 0.33 67 0. 0.42 0.33 25  Porosity (-) RLP  RDP  0.719 0.723 0.728 0.713 0.716 0.713 0.725 0.724 0.705 0.700 0.718  0.632 0.626 0.624 0.599 0.607 0.600 0.618 0.621 0.579 0.584 0.620  Results and discussion  6.3.1 Surrogate particle packing The porosity of the blend matrix for the surrogate mixtures for RLP and RDP samples are shown in Table 6.4. The porosity of the RDP samples is lower than the RLP by about 0.08 or 15%. In general, the porosity of both RLP and RDP samples decreased with increasing amounts of corefine and medium blocks. The porosity in the RLP was highest at 0.605 in a binary blend containing 80% coarse and 20% core-fine blocks. This could be attributed to the wedging effect of the core-fine blocks which inhibited direct contact between large blocks and created smaller sized but more numerous voids.  130  Table 6.4  Porosity of surrogate particles for RLP and RDP. Porosity (-)  Particle volume fraction Core-fine  Medium  Coarse  RLP  RDP  1 0 0 0.2 0.2 0.2  0 1 0 0 0.2 0.4  0 0 1 0.8 0.6 0.4  0.600 0.581 0.577 0.605 0.581 0.578  0.484 0.498 0.507 0.488 0.490 0.497  0 0 0.4 0.33 0.25 0.25 0.15  0.2 0.4 0.2 0.33 0.25 0.2 0.25  0.8 0.6 0.4 0.33 0.5 0.55 0.60  0.574 0.554 0.560 0.561 0.565 0.582 0.567  0.501 0.486 0.476 0.469 0.488 0.481 0.497  The highest porosity in the RDP samples occurred in the 100% pure blend coarse blocks. As the blocks were shaken after the RLP measurements, they were observed to align themselves mostly along the wider block faces creating voids at their edges. It must be noted that the surrogate blocks were planed and could easily slide on one another which may not be the case with industrial particles due to their surface roughness and irregularity.  One of our goals was to identify interactions between the three particle sizes and their effects on voids. A reduced Scheffe’s canonical special cubic model (SCM) as described in Cornell (1981) and Khuri and Cornell (1987) was initially fitted to the mean void ratio of the 13 test points of the surrogate blocks for both the RLP and RDP samples. However, the model had a significant lack-of-fit (LOF) for the RLP sample. The orderliness of the large surrogate blocks in a random loose state was very low and might have contributed to the difficulty in finding a fit for the RLP 131  sample. The SCM fitted to the RDP results in Equation 6.4, where x1, x2, and x3 represents corefine, medium, and coarse blocks respectively, is shown in Figure 6.5. The models had no LOF and the parameters of all blends were significantly different from zero at p<0.0001. A plot of predicted and actual P yielded an adjusted R2 of 0.71. The values in parenthesis below each term in Equation 6.4 are the estimated standard errors (e.s.e) of the parameter estimates. Note: All statistical analyses were done with JMP 7.0.  P = 0.484 x1 + 0.498 x2 + 0.507 x3 - 0.314 x1 x2 - 0.09 x1 x3 - 0.071x2 x3 + 1.03 x1 x2 x3 ( 0.003 ) ( 0.003 )  ( 0.003 ) ( 0.092 )  ( 0.022 )  ( 0.014 )  ( 0.248 )  6.4  The magnitudes of the coefficients in Equation 6.4 indicate the extent to which the components influenced porosity; negative coefficients corresponded to a decrease in porosity and positive coefficient to an increase. Comparing the coefficients of the linear terms, porosity of the particle size types increased from core-fine to medium, and to coarse blocks. Contrary to the linear terms, all of the coefficients for the binary blends indicated a reduction in porosity and was largest for the core-fine and medium binary blends. Interestingly, the porosity of the ternary blend increased. This was attributed to the presence of coarse blocks because the two ternary blends with the coarse particles (i.e. x1x3 and x2x3) had lower negative parameter estimates and the coarse pure blend had the highest positive coefficient.  132  core-fine 0.475  0.475  0.484  0.493  0.501  medium  Figure 6.5  Fractional solid volume  coarse  Porosity of the RDP sample for surrogate blocks. Open dots indicate test mixtures  6.3.2 Industrial particle packing Owing to the constraints placed on the components, the following equations were used; x1′= x1/0.6; x2′ = x2/0.6; x3′= (x3-0.4)/0.6, where the xi′s are the various pseudo-components as shown in Table 6.3. The new vertices for the region of interest as shown in Figure 6.6 were determined using pseudo-components. In a pattern similar to the surrogate blocks, the estimated void volume of the industrial particles of the RDP was about 15% less than that of the RLP. There was also a general trend of decreasing porosity with increasing core-fine and medium particle contents for both the RLP and RDP samples. The many similarities between the packing behaviours of surrogate and industrial particles indicated that the wooden blocks were a good particle model for simulating the packing of core particles from industrial particleboard furnish.  133  As shown in Table 6.3, the highest porosity for the RLP industrial particles was 0.728 for the binary blend with 60% medium and 40% coarse particles, while the porosity of the pure coarse blend of the RDP samples was 0.632 and showed a similar trend as the surrogate blocks. The lowest porosity or the highest packing density was recorded for the mixture with the highest (40%) proportion of core-fine particles for both packing types. During mat formation, fine and medium particles result in more voids initially and as densification proceeds these particles offer less resistance to compression and are better packed as a result. The effect of chunky and irregular coarse particles becomes dominant in retaining voids at the latter stages of densification as was the case for RDP sample. This confirms other studies that void volume is dependent on the thickness of the wood elements (Dai and Steiner 1993, Dai et al. 2005, Li et al. 2009). The dominance of edge voids at the later stages of consolidation is also obvious, as indicated by Dai et al. (2005).  Reduced models consisting of the linear and cubic terms were fitted to the porosity of the RLP industrial particle samples as shown in Equation 6.5. The adjusted R2 was 0.89 with no LOF for all significant terms unlike the surrogate blocks. The model shows no binary interaction but there was an antagonistic interaction in the ternary blend. The highest porosity recorded at the test point with the highest proportion of medium particles, (Table 6.3), indicates that the presence of medium particles can also increase voids, especially for particle combination containing coarse particles. As shown in Figure 6.6, porosity of both RLP and RDP samples within the region of interest increases as the core-fines content of the mixture is reduced  134  P = - 0.48 + 1.15 x1 + 1.212 x2 + 1.201x3 - 0.243 x1 x 2 x3  6.5  ( 0.0053 ) ( 0.0015 ) ( 0.0015 ) ( 0.0456 ) core-fine  core-fine 0.58  0.58  0.7 0.708  0.594  0.715  0.608 0.723  0.621  0.73  medium  Fractional solid volume  coarse  medium  Fractional solid volume  coarse  (a)Industrial particles- porosity for  (b)Industrial particles- porosity for the  the RLP sample mats  RDP sample mats  Figure 6.6  Porosity of industrial particles for (a) RLP and (b) RDP samples. Dots indicate  test mixtures and the trapezium bounded by the dotted lines is the region of interest.  The RDP was fitted with an SCM as shown in Equation 6.6 and the relationship between actual and predicted P had an adjusted R2 of 0.91; this result differed from the RLP samples in that the binary terms were significant. Comparing the three quadratic terms to the cubic term, the coarse particle component has the strong net effect on increasing porosity and the medium-coarse blend (x2x3) had the lowest negative parameter estimate. However, the parameter of the core-finemedium blend (x1x2) with the highest negative coefficient, 0.666, indicates that the presence of core-fine particles reducing porosity substantially. The results indicate that the coarse particles have much greater impact on increasing porosity for the RDP samples compared to the RLP samples in which the presence of medium particles increases the porosity the most.  135  P = - 0.421 + 1.084 x1 + 1.068 x2 + 1.052 x3 - 0.666 x1 x2 - 0.247 x1 x3 - 0.069 x2 x3 + 1.257 x1 x2 x3 ( 0.0093 ) ( 0.003 )  ( 0.003 )  ( 0.0217 )  ( 0.0241 )  ( 0.0103 )  ( 0.0818 )  6.6  From Equation 6.6, the parameter estimates decreased linearly from fine to medium to coarse. For a constant volume system, more fine particles will be required to fill the system than coarse particles; hence numerous inter-particle voids will produce higher void volume and increase in porosity. As densification increases during board pressing, the presence of core-fines on porosity decreases drastically to 0.599 for a mixture containing 20% of core-fine compared with 0.632 for a pure coarse particle mat (Table 6.3). The effect of the coarse particles can also be seen in Figure 6.6b, where porosity decreased as the amount of coarse particles in the ternary blend decreased. These particle blend effects could be used to determine which combinations increase permeability during board manufacturing. It could also be exploited to achieve more interparticle contacts during compression to enhance inter-particle bonds.  6.3.3 Relationship between core void volume in uncompressed mat and finished panel Table 6.5 lists fractional solid volume contained in the novel mixtures used for panel manufacturing and their corresponding percentage voids. As expected the lowest percent voids were recorded in panels with the highest amounts of core-fines. Panels made from medium and coarse binary mixtures had higher void volumes. Interestingly, the coarse-only mixture did not produce the highest void percentage but a mid-range value of 6%. This could be due to the effect of thickness ratio (panel-to-element thickness ratio), compaction ratio, higher inter-particle contact, and to a lesser extent cell collapse, because of the higher compression resistance of thicker particles during hot-pressing. 136  Table 6.5  Core void volume in pressed panels for various industrial particle mixtures.  Panel  1  2  3  4  5  6  7  8  9  10  11  Particle  core-fine  0  0  0  0.2  0.2  0.2  0  0  0.4  0.4  0.15  volume  medium  0  0.3  0.6  0  0.2  0.4  0.2  0.4  0.2  0  0.25  fraction coarse  1  0.7  0.4  0.8  0.6  0.4  0.8  0.6  0.4  0.6  0.6  Voids (%)  6.03 9.53 11.99 9.21 8.15 6.64 10.99 11.47 5.08 4.84 8.50  Figure 6.7a show the correlations between panel porosity determined by X-ray CT and the fractions of the three particle size classes contained in the panel. The graph indicates an inverse relationship between panel porosity and percentage core-fine with an R2 of 0.6, while porosity shows an increasing trend with increasing medium particle content. Contrary to expectations based on the RDP results, there was no measurable correlation between panel porosity and coarse particle content (R2 = 0.064). This could be due to the range of coarse particle content used for making boards. The RDP state along the spectrum of particle mat compaction could be considered as the packing state between the moment just after pre-pressing and the moment during hot-pressing just before the wood plasticizes and resin viscosity increases due to curing. This is because particles of the RDP sample could not be packed any further without flattening or altering their shapes. The ease with which fine material is compressed has a direct influence on the overall porosity of the finished panel.  Figure 6.7b illustrates the relationship between percentage panel void and void fraction of uncompressed mat of the RDP and RLP samples. As expected, there is a general positive linear to exponential trend for both data sets. Void fraction in the RLP sample correlated better with panel void fraction with an R2 of 0.8, than the RDP sample, R2 of only 0.4. This suggests that panel porosity is determined more during the pre-press stage rather than at the final pressing  137  stage. Before the pre-press stage, particles are loosely touching each other and lots of particlewedging is in play, hence the presence of larger and numerous voids. This is exemplified in the RLP samples. During pre-pressing the voids are drastically reduced with higher inter-particle contact. In the hot-press, as the mat is compressed, there is lesser opportunity for rearrangement of the particles due to (i) increased friction through higher inter-particle contact resulting in reduced particle-particle slippage and (ii) increased tackiness of the resin due to curing. This might have led to the lower correlation between the void fractions in the RDP sample, which typifies the early stages of hot-pressing. Better wood particle packing can be obtained in wood composite production if pre-pressing is incorporated into the manufacturing process.  Percent void in pressed panel (%)  14  12  R² = 0.4141  12  10  10 R² = 0.0644  8  8 6  6 4  R² = 0.6012  2  4 Core-fine Medium Coarse  2  0 0  (a)  Figure 6.7  0.2 0.4 0.6 0.8 Particle size fraction in mixture (-)  1  y = 0.0056e11.915x R² = 0.428 RDP  y = 5E-10e 32.767x R² = 0.776 RLP  0 0.57 0.59 0.61 0.63 0.65 0.67 0.69 0.71 0.73 0.75  (b) Void fraction of uncompressed particle mat (-)  Correlations between porosity of particleboard panel core for (a) fractions of core-  fine, medium, and coarse particles used in the panel and (b) void fraction of uncompressed particle mat of the RDP and RLP samples.  6.3.4 Effects on IB strength and edge SWR of voids in RDP mat and finished panel Figure 6.8 shows the effect of voids in the RDP sample and panel void fraction on IB strength and edge SWR of particleboard panels. The mixtures have the same particle size fractions as in panels 1, 5, and 9 respectively listed on Table 6.5. Figure 6.8a IB and SWR increased to a peak 138  value and decreased with further increase in voids for lower density (LD) panels, while the opposite was true for higher density (HD) panels. The sensitivity of IB to changes in void fraction is also obvious. The LD M1 panels, which contained only coarse particles, had higher property values than their counterparts in the HD panels, indicating the effects of compaction in HD panels and packing in LD panels. As thicker particles are unduly compacted to higher densities cell collapse and crushing may occur resulting in reduced inherent particle strength. The HD M2 and M3 panels resulted in higher property values than M1, indicative of the greater packing effect as density increases in core-fine and medium particles. The compression of coarser particles resulted in greater inter-particle contact, better bonds, and mid-sized voids which led to better panel properties. The core-fine and medium particles were better packed during compression to a higher density. This likely resulted in a greater number of inter-particle bonds and better panel properties, even though the numerous particles contributed to a higher void fraction.  M2  0.80  1000  0.75 900  M3  0.60 M1  0.55 0.50  700  M3  600  0.45 0.40  IB (650 kg/m3)  IB (700 kg/m3)  0.35  SWR (650 kg/m3)  SWR (700 kg/m3)  0.30 0.57  0.58  0.59  0.6  0.61  0.62  0.63  M2  0.60 0.55 0.50  800  M1 700 M3  600  0.45 0.40  500  IB (650 kg/m3)  IB (700 kg/m3)  SWR (650 kg/m3)  SWR (700 kg/m3)  500  0.35 0.30 5.00  400 0.64  5.50  6.00  6.50  7.00  7.50  8.00  400 8.50  Percent void in pressed panel (%)  Void fraction in RDP (%)  Figure 6.8  M2  0.65  800 M2  IB strength (MPa)  0.65  900  0.70 M3  M1 Edge SWR (N)  IB strength (MPa)  0.70  1000 M1  0.75  Edge SWR (N)  0.80  Effect on IB strength and edge SWR of (a) percent void in RDP samples (b)  percent void in pressed panel relative to density. The IB and SWR results were taken from Sackey et al. (2008); Each point is an average of 48 samples for IB and 24 samples for edge SWR. 139  As shown in Figure 6.8b, IB and SWR increased gradually to a maximum with increasing percentage void in LD pressed panels and decreased again at a much higher void content. The peak was reached in the HD panels more rapidly compared to that in the LD panels. The best particle mixture for maximizing the board mechanical properties is one made from coarse particles and pressed to a lower density.  6.4  Conclusions  Within the confines of this study the following conclusions were made: 1. Packing of industrially produced hammer-milled wood particles can be simulated with wood blocks cut according to the major shapes and sizes of the actual particles using random loose packing (RLP) and random dense packing (RDP) techniques. 2. Using a three component (core-fine, medium, and coarse) mixture design, an RLP has about 15% more voids than a densely packed sample which has a maximum void fraction of 63.2%. 3. A reduced cubic model with only linear and cubic terms was suitable for predicting void fraction of a customized particle mat with RLP producing an adjusted R2 of 0.89, while a full cubic predictive model with adjusted R2 of 0.91 was capable of predicting void fraction in a RDP particle mat. 4. Increasing the core-fine content in an industrial particle mixture decreased panel void volume and increased particle packing efficiency. However, void volume increased with increasing coarse particles in the panel core; pure coarse particles have the highest void fraction.  140  5. Void fraction in finished panels decreased with increasing core-fine particles and to a lesser extent increased with increasing medium particles. 6. Void ratio in pressed panels increased exponentially with fractional void volume of RLP and RDP samples reminiscent of pre-pressed particle mat and early stages of hotpressing. 7. Depending on the purpose, the developed predictive models can be used to increase or decrease permeability of particle mat or increase packing efficiency to increase interparticle bond. 8. The pure coarse mixture was found to be the best particle mixture from which to manufacture particleboard with better mechanical properties at low density  6.5  Recommendation  Since resin will influence inter-particle friction and slippage of particles over one another, it is recommended that a further study be undertaken to ascertain the effect of resin on the particle behaviour with respect to void ratio and packing efficiency.  6.6  Acknowledgements  The authors would like to thank the Natural Science and Engineering Research Council of Canada (NSERC) and Natural Resources Canada’s Value to Wood Program for financial support of the work.  141  6.7  Literature cited  Cox, E.P. 1927. A method of assigning numerical and percentage values to the degree of roundness of sand grains. J. of Palaeontology 1(3), 179-183. Cornell, J.A. 1981. Experiments with mixtures: Designs, models, and the analysis of mixture data.. John Wiley & Sons Inc., New York, NY, 1981. Dai, C. and P.R. Steiner. 1993. Compression behaviour of randomly formed wood flake mats. Wood and Fiber Sci., 25(4), 349-358. Dai, C. 2001. Viscoelasticity of wood composite mats during consolidation. Wood and Fiber Sci., 33(3), 353-363. Dai, C., C. Yu, and X. Zhou. 2005. Heat and mass transfer in wood composite panels during hot pressing. Part II. Modelling void formation and mat permeability. Wood and Fiber Sci., 37(2), 242-257. Dai, C., C. Yu, and C. Zhou. 2007. Theoretical modelling of bonding characteristics and performance of wood composites. Part I. Inter-element contact. Wood and Fiber Sci. 39: 48-55. Ellis, S., J. Dubois, and S. Avramidis. 1994. Determination of parallam macroporosity by two optical techniques. Wood and Fiber Sci., 26(1), 70-77. Geimer, R. L., J. W. Evans, and D. Setiabudi. 1999. Flake furnish characterization - Modeling board properties with geometric descriptors. Res. Pap. FPL-RP-577. Madison, WI: U.S. Department of Agriculture, Forest Service, Forest Products Laboratory. 36p. German R. M. 1989. Particle Packing Characteristics. Metal Powder Industries Federation, Princeton, NJ, 1989 142  Huck, D. 2005. Image analysis coupled with classification – A powerful combination for the study of agglomeration. Technical Report, Malvern Instrument Ltd. Malvern, Worcestershire UK. Humphery, P.E. and A. J. Bolton. 1989. The hot pressing of dry-formed wood-based composites. Part II. A simulation model for heat and moisture transfer, and typical results. Holzforshung 43(3):199-206. JMP 8 Statistical Discovery Software. 2008. SAS Institute Inc. SAS Campus Drive, Building S, Cary, NC, 27513, Cary, NC, USA. Kamke, F.A. and M. P. Wolcott. 1991. Fundamentals of flakeboard manufacture: Woodmoisture relationships. Wood Sci. Technol. 25:57-71. Kelly, M.W. 1977. Critical literature review of relationships between processing parameters and physical properties of particleboard. Forest Prod. Lab. Gen. Tech. Rep. FPL 10, 64 pp. Khuri, A. I. and J.A. Cornell. 1987. Response surfaces: Designs and analyses. Marcel Dekker, Inc., New York, NY 10016, 1987. Lee, J. N., L. T. Watson, and F. A. Kamke. 2006. Compression behaviour of resinless oriented strandboard mat: Effect of internal mat environment and void volume. In: Recent developments in wood composites. Ed. Shuppe, T.F. Forest Prod. Soc. Madison WI, USA. pp 41-49. Li, P., C. Dai, and S. Wang. 2009. A simulation of void variation in wood-strand composites during consolidation. Holzforshung 63:357–361. Milewski , J.V. 1978. The combined packing of rods and spheres in reinforcing plastics. Ind. Eng. Chem. Prod. Res. Dev. 17 (4) 363-66  143  Nakamura, M., M. J. Castaldi, and N. J. Themelis. 2005. Measurement of particle size and shape of New York City Municipal solid waste and combustion residues using image analysis. Dept. of Earth and Environmental Eng., Columbia University, USA. River, B. H. 1994. Fracture of adhesive bonded wood joints. In: Handbook of adhesive technology. Eds. Pizzi, A., Mittal, K.L. New York: Marcel Dekker Inc. Chapter 9. Sackey, E., K. Semple, S-W. Oh, and G. D Smith. 2008. Improving core bond strength of particleboard through particle size redistribution. Wood and Fiber Science, 40(2), 214224. Scott, G.D. 1960. Packing of spheres. Nature. 188 (12) 908-909. Scott, G.D. and D. M. Kilgour. 1969. The density of random close packing of spheres. Brit. J. Appl. Phys. Ser. 2, Vol. 2, 863-66. Senden T. and O. Morrison. 2004. Picking the glue from the wood: Distinguishing resin in wood composites. Materials Monthly produced by the Australian National University Center for Science & Engineering of Materials. 5(8) 1-2. Standfest, G., A. Petutschnigg, M. Dunky, and B. Zimmer. 2009. Determination of density profile in wood based panels by means of computer tomography. Eur. J. Wood Prod. 67:83-87. Stat-Ease, Inc. 2001. Design-Expert 6 User’s guide. Statistical details – Design selection. Hennepin Square, Minneapolis, MN 55413-2726. Sugimori M. and F. Lam. 1999. Macro-void distribution analysis in strand-based wood composites using an X-ray computer tomography technique. J. Wood Sci. 45:254-257.  144  Westman, A.E.R. and H. R. Hugill. 1930. The packing of particles. J. Amer. Ceram. Soc., 13 (10) 767-79. Wolcott, M. P., F. A. Kamke, and D. A. Dillard. 1994. Fundamentals aspects of wood deformation pertaining to manufacture of wood-based composites. Wood and Fiber Sci. 26:496-511. Wu, Q., and J. N. Lee. 2002. Predicting the influence of voids on the engineering constants of oriented strandboard: a continuum model. In: Proc of the 6th Pacific Rim bio-based composites symposium & workshop on the chemical modification of cellulosics. Portland, OR, USA. Vol.2. 372-380. Wu, Q., B. Zhang, L. Wang, and G. Han. 2006. The application of 3-D X-ray tomography with finite element analysis for engineering properties of strand-based composites. In: Proc. of the 8th Pacific Rim bio-based composites symposium, advances and challenges in biocomposites: 20-23 November, 2006, Kuala Lumpur, Malaysia. pp 200-209 Yang, R.Y., R-P. Zou, and A-B. Yu. 2000. Computer simulation of the packing of fine particles. Physical review E. 62 (3) 3900-08. Yu, A-B. and N. Standish. 1991. Estimation of porosity of particle mixtures by linear-mixture packing model. Ind. Eng. Chem. Res. 30 (6) 1372-85. Yu, A-B, and R-P. Zou, and N. Standish. 1992. Packing of ternary mixtures of non-spherical particles. J. Am. Ceram. Soc. 75 (10) 2765-72. Yu, A-B, N. Standish. and A. McLean. 1993. Porosity calculation of binary mixtures of nonspherical particles. J. Am. Ceram. Soc. 76 (11) 2813-16. Yu, A-B., R-P. Zou, and N. Standish. 1996. Modifying the linear packing model for predicting the porosity of nonspherical particle mixtures. Ind. Eng. Chem. Res. 35 (10) 3730-41. 145  Zhang, B., Q. Wu, L. Wang, and G. Han. 2005. Characterization of internal void structure of strand-based wood composites using X-ray tomography and digital tools. In: Proc. of the McMat2005: 2005 joint ASME/ASCE/SES conference on mechanics and materials. June 1-3, 2005. Baton Rouge, LA. Paper 257. Zou, R-P., and A-B. Yu. 1996. Wall effect on the packing of cylindrical particles. Chemical Engineering Science, 51(4) Number 7, pp. 1177-1180. Zou, R-P. and A-B. Yu. 1996. Evaluation of the packing characteristics of mono-sized nonspherical particles. Powder technology 88:71-79. Zou, R-P., A-B. Yu, and X-Y. Lin. 1999. Packing of quaternary mixtures of fibrous particles. J. Am. Ceram. Soc. 82 (4) 933-938. Zou, R-P., C-L. Feng, and A-B. Yu. 2001. Packing density of binary mixtures of wet spheres. J. Am. Ceram. Soc. 84 (3) 504-508.  146  Chapter 7 Summary and conclusions 7.1  Project summary and conclusions  As particleboard is the major wood composite panel product used for furniture worldwide, this study focused on the two most important mechanical properties (IB strength and SWR) of particleboard for RTA furniture manufacturing and a means of improving them without increasing production cost. The overall research hypothesis was that redistributing particles within the panel mat by dividing contemporary commercial particleboard core particles into fines, medium, and coarse and systematically remixing them will increase the bonding properties of the core and the overall panel properties.  The project commenced by assessing the raw material and production process concerns and difficulties faced by the North American RTA furniture and particleboard producers with the focus on Canada. Particleboard plants were mostly concerned with wood and resin cost, with wood cost increasing more rapidly. Most of the furniture producing plants surveyed were faced with difficulties with panel surface smoothness and panel edge particle pull-out which leads to high adhesive consumption during edge banding. This required the assessment of the properties of particleboard produced in Canada.  A benchmarking study on physical and mechanical properties of RTA furniture grade M2 and MS panels produced by six different particleboard plants across Canada was conducted to ascertain the validity of the concern of particle edge pull-out by the PB users. Contrasting the 147  results against the ANSI 208.1 (1999) standards showed that IB strength of panels from all plants met the standard, while panels from only a third of the plants surveyed complied with the standard for MOR. Particle edge pull-out was assessed through edge SWR test. The results showed that over 80% of the tested panels did not meet the ANSI standard of 900N for M2 furniture grade particleboard, thus validating the concerns of particle edge pull-out by the RTA furniture producers.  Although MS grade particleboard is formulated and distributed as a lighter weight product with lower minimum property standard compared to the M2 grade, the experimental results suggest that the M2 and MS grade particleboards on the market from different producers are not standardized products. The survey also found that the standard screw prescribed for testing SWR is difficult to procure and is not the commonly used screw by RTA products. The use of a different screw with a thread pitch of 10 tpi resulted in higher SWR values compared with the standard 16 tpi sheet metal screw prescribed in the ASTM 1087 standard (2000).  After establishing that the edge SWR of most contemporary particleboard produced in Canada did not meet the minimum requirements, the next step was to develop a new panel with improved bonding in the core. The core particle configuration of the panels from the different particleboard plants were characterized and evaluated. A systematic study of the core particle configurations revealed significant differences in the particle size distributions of panel samples taken from different plants which partially explained the observed differences in panel properties. Mass fractions of the different particle sizes in the panels were compared by sieving the particles into seven particle size classes. The measurements indicated that 80% of the  148  particles from the panels surveyed were above the 0.5 mm mesh opening with about 35% of the total being retained on the 1 mm mesh opening. However conventional 3-layer particleboard furnish is divided into surface fine and core coarse particles where the cut-off is around 0.75 mm mesh opening (Besselt 2005). The size boundary between surface fine and core coarse particles was changed to a 0.5 mm mesh to reduce the amount of coarser particles in the surface and increase surface smoothness. The new size boundary value also increases the mass fraction of finer particles in the core that fills inter-particle voids and increases the frequency with which bonds are formed between particles.  The core particle configuration (> 0.5 mm mesh opening) was further partitioned into core-fine, medium, and coarse particle size classes as opposed to the conventional non-partitioned coarse core particle. Using particle dimensions (length, width, and thickness), aspect ratio (AR), and slenderness ratio (SR) as geometrical descriptors, the particle sizes of the core configuration of conventional particleboards from the six plants were characterized. With the exception of width for the medium particles, significant differences (p<0.0001) were found between plants for all geometrical descriptors. As expected it was observed that the finer particles had smaller size differences between plants, which were similar to the observation for mass percentages. The aspect ratio (AR) of the medium and coarse particles from the panels was found to correlate highly with edge SWR with an R2 of 0.82 for the medium particles and 0.84 for the coarse particles in agreement with the results of Lehmann (1974) and Lin et al. (2002), whereas AR of the fine particles correlated fairly well with IB strength with an R2 of 0.67.  149  Distribution models for the geometrical descriptors (length, AR, and SR) were developed for core-fine, medium, and coarse particles. Using the goodness-of-fit test and maximum likelihood, a lognormal density distribution function was found to best describe particle length, SR and AR of the entire core particle distribution and most especially the core-fine, medium, and coarse particle size classes. This is because most of the particle distributions have a heavier right tail. The results support an earlier work by Suchsland (1959) who found that the logarithm of particle width followed a normal distribution. The Gamma distribution function also partially fit the length and AR of the three particle core size classes, unlike the two-parameter Weibull distribution function. This is in contrast to a recent work by Lu et al. (2007) who found the 2parameter Weibull distribution to be the best fit for core particles of particleboard. The discrepancy between the two results may be attributed to the lower variability in the samples collected by Lu et al. since their core particle sample was from a single source.  The research on the conventional core particle distributions was used to develop a novel core particle configuration to improve panel properties for lower density, i.e. lower wood content, single- and 3-layer laboratory panels. The hypothesis was that adding fine particles to the core of particleboard would improve consolidation and bonding in the core by filling voids which would in turn improve IB strength and screw holding properties of PB. Results from the single layer panels partially supported this hypothesis in that the IB strength and edge SWR increased by 34% and 18% respectively after replacing 40% of the coarse particles with medium and core-fine particles, supporting the previous findings of Talbott and Maloney (1957), Nemli (2003), and Kakaras and Papadopoulos (2004). However, a further increase in core-fines above 40% decreased these properties indicating there is a threshold value for fines content above which the  150  properties decrease. Although the un-screened, as-received industrial furnish (control) furnish had fewer fines than 2 of the novel particle mixtures (M2 and M3), their IB and SWR results were lower. This initiated the proposition that changes in the property values were not caused purely by changes in fines content, but also by the proportions of the various particle sizes in the furnish.  While the findings from the 3-layer panels, especially for the high density panels, did not follow the same trend as the single layer ones due to the masking effect of VDP and density, the IB strength increased 40% and edge SWR 18%. The effect of fines in the particle mixtures of the 3layer panels was particularly pronounced in the low density panels. Furthermore particle size proportions in commercial panels are not optimized to produce the best particle packing efficiency.  Using mixture design methods to understand the particle size class interactions within the novel particle configuration, model wooden blocks and industrial particles were used to simulate particle mats. Macro-voids fraction in the mats was measured and regression models developed. X-ray CT was used to determine the ensuing macro-voids of the laboratory produced panels and correlated with the particle mats. By means of a three component (core-fine, medium, and coarse) mixture design, it was observed that a random, loosely packed wood particle system of irregular shape had a maximum void fraction of 63.2%, 15% greater than a densely packed system. Reduced and full cubic regression models with adjusted R2 of 0.89 and 0.91 were fit to loosely packed and densely packed particle mats respectively. Findings from the particle mats using the novel particle configuration suggested that increasing core fine content decreases void  151  ratio supporting the report by Nemli (2003). In contrast, increases in coarse particles increased void fraction, and is the reason the pure coarse particle mixture had the highest measured void fraction. Using X-ray CT, a similar trend was observed with void ratio in pressed panels, An increase in medium particles also slightly increased void ratio in the pressed panels. Consequently, void fraction in pressed panels increased exponentially with those in modeled particle mats. It was concluded from the research that a pure coarse mixture was the most cost effective particle mixture to manufacture low density particleboard with high IB strength and edge SWR properties.  7.2  Significance and limitations of the studies  As the panel property benchmark is the first study of its kind on particleboard panels produced across Canada, the results serves as a baseline for furniture manufacturers as to what they can realistically expect in terms of panels properties. It also serves as an independent assessment of board properties for the particleboard industry and as a quality control tool. This will assist the industry in identifying which properties need to be improved. It is known that about 80% of consumers who buy RTA furniture are satisfied with their purchase (French 2009); this work informs the public about the properties of the RTA furniture on the market, especially with the issue of edge SWR. However, the benchmark study is limited in its scope, i.e. it did not consider panels imported into Canada, which may be lower in quality. In addition the panels tested were limited in space and time, because properties from panels produced from the same line and plant may differ from season to season or even from day to day production.  152  Knowledge of the particle distribution models opens avenues for further work in obtaining optimal particle packing efficiencies for particulate wood composites. These models will also assist in formulating particle distributions that will increase or decrease porosity of a panel mat, which affects water vapour flow and permeability during hot pressing and could lead to particle mixtures that reduce degassing times (Kamke 2003, Thoemen and Klueppel 2008). Mat compression behaviour can also be influenced with the knowledge of particle size distribution and its variability. However due to their empirical nature, they are limited in their application and are likely not valid for different experimental conditions.  As wood prices keeps increasing due to saw mill closures and distance from particle source, the developed novel furnish opens an avenue by which the beleaguered panel industry can save money. The savings could be realized by reducing particle consumption through production of low density panels with increased mechanical properties. The particle mixture will also help reduce the edge particle-pull-out and reduce glue consumption during edge banding in the furniture plants. The developed empirical models can be used to formulate particle mixture with a targeted porosity or permeability. It can also be used to influence packing efficiency in particle furnish to enhance inter-particle bonds and serve as a tool in the formulation of particle configuration in panel production.  However, the implementation of the novel mixture requires initial investment in the particle sifting process. The entire particles must be screened into surface fines, core-fines, medium, and coarse. The latter 3 particle types will then be metered and remixed as core particles before blending. Depending on the blending strategy of a company, the initial investment will involve  153  adding 2 new particle storages for core-fine and medium particles, a metering system for remixing, and a conveyer which will transport the new mixture to join the existing line for blending.  The efficacy of the novel mixture was tested only for the mechanical properties and not for thickness swell and water absorption. It has been reported that dimensional stability of particleboard increases with particle thickness pressed to high densities (Suchsland 1959, Post 1961, Kimoto et al. 1964, Mottet 1967, Maloney 1993). Hence it is possible that the novel furnish with more coarse particles could have an increased dimensional stability if pressed to higher density. However, the models can further be used to find formulations which will optimize all the properties. Similar to the distribution models, this is also an empirical model and hence is dependent on the experimental conditions.  In conclusion, a new particle mixture was developed for the particleboard industry to be used as particleboard core furnish. The mixture consists of core-fines, medium, and coarse particles. Empirical models were also developed to configure the 3 component mixture for panel property optimization purposes. The mixture has the potential of improving contemporary particleboard IB strength by about 40% and edge SWR by about 18% for a panel 7% lower in density compared with current conventional particleboards.  7.3  Recommendations and further research  Procuring a Type A sheet metal screw with 16 tpi as required by the ASTM 1087 standard proved difficult and could not be supplied by any local hardware company. Furthermore, almost 154  all RTA furniture manufacturers surveyed do not supply this type of screw with their products. A similar screw but with 10 tpi used for the study also gave higher SWR values than the standard screw required. It is therefore recommended that a further study be conducted to standardize and harmonize an appropriate screw for SWR testing for quality control purposes.  Particles in 3-layer conventional particleboard is generally partitioned into fine surface and coarse core material, but my findings suggests that significant benefit in IB strength and edge SWR could be obtained by adopting three size classes of particles: coarse, medium, and fine. From the study, pure coarse particles may be recommended as the best core particles furnish to yield the highest mechanical properties. This particle size class can be produced by removing all particles passing through mesh opening of 2 mm from the core particleboard furnish. In order to avoid a possible decrease in panel dimensional stability, a combination of core-fine, medium, and coarse particles in the ratio of 20/20/60 is further recommended.  Since resin influences inter-particle friction and slippage of particles over one another, it is recommended that a further study be conducted to ascertain the effect of resin on the particle behaviour with respect to void ratio and packing efficiency. A further investigation on the effect of the novel mixtures in 3-layer boards on dimensional stability using mixture design is also appropriate. A study to incorporate the void ratio empirical models into the distribution models is further recommended.  155  7.4  Literature cited  American National Standards Institute (ANSI). 1999. ANSI A208.1 Particleboard. 11 pp. Besselt, N. 2005. Technical discussion at NewPro particleboard plant in Wahnam AB with the Technical Director on May 9th 2005. French, D. 2009. A consumer’s view of RTA. Furniture/Today and HGTV 2009 Consumer Views Survey. [Cited on October 2, 2009]. Available from http://www.furnituretoday.com/blog/Research_Says/23540-A_consumer_s_view_of_RTA.php  Kakaras, I.A. and Papadopoulos, A.N. 2004. The effects of drying temperature of wood chips upon the internal bond strength of particleboard. Journal of the Institute of Wood science. 16 (5):277-279. Kamke, F. A. 2003. Physics of hot pressing. Winandy, J. E.; Kamke, F. A., Eds. Fundamentals of composite processing. Proceedings of a workshop; November 5.6, 2003; Madison, WI. Gen. Tech. Rep. FPL-GTR-149. Madison, WI: U.S. Department of Agriculture, Forest Service, Forest Products Laboratory. 118 pp. Kimoto, K., Ishimori, E., Sasaki, H., and Maku, T. 1964. Studies on the particle boards Report 6:.Effects of resin content and particle dimension on the physical and mechanical properties of low-density particle boards. Departmental Bulletin Paper. KURENAI: Kyoto University Research Information Repository.15pp. Lehmann, W.F. 1974. Properties of Structural Particle boards. For. Prod. J. 24(1): 19-26. Lu, J. Z., Monlezun, C.J., Wu, Q., and Cao, Q.V. 2007. Fitting weibull and lognormal distributions to medium-density fiberboard fiber and wood particle length. Wood and Fiber Science, 39(1), 82-94.  156  Mottet, A.L. 1967. The particle geometry factor in particleboard manufacturing. In Proceedings 1st Washington State University Symposium on Particleboard, Pullmann, Washington, ed T.M. Maloney. pp. 23-73. Nemli, G. 2003. Effects of some manufacturing factors on the properties of particleboard manufactured from Alder (Alnus glutinosa subs. Barbata). Turkish Journal of. Agriculture and Forestry. 27: 99-104. Post, P.W. 1961. Relationship of flake size and resin content to mechanical and dimensional properties of flake board. Forest Prod. J. 11(9):34-37. Suchsland, O. 1959. An analysis of particleboard process. Quarterly Bulletin, Michigan Agricultural Experiment Station, Michigan State University, 42(2):350-372. Suchsland, O. and W.S. Good. 1968. The selection of panel materials by furniture and cabinet manufacturers. Michigan Agr. Expt. Sta. J. Article No. 4378. Michigan State Univ., East Lansing, MI. 11 pp. Talbott, J.W. and T.M. Maloney. 1957. Effect of several production variables on modulus of rupture and internal bond strength of boards made from green Douglas fir planer shavings. Forest Prod. J. 7(10):395-398. Thoemen, H. and Klueppel, A. 2008. An investigation on the permeability of different wood furnish materials. Holzforschung 62:215-222.  157  Appendices  158  Appendix A.  Questionnaire for particleboard manufacturers  Objectives: To identify the types and grades of industrial particleboard used in the secondary wood products industry for making furniture and to benchmark the properties of these in terms of internal bond strength, screw-holding ability, and other relevant properties.  1. Please where is your manufacturing plant situated? 2. What types of particleboard products do you produce at your mill and which thickness are the most common? (Example: H-1, 2, 3; M-1, S, 2, 3; LD-1, 2 etc or 8mm, 11mm, and 25mm)? 3. What is your average total production of your mill/year? a. Can you give me the breakdown into specific products? 4. What types of resin do you use and which of them is the most common? 5. How much resin do you use per year? Average amount of liquid resin used/year Non-volatiles content (solids content) of resin (%) 6. How is the furnish resinated in your plant? 7. Do you currently have any method of detecting the resin content either on the furnish or in the finished board? If yes, please describe. 8. Do you think the resination can be improved? If so, please describe. 9. In your mill, which board property is most sensitive to adhesive content? [ ] Face screw holding [ ] Edge screw holding [ ] Internal bond 159  [ ] Thickness swell [ ] MOR [ ] MOE 10. Please rate how important the following items are to your company: not important (1) ; of interest (2);  keen to know more (3); very keen (4)  A. Knowledge of the relationship between internal bond and edge or face screw holding ability? B. Knowledge of how the resin is distributed on the furnish? C. Measurement of the resin content and the distribution on the resinated furnish? D. Measurement of the resin content and the distribution on the finished board? 11. Do you have a method that estimates the amount of furnish surface area covered by resin? 12. If a method were available to accurately measure the amount of surface area covered by resin on the particles, how frequently would this information be needed? 13. Do you do screw and nail test in your mill and if yes which type of screws, nails, and other fasteners do you normally test on your products? 14. Who buys your products? i.e., Furniture manufacturers, Home Depot, Lowe, small retailers, home owners, etc. 15. What are the most common questions asked and demands made by your customers? 16. Is there anything else about the resination of particleboard that we haven’t asked but would like to comment on?  160  Appendix B.  Questionnaire for furniture manufacturers  Objectives: To identify the types and grades of industrial particleboard used in the secondary wood products industry for making furniture and to benchmark the properties of these in terms of internal bond strength, screw-holding ability, and other relevant properties.  1. What types of furniture products do you produce in your company and which are most common? 2. Do you manufacture sub-components for furniture parts in-house or are they purchased from suppliers? 3. How much particleboard does your company use on an annual basis? 4. Do you manufacture particleboard in-house or purchase it from suppliers? If the particleboard is purchased, from who is it bought? 5. How is the particleboard used? For example what percentage of your annual consumption goes into cabinets, chairs, etc?) 6. What grade and type of particleboard is used (H-1, 2, 3; M-1, S, 2, 3; LD-1,2 etc.)? 7. What are your specifications for each type and grade of particleboard? Do your suppliers consistently meet these requirements? 8. What are the key properties of the particleboard for your application? 9. How does your company monitor if the supplier is providing board with the specified properties?  161  10. What property needs to be improved the most and how will that change affect your operation? Internal bond strength, edge and face screw holding ability. 11. Which fastener tests are routinely performed in your company? Are there other tests that are used only occasionally–what are they? 12. Is there another test that may work better for characterizing board properties for your application? 13. What type of nails, screws, and other fasteners do you normally use for test samples? 14. Who is the supplier of the fasteners (nails, screws, etc.)? 15. Who buys your products? i.e., Home Depot, Lowe, Small retailers, home owners, etc. 16. What are the most common questions asked and demands made by your customers? 17. Is there anything else about the use of particleboard for manufacturing RTA furniture that we haven’t asked, but you would like to comment on?  162  Appendix C. Photographs and schematic drawing of screw used for screw withdrawal test  Screw A  Screw B  (a)Photographs of screws used for SWR test  (b)Micrograph of a screw showing thread and root  Where: Th = thread height, mm Tp = thread pitch, mm Dt = root diameter, mm Dr = shank diameter, mm Th = Dt – Dr  (c)Schematic drawing of a screw thread and root  163  Appendix D.  Design of screw withdrawal testing apparatus  SWR TESTING APPARATUS (ALL TOLERANCES =±0.51 mm)  266.7 mm  ⅜”  108mm  101.6 mm CLAMP PLATE  9.53 mm  6.35 mm HOLDING STEEL PLATE  YOKE (FRONT VIEW)  164  120.65mm  266.7 mm 254 mm 190.5 mm  42.86 mm  152.4 mm 3.125 mm R 101.6 mm  101.6 mm  25.4 mm 9.53 mm D, x 4  25.4 mm  15.88 mm D  YOKE (TOP VIEW)  228.6 mm 23.81 mm  190.5 mm  9.53 mm UNF, x 4  12.7 mm R 152.4 mm 101.6 mm 101.6 mm  25.4 mm 25.4 mm  CLAMP PLATE  165  101.6 mm  152.4 mm  15.88 mm D  HOLDING PLATE 11.11 mm UNF x 31.75 DEEP  19.05 mm D  254 mm  12.7 mm D MILL 0.0625 mm R, MILL HOLE BOTTOM & SIDES FLAT 6.35 mm 4.76 mm SLOT, 0.48 19.05 mm  COUPLING  166  Appendix E. resistance  (a)  Photographs showing testing for screw withdrawal  (b)  (a)Figure shows the testing for face SWR and (b) shows testing for edge SWR  167  Appendix F. Histograms of particle sizes of hydrolyzed sampled panels superimposed with the best distribution fit for particle length, AR, and SR 12 10  8  Lognormal Weibull Gamma  8  10 Plant A size : fine mean: 4.72mm n : 40  6  Plant A size : fine mean: 17.83mm n : 40  8  Lognormal Weibull Gamma  6 Count  Count  10  Plant A size : fine mean: 4.13mm n : 40  Lognormal Weibull Gamma  6  Count  14  4  4  2  2  4  0  1  2  3 4 5 6 7 Particle length, mm  8  9  0  10  14  Plant B size : fine mean: 2.04mm n : 40  12  12  6 4  14 12  Count  10 8  8 6  0  0.5 1  1.5 2 2.5 3 3.5 4 Particle length, mm  0  4.5  4  6 4 2 0 0 1 2 3 4 5 6 7 8 9 10 11 12 Particle length, mm  2 0  0  1  2  3 4 5 Aspect ratio  6  7  8  14 12  Lognormal Weibull Gamma  10 8  10  20  30 40 50 60 Slenderness ratio  70  80  14 12  8  4  4  2  2 2  4  6  8 10 12 14 16 18 Aspect ratio  Lognormal Weibull Gamma  10  6  0  Plant C size : fine mean: 15.13mm n : 40  16  6  0  0  20 Plant C size : fine mean: 4.51mm n : 40  16  Lognormal Weibull Gamma  8 6  20  Plant C size : fine mean: 3.94mm n : 40  Lognormal Weibull Gamma  10  2  Count  0  Lognormal Weibull Gamma  4  2  Plant B size : fine mean: 13.11mm n : 40  14 12  10 Count  Count  8  0 5 10 15 20 25 30 35 40 45 Slenderness ratio  16  Plant B size : fine mean: 2.22mm n : 40  10 Lognormal Weibull Gamma  0  0 1 2 3 4 5 6 7 8 9 10 11 12 Aspect ratio  Count  0  Count  2  0  0  10  20  30 40 50 60 Slenderness ratio  App F-1a Histograms of fine particle size data sets from panels of plants A, B, and C.  168  70  80  12  18  Plant D size : fine mean: 2.87mm n : 40  10  20  Plant D size : fine mean: 3.14mm n : 40  16  16  14 12 Count  Count  8  Lognormal Weibull Gamma  6 4  10 8  14 Count  Lognormal Weibull Gamma  2  3 4 5 Particle length, mm  6  7  Plant E size : fine mean: 3.15mm n : 40  16 14  Lognormal Weibull Gamma  Count  12 10  Plant E size : fine mean: 3.99mm n : 40  10  6  6  4  1  2  3 4 5 6 particle length, mm  7  0  8  6 4 2 0  0  1  2  3 4 5 6 particle length, mm  7  4  6 8 10 Aspect ratio  8  12  14  Lognormal Weibull Gamma  0  10 20 30 40 50 60 70 80 90 Slenderness ratio  12  Plant F size : medium mean: 21.34mm n : 40  11 10 9  Lognormal Weibull Gamma  14 12 10  7 6 5 4  6  3  4  2  2  1 0  0  2  4  6 8 10 Aspect ratio  12  14  Lognormal Weibull Gamma  8  8  0  Plant E size : fine mean: 25.47mm n : 40  6  16  Plant F size : fine mean: 5.04mm n : 40  16 Count  Count  8  2  18  Lognormal Weibull Gamma  40  0 0  20  10  35  2  22  Plant F size : fine mean: 4.23mm n : 40  12  15 20 25 30 Slenderness ratio  4  2  0  10  8  Count  0  5  10  4 2  0  12  Lognormal Weibull Gamma  8  8  0  0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Aspect ratio  12  Count  18  2  14  20  10  4  Count  1  0  12  6  4 2  0  Lognormal Weibull Gamma  8  6  2  Plant D size : fine mean: 12.42mm n : 40  18  0  5  10 15 20 25 30 35 40 45 50 Slenderness ratio  App F-1b Histograms of fine particle size data sets from panels of plants D, E, and F.  169  Lognormal Weibull Gamma  50 40 30  180 160  Lognormal Weibull Gamma  80 60  100 60 40  10  20  0  0 2  4  6 8 10 12 14 16 18 20 Particle length, mm  140  220  Plant B size : medium mean: 5.97 mm n: 600  100  180  Lognormal Weibull Gamma  80  0  Lognormal Weibull Gamma  140 120 100  0  2  4  140  6 8 10 12 14 Particle length, mm  16  120  20  25 0  0  2  4  6  0  8 10 12 14 16 18 20 22 Aspect ratio  Plant C size : medium mean: 3.62 mm n : 600  Count  120 100  60  80  40  60 40 20 0  2  4  6 8 10 12 Particle length, mm  14  16  0  0  2  4  6  8 10 12 14 16 18 Aspect ratio  Count  Lognormal Weibull Gamma  140  20  125  50  160  80  Lognormal Weibull Gamma  75  180  Lognormal Weibull Gamma  100  80  100  200  Plant C size : medium mean: 5.80 mm n : 600  70  Plant B size : medium mean: 12.55 mm n : 600  150  40 0  30 40 50 60 Slenderness ratio  175  60  20  20  200  80  40  10  225  160  60  0  250  Plant B size : medium mean: 3.43 mm n: 600  200  Count  120  0  0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 Particle aspect ratio  Count  0  Count  120 80  40 20  Count  Lognormal Weibull Gamma  140  20  0  Plant A size : medium mean: 15.02 mm n: 600  200  Count  60  Plant A size : medium mean: 5.64 mm n : 600  100  Count  70  Count  120  Plant A size : medium mean: 9.44 mm n : 600  80  300 275 250 225 200 175 150 125 100 75 50 25 0  10  20 30 40 Slenderness ratio  50  Plant C size : medium mean: 11.54 mm n : 600 Lognormal Weibull Gamma  0  10  20  30 40 50 60 Slenderness ratio  App F-2a Histograms of medium particle size data sets from panels of plants A, B, and C.  170  60  70  80  100 80  140  120  30 20 10 0 2  4  6  250  8 10 12 14 Particle length, mm  16  200  60  40  40  20  20 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Aspect ratio  180 160  Lognormal Weibull Gamma  125 100  100 80  4  6  80 70 60  125  40  75  Count  Count  50 40 30 20 10 0  50 25 0  Lognormal Weibull Gamma  0  2  4  6  8 10 12 14 16 18 20 22 Particle length, mm  150 100  8 10 12 14 16 18 20 Particle length, mm  Plant F size : medium mean: 8.03 mm n : 600  175  60  130 120 110 100 90 80 70 60 50 40 30 20 10 0  2  4  6 8 10 Aspect ratio  12  14  0  16  Lognormal Weibull Gamma  0  220  Plant F size : medium mean: 4.72 mm n : 600  10  20  30 40 50 60 Slenderness ratio  70  80  Plant F size : medium mean: 14.24 mm n : 600  200 180 160  Lognormal Weibull Gamma  140 Count  2  Lognormal Weibull Gamma  200  0  0  Plant E size : medium mean: 12.05 mm n : 600  250  20  25  10 15 20 25 30 35 40 45 50 Slenderness ratio  275  75 50  5  225 Lognormal Weibull Gamma  120 Count  150  0  300  Plant E size : medium mean: 4.34 mm n : 600  140  175  80  60  0  18  Plant E size : medium mean: 7.57 mm n : 600  225  80  Lognormal Weibull Gamma  100 Count  50  Lognormal Weibull Gamma  100 Count  60  Plant D size : medium mean: 11.18 mm n : 600  140  Count  Lognormal Weibull Gamma  40  Count  160  Plant D size : medium mean: 4.00 mm n : 600  120  70 Count  160  Plant D size : medium mean:6.93 mm n : 600  90  120 100 80 60 40 20  0 2 4 6 8 10 12 14 16 18 20 22 24 26 Aspect ratio  0  0  10 20 30 40 50 60 70 80 90 100 Slenderness ratio  App F-2b Histograms of medium particle size data sets from panels of plants D, E, and F.  171  Plant A size : coarse mean: 3.50 mm n : 900  120  Lognormal Weibull Gamma  160  100 Count  120 80  Plant A size : coarse mean: 12.13 mm n : 900  200  Lognormal Weibull Gamma  80  Count  160 Count  140  Plant A size : coarse mean:12.26 mm n : 900  200  60  Lognormal Weibull Gamma  120 80  40  40  40  20  0  0  5  400  10 15 20 25 Particle length, mm  30  0  35  2  3  4  5 6 7 8 Aspect ratio  300  Plant B size : coarse mean: 8.67 mm n : 900  350  0 0 1  9 10 11  5  10  15 20 25 30 Slenderness ratio  300  Plant B size : coarse mean: 2.20 mm n : 900  250  0  35  40  Plant B size : coarse mean: 8.80 mm n : 900  250  300  150  Lognormal Weibull Gamma  200 Count  Count  200  Lognormal Weibull Gamma  200 Count  Lognormal Weibull Gamma  250  150  150  100  100  50  50  100  0  5  350  10 15 20 25 Particle length, mm  0  35  250  2  3  4  5 6 7 Aspect ratio  150  8  9  12 16 20 24 Slenderness ratio  26  30  Plant C size : coarse mean: 8.25 mm n : 900  300  Lognormal Weibull Gamma  150  8  4  350  Lognormal Weibull Gamma  250 200 150  100  100  100  50  50 0  0  10  Plant C size : coarse mean: 2.39 mm n : 900  200  Lognormal Weibull Gamma  200  1  250  Plant C size : coarse mean: 9.16 mm n : 900  300  Count  30  Count  0  Count  50  0  5  10 15 20 25 Particle length, mm  30  35  0  50 0 1  2  3  4  5 6 Aspect ratio  7  8  9  0  5  10 15 20 25 30 35 40 45 Slenderness ratio  App F-3a Histograms of medium particle size data sets from panels of plants A, B, and C.  172  160  Lognormal Weibull Gamma  150  Lognormal Weibull Gamma  120 100  100  40  50  20 0 0  0 2 4 6 8 10 12 14 16 18 20 22 24 26 Particle length, mm  3  4 5 6 Aspect ratio  8  0  9  Lognormal Weibull Gamma  Lognormal Weibull Gamma  120  100  0  5  350  140  150  7  Plant E size : coarse mean: 2.85 mm n : 900  160  Count  Count  2  180  Plant E size : coarse mean: 9.27 mm n : 900  200  1  100 80  10 15 20 25 30 35 40 45 Slenderness ratio  Plant E size : coarse mean: 8.53 mm n : 900  300 250 Count  250  200 150  60  0  Lognormal Weibull Gamma  250  80  50  Plant D size : coarse mean: 7.69 mm n : 900  350 300  140  100  400  Plant D size : coarse mean: 2.77 mm n : 900  180  Count  200  Count  200  Plant D size : coarse mean: 9.64 mm n : 900  Count  250  Lognormal Weibull Gamma  200 150  60  100  40  50  50  20 1  2  3  4 5 6 Aspect ratio  7  Lognormal Weibull Gamma  200 150  160  120  80  5  10 15 20 25 Particle length, mm  30  35  0  8 12 16 20 24 28 32 36 40 44 Slenderness ratio  Plant F size : coarse mean: 10.38 mm n : 900 Lognormal Weibull Gamma  150 100  40  0  4  200 Lognormal Weibull Gamma  100 50  0  250  Plant F size : coarse mean: 2.60 mm n : 900  Count  Plant F size : coarse mean: 10.53 mm n : 900  250  0  0  8  200  300  Count  0  0 2 4 6 8 10 12 14 16 18 20 22 24 26 Particle length, mm  Count  0  50  1  2  3  4  5 6 7 8 Aspect ratio  9  10  0  0  5 10 15 20 25 30 35 40 45 50 55 Slenderness ratio  App F-3b Histograms of medium particle size data sets from panels of plants D, E, and F.  173  Appendix G. Table of model parameters and maximum likelihood of lognormal and Weibull distributions fit to coarse, medium, and fine particle sizes  Plant  Scale µ  A B C D E F  1.2887 0.6246 1.132 0.9823 1.0847 1.3696  A B C D E  1.4040 0.6679 1.1526 1.0145 1.2608 1.4833  F A B C D E F  2.7771 2.2075 2.3306 0.5717 2.8491 2.9102  Fine particles Lognormal Shape Maximum Scale σ loglikelihood β Length 0.5324 -31.0366 2.0961 0.4335 -22.8159 2.4880 0.6881 -41.2961 1.4997 0.3814 -17.6930 2.6758 0.3475 -13.9742 2.6822 0.3809 -17.6395 2.6872 Aspect ratio 0.5985 -34.3854 2.0132 0.5075 -29.12 2.5226 0.8286 -48.7316 4.8427 0.4961 -28.2102 3.5537 0.4763 -26.5811 4.5219 0.5213 -30.1924 5.7259 Slenderness ratio 0.7805 -46.3393 23.8696 0.8257 -48.5887 13.848 0.8103 -47.8386 15.839 2.3446 -33.8865 14.047 0.9472 -54.0805 27.152 0.5613 -33.1542 24.221  174  Weibull Shape Maximum γ loglikelihood 4.6790 2.3096 4.4023 3.2303 3.5320 4.7621  -31.7272 -24.6865 -44.7904 -20.9986 -19.5397 -21.0038  5.3756 1.8939 1.2095 1.8127 1.8521 1.9784  -35.7149 -33.9516 -52.9478 -34.5849 8.532488 -33.4204  1.2379 1.1357 1.1076 1.6643 1.1911 1.9534  -51.0464 -54.0194 -55.0479 -34.5849 -54.5547 -34.8597  Medium particles Plant  Scale µ  A B C D E F  2.4281 2.0871 2.1359 2.1886 2.1296 2.1994  A B C D E F A B C D E F  1.6137 1.1191 1.1763 4.1153 1.3780 1.3698 2.5681 2.3699 2.2627 2.2797 2.3572 2.4801  Lognormal Weibull Scale Shape Maximum Shape Maximum σ loglikelihood β γ loglikelihood Length 0.3721 -254.7833 13.5318 3.0613 -258.1943 0.3409 -268.7444 9.5972 2.8406 -282.6799 0.3591 -245.3903 10.1496 2.7845 -253.5037 0.3449 -389.0745 10.5712 3.0778 -339.4733 0.3535 -190.8005 10.0014 3.0843 -216.1705 0.3988 -434.4120 11.0006 2.5638 -413.1292 Aspect ratio 0.4821 -413.101 6.3892 2.0049 -472.431 0.4576 -381.768 3.8752 1.9123 -481.245 0.4984 -433.069 4.0903 2.0306 -458.079 0.5668 -510.174 4.5133 3.9973 -472.908 0.4208 -331.468 4.8963 4.3377 -390.851 4.7788 -567.371 5.3213 4.7507 -591.981 Slenderness ratio 0.5187 -457.022 16.9942 1.7710 -538.44 0.5480 -489.926 14.1746 12.6571 -572.29 0.5689 -512.391 12.9462 1.5047 -623.94 0.5201 -458.575 12.6553 1.9025 -508.99 0.4942 -427.982 13.6482 1.8184 -520.97 0.5884 -532.625 16.0479 1.6438 -593.18  175  Coarse particles Plant  Scale µ  A B C D E F  2.4344 2.1011 2.1518 2.1765 0.3664 2.2719  A B C D E  1.1468 0.6882 0.7690 0.9868 0.9711 0.8503  F A B C D E F  2.3849 2.0627 2.0197 1.9987 2.0258 2.2166  Lognormal Scale Shape Maximum σ loglikelihood β Length 0.3873 -422.753 13.7822 0.3334 -287.847 9.7188 0.3510 -334.169 10.2805 0.3707 -383.38 10.5782 2.1397 -372.83 10.1633 0.4054 -464.042 11.8878 Aspect ratio 0.4707 -598.440 3.9647 0.4313 -519.653 2.4956 0.4448 -547.326 2.7063 0.4215 -499.077 3.3027 0.4332 -523.653 3.2714 0.4531 -564.131 2.9443 Slenderness ratio 0.4697 -596.391 13.7351 0.4064 -466.085 9.6986 0.4184 -492.370 9.3291 0.3777 -400.301 8.9863 0.4229 -391.377 9.4672 0.4828 -621.188 11.7654  176  Weibull Shape Maximum γ loglikelihood 2.7813 2.7262 2.7700 2.8572 2.9808 2.4538  -452.461 -421.867 -427.961 -423.846 -397.335 -541.753  2.2910 2.0887 2.2234 2.5320 2.4289 2.1999  -629.397 -662.568 -634.361 -537.753 -569.468 -644.940  2.1103 2.2584 2.2011 2.1543 1.9702 1.9025  -674.213 -592.133 -612.834 -591.334 -532.979 -746.684  Appendix H. Table of AIC and -2loglikelihood values for testing lognormal, gamma, and Weibull distributions fit to coarse, medium, and fine particle sizes  Plant A B C D E F A B C D E F A B C D E F  Fine particles Length Lognormal Gamma -2*LL AICc -2*LL AICc 161.74 166.06 160.98 165.30 95.60 99.92 96.05 100.38 173.16 177.48 177.52 181.84 113.97 118.29 115.39 119.72 114.72 119.05 117.23 121.55 144.84 149.17 146.34 150.66 Aspect ratio 178.83 183.15 176.35 180.68 111.65 115.98 115.24 119.56 189.67 194.00 196.29 200.61 137.67 142.00 142.35 146.68 154.04 158.36 159.80 164.12 179.06 183.38 181.49 185.82 Slenderness ratio 286.79 291.12 287.10 291.43 273.77 278.10 282.20 286.53 282.12 286.45 294.42 298.74 255.34 259.67 261.29 265.62 336.09 340.41 336.15 340.48 299.13 303.45 300.25 304.57  Weibull -2*LL AICc 163.13 167.45 99.34 103.66 180.14 184.47 120.58 124.90 125.86 130.18 151.57 155.90 176.77 121.32 198.10 150.35 168.33 185.51  181.09 125.64 202.42 154.68 172.66 189.83  288.52 284.64 296.54 266.47 337.04 302.54  292.85 288.96 300.87 270.80 341.36 306.86  where -2*LL = -2LogLikelihood  177  Plant A B C D E F A B C D E F A B C D E F  Medium particles Length Lognormal Gamma -2*LL AICc -2*LL AICc 3124.98 3129.00 3108.69 3112.71 2600.90 2604.92 2582.44 2586.46 2445.24 2449.26 2447.73 2451.75 2878.76 2882.78 2852.10 2856.12 2746.33 2750.35 2739.30 2743.32 3234.44 3238.46 3190.29 3194.31 Aspect ratio 2762.42 2766.44 2790.67 2794.69 2106.23 2110.25 2170.16 2174.18 2275.49 2279.51 2327.52 2331.54 2526.19 2530.21 2430.45 2434.47 2316.69 2320.71 2338.42 2342.44 2778.27 2782.29 2778.09 2782.11 Slenderness ratio 3995.80 3999.82 4055.05 4059.07 3823.69 3827.71 3892.52 3896.54 3740.18 3744.20 3851.67 3855.69 3652.80 3656.82 3674.70 3678.72 3684.60 3688.62 3757.85 3761.87 4041.41 4045.43 4086.35 4090.37  where -2*LL = -2LogLikelihood  178  Weibull -2*LL AICc 3131.80 3135.82 2628.77 2632.79 2523.98 2528.00 2868.60 2872.62 2797.07 2801.09 3191.88 3195.90 2881.24 2305.33 2235.67 2450.98 2435.35 2827.70  2885.27 2309.35 2239.69 2455.00 2439.37 2831.72  4158.58 3988.46 3963.18 3753.65 3870.60 4162.53  4162.60 3992.48 3967.20 3757.67 3874.62 4166.55  Coarse particles Length Gamma Plant -2*LL AICc A 5217.27 5221.29 B 4420.91 4424.92 C 4579.49 4583.50 D 4726.08 4730.10 E 4622.40 4626.42 F 5047.65 5051.66 Aspect ratio A 3261.42 3265.43 3259.26 3263.27 B 2278.15 2282.16 2381.61 2385.63 C 2478.79 2482.80 2532.35 2536.36 D 2712.00 2716.01 2719.54 2723.55 E 2794.62 2798.63 2815.89 2819.90 F 2658.91 2662.93 2820.61 2824.62 Slenderness ratio A 5485.72 5489.73 5523.01 5527.02 B 4645.06 4649.07 4715.42 4719.43 C 4620.02 4624.03 4683.02 4687.04 D 4641.22 4645.24 4531.69 4535.70 E 4757.78 4761.80 4906.29 4910.30 F 5232.38 5236.39 5320.97 5324.98 where -2*LL = -2LogLikelihood Lognormal -2*LL AICc 5227.39 5231.40 4357.60 4361.61 4541.54 4545.55 4730.89 4734.90 4634.73 4638.75 5017.54 5021.55  Weibull -2*LL AICc 5286.81 5290.82 4625.64 4629.65 4729.13 4733.14 4814.89 4818.90 4680.11 4684.12 5172.96 5176.98 3323.25 2564.02 2652.85 2794.74 2912.55 2705.92  3327.26 2568.03 2656.86 2798.75 2916.56 2709.93  5641.33 4897.19 4861.05 4704.71 5145.86 5483.32  5645.34 4901.20 4865.06 4708.72 5149.88 5487.34  179  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

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"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0069234/manifest

Comment

Related Items