Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Drying and equalization of western hemlock to Japanese equilibrium moisture content Wallace, John Wilkes 2001

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

Item Metadata

Download

Media
831-ubc_2001-0286_fixed.pdf [ 9.98MB ]
Metadata
JSON: 831-1.0090424.json
JSON-LD: 831-1.0090424-ld.json
RDF/XML (Pretty): 831-1.0090424-rdf.xml
RDF/JSON: 831-1.0090424-rdf.json
Turtle: 831-1.0090424-turtle.txt
N-Triples: 831-1.0090424-rdf-ntriples.txt
Original Record: 831-1.0090424-source.json
Full Text
831-1.0090424-fulltext.txt
Citation
831-1.0090424.ris

Full Text

DRYING AND EQUALIZATION OF WESTERN HEMLOCK TO JAPANESE EQUILIBRIUM MOISTURE CONTENT By JOHN WILKES WALLACE B.Sc, The University of British Columbia, 1994  A THESIS SUMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in  THE FACULTY OF GRADUATE STUDIES DEPARTMENT OF WOOD SCIENCE ;"'_' ' FACULTY OF FORESTRY We accept this thesis as conforming To the required standard  The University of British Columbia April 20, 2001  © John Wilkes Wallace 2jQOl  In  presenting  degree freely  at  this  the  thesis  in  partial  fulfilment  of  University  of  British  Columbia,  I agree  available for  copying  of  department publication  this or of  reference  thesis by  this  for  his thesis  and  scholarly  or  her  for  of  The University of British Vancouver, Canada  Columbia  Date , >Ayr<v \ Z&.  O/.  DE-6 (2/88)  I further  purposes  gain  requirements that  agree  may  representatives.  financial  permission.  Department  study.  the  be  It  shall not  that  the  be  an  advanced  Library shall make  permission  granted  is  for  by  for  the  understood  that  allowed  without  it  extensive  head  of  my  copying  or  my  written  Abstract Density sorted, matched-samples of Western hemlock {Tsuga heterophylla (raf.)[SargP were dried in three different drying technologies in order to quantify the physical changes of kiln dried wood when exposed to the equilibrium moisture contents of a typical Japanese winter.  Moisture sorption and desorption was monitored for 14 weeks by  sampling board weight, dimensions, shell and core moisture contents and warp.  All three technologies, a conventional kiln, a radio frequency vacuum kiln and a superheated steam vacuum kiln are capable of drying to a target average moisture content of 19% at the core with acceptable standard deviation. Product quality was good with all three drying technologies. The difference of 7% in equilibrium moisture content between Tokyo and Vancouver is large enough to elicit a response in shell moisture content, dimensions and warp. However, the responses found should be expected when wood is examined in context of its natural variability. Drying Western hemlock to 19% moisture content at the core should be considered the maximum target MC. A target MC of 15% moisture content at the core would optimize the drying time and stability as the lumber is equalized to Japanese conditions.  n  Table of Contents ABSTRACT  [___.'  ii  LIST OF TABLES  V  LIST OF FIGURES  •.  :^J  SYMBOLS AND ABBREVIATIONS  :'</>i»  ACKNOWLEDGEMENTS CHAPTER 1 1.1 1.2  LITERATURE REVIEW  7 :  ....7  Wood Chemistry Wood Ultrastructure  7 9 14 23  Conventional Kilns Superheated Steam Vacuum Kilns Radio Frequency Vacuum Kilns  26 31 32  ,  35  GENERAL METHODS AND MATERIALS  MATERIALS PHASE I: DRYING  3.2.1 3.2.2 3.3.3 3.3.4 3.3  6  PHYSICAL PROPERTIES OF WESTERN HEMLOCK  CHAPTER 3 3.1 3.2  4 5  WATER AND WOOD INDUSTRIAL DRYING  2.3.1 2.3.2 2.4.3 2.5  1 4  Kiln Drying Post-drying Equalization to Japanese EMC  CELLULAR STRUCTURE OF WOOD  2.1.1 2.1.2 2.2 2.3  1  OVERVIEW OF THESIS  CHAPTER 2 2.1  GENERAL OVERVIEW  INTRODUCTION OBJECTIVES  1.2.1 1.2.2 1.3  llK  37  :  Conventional drying: Superheated Steam Vacuum Drying: Radio Frequency Vacuum Drying: Kiln Dry Measurements  ,  ;  37 ., 39  39 39 40 41  PHASE II: EQUALIZING TO WINTER JAPANESE EQUILIBRIUM MOISTURE CONTENT  42 CHAPTER 4 4.1 4.2 4.2 4.3 4.4 4.5  DRYING ANALYSIS AND DISCUSSION  44  DENSITY SORTING CONVENTIONAL KILN DRYING SUPERHEATED STEAM VACUUM KILN DRYING RADIO FREQUENCY VACUUM KILN DRYING MOISTURE CONTENT SUMMARY SHRINKAGE SUMMARY  44 44 51 58 64 65  iii  CHAPTER 5  5.1 5.2 5.3 5.4 5.5  EQUALIZING ANALYSIS AND DISCUSSION  EQUALIZING CONDITIONS CONVENTIONALLY DRIED SAMPLES SUPERHEATED STEAM VACUUM DRIED SAMPLES RADIO FREQUENCY VACUUM DRIED SAMPLES STABILITY SUMMARY  67  :.....'. :  67 68 72 :. 76 80  CHAPTER 6  CONCLUSIONS AND RECOMMENDATIONS  85  REFERENCES  88  APPENDIX A  PHOTOGRAPHS  91  APPENDIX B  ANOVA TABLE  99  IV  List of Tables Table 1. Conventional kiln schedule  39  Table 2. Superheated Steam Vacuum kiln schedule  40  Table 3. Moisture content results from conventional drying  45  Table 4. Shrinkage results from conventional drying  46  Table 5. Casehardening results from conventional drying  50  Table 6. Moisture content results from SSV drying  52  Table 7. Shrinkage results from SSV drying  53  Table 8. Casehardening results from SSV drying  57  Table 9. Moisture content results from RFV drying...  59  Table 10. Shrinkage results from RFV drying  60  Table 11. Casehardening results from RFV drying  63  J  List of Figures Figure 1: The molecular structure of cellulose  8  Figure 2. The structure of softwood lignin  9  Figure 3. The cell wall (Adapted from Walker, 1993)  11  Figure 4. The three phases of the drying curve  19  Figure 5. Warping of Timbers (adapted from Simpson, 1999)  :  21  Figure 6. Cutting pattern of green lumber  38  Figure 7. Casehardening measurements  41  Figure 8.  46  Initial and final moisture content distribution of conventional samples  Figure 9. Warp results for conventional HBD and LBD drying Figure 10. Drying curves and drying rates for conventional drying Figure 11.  49 ,  Initial and final moisture content distribution of SSV samples  50 53  Figure 12. Warp results for LBD and HBD SSV drying  56  Figure 13. Drying curves and drying rates for SSV drying  57  Figure 14. Initial and final moisture content distribution of RPV samples  60  Figure 15. Warp results for LBD and HBD RFV drying  62  Figure 16. Drying curves and drying rates for RFV drying  64  Figure 17. Equilibrium moisture content conditions in the humidity chamber  67  Figure 18. Weight changes during equalization  69  Figure 19. Moisture content and dimensional response of conventionally dried samples during equalization  70  Figure 20. Warp results between conventionally dried and equalized conditions  71  Figure 21. Weight changes during equalization  73  Figure 22. Moisture content and dimensional response of SSV dried samples during equalization  74  Figure 23. Warp results between SSV dried and equalized conditions  75 .  Figure 24. Weight changes during equalization  77  Figure 25. Moisture content and dimensional response of RFV dried samples during equalization Figure 26. Warp results between RFV dried and equalized conditions  78 79 VI  Figure 27. Coefficient of variation for stability of conventionally dried samples  82  Figure 28. Coefficient of variation for stability of SSV dried samples  83  Figure 29. Coefficient of variation for stability of RFV dried samples  84  Symbols and Abbreviations dM/dt  Drying rate  Gb  Specific gravity (basic)  Gn  Specific gravity (nominal)  G0  Specific Gravity (ovendry)  h  Partial relative humidity  HBD Ki K2 LBD  High basic density Coefficient from Siau's model of equilibrium moisture content Coefficient from Siau's model of equilibrium moisture content Low basic density  M  Moisture content  Mc  Core moisture content  Me  Equilibrium moisture content  Mf  Final moisture content  MfSp  Fibre saturation point  Ms  Shell moisture content  Vg  Green or wet volume  W  Coefficient from Siau's model of equilibrium moisture content Weight ofovendry sample  Wovendry W mc  Weight of sample at specified M  viii  Acknowledgements I would like to acknowledge the following people for contributing so generously their time and consideration for this project.  My supervisor, Dr. Stavros Avramidis, my  advisory committee: Drs. J. David Barrett, Simon Ellis and Ian D. Hartley. Also, thanks are due to Dr. Luiz Oliveira for his encouragement and stewardship. Forintek Canada Corp. provided the opportunity to undertake this project; their financial and equipment support is appreciated. Additionally, Weyerhaeuser Company and HeatWave Drying Systems for their respective in-kind contributions of the Western hemlock material and the use of the Radio Frequency Vacuum kiln.  IX  Chapter 1 1.1  General Overview  Introduction  On January 17, 1995, the Hyogoken-Nanbu earthquake, measuring 7.2 on the Richter scale (JMA) shook the Kobe, Japan region and caused billions of dollars of damage. The earthquake destroyed 150,000 buildings and left 300,000 people homeless. The ensuing investigations said many factors came together to cause such massive destruction. It was determined that the primary reason for such wide destruction was that the region had grown quickly since World War II and had primarily been built on reclaimed landfill. When the earthquake struck, the reclaimed land liquefied, the old structures, often not built to resist lateral shaking, crumbled to the ground. Too often crumbling structures knocked down adjacent buildings, which had withstood the shaking.  Inadequate  engineering of buildings and poor maintenance are cited as the two main structural factors contributing to the mass destruction.  Lack of reinforced concrete found in  modern buildings explained why some newer concrete buildings crumbled while others remained standing. Poor maintenance of the traditional Japanese Post-and-Beam house was cited as to why so many wood-framed homes collapsed. Until recently, most homes in Japan were not maintained to last more than one generation and therefore were highly susceptible to decay. The violent shaking of the earthquake caused the decay-weakened structures to crumble under the weight of their tile roofs or fall off their foundations due to poor connectors.  In June 2000, the Japanese government moved to implement performance guarantees for all buildings now built in Japan with the Housing Quality Assurance Law. This new law 1  was intended to lessen the impacts of another large earthquake hitting a major urban centre by ensuring that buildings used new construction techniques to resist strong lateral loads and are built with the best-suited products available in the market. In the Japanese case, the best-suited product is wood.  When dry, wood's physical attributes make it an excellent building material. Its strength to weight ratio is very high, and when used properly, it is able to withstand the strong lateral inertia loads typical of strong earthquakes. Many species have an inherently high resistance to rot. Species with low natural durability can be treated with preservatives, such as Chromated-Copper-Arsenate (CCA) or borate salts, to produce products with excellent resistance to decay. The Athena Institute found that when compared to steel and concrete, wood is more environmentally friendly when the materials whole life cycle, cradle-to-gate, is considered (Trusty and Meil, 1997). Solid wood also compliments the Japanese "Healthy House" program for its lack of harmful emissions which are frequently associated with engineered lumber products.  There were approximately 500,000 wooden housing starts in Japan in 1998. Increasingly, Japanese homebuilders are tending to use more kiln-dried lumber in their construction projects to take advantage of the superior strength, stability and durability attributes of kiln-dried lumber. With the shear magnitude of these construction projects coupled with an aging, skilled workforce, the Japanese have turned to automation to meet their demand issues. Pre-cut facilities now perform the complex joinery work in a factory that was previously done on the job-site by master carpenters. The number of  2  pre-cut facilities has grown from 181 plants in 1986 to 890 plants in 1998 (Gaston et al. 2000).  These automated facilities now produce enough joinery to supply 210,000  housing units annually. The raw material required for these facilities must be relatively free of warp, otherwise the machine centres will reject the lumber before manufacturing. If the pre-cut timbers distort after manufacturing due to moisture content issues, the precise joinery will not fit during final assembly at the job site. Either scenario is undesirable and the resultant economics have further driven the demand for change from wet or 'green' material to a consistently stable kiln-dried material.  "Performance-  guarantees", mandated by the Japanese government, stipulates that homebuilders will have to guarantee the performance of the house for at least 10 years.  This has  compounded the growing demand for consistently stable building materials.  The changes in consumer requirements have been a challenge for lumber producers in coastal British Columbia (BC) who had previously exported large volumes of green lumber to Japan. Many producers only had limited experience in producing a consistent, kiln-dried, solid wood product for the Japanese market. Furthermore, the necessary kiln capacity to dry the volume of wood produced for Japan simply does not exist, considering it can take 20 days to kiln-dry large wooden posts. In an effort to remain competitive, BC's coastal producers are increasingly investing in research to investigate new methods of producing a consistently stable, kiln-dried solid wood product.  This study was designed to add necessary information for lumber producers, by kilndrying 114 mm by 114 mm Western hemlock (Tsuga heterophylla (Raf)[Sarg.p for the  3  Japanese market.  It takes the kiln-drying process one step farther, and qualitatively  assesses the performance of BC's kiln-dried lumber under simulated Japanese climate conditions.  1.2  Objectives  1.2.1  Kiln Drying  Developing and implementing new technologies is a fundamental component of the growth in any industry. There have been three successful cases in coastal BC for kilndrying Western hemlock (called hemlock hereafter). The first and most common is the conventional kiln dryer. Essentially, hot air is circulated in a closed structure. The hot air transfers heat to the lumber and removes excess water as it dries. It is a slow process but it is a relatively inexpensive technology.  The second technology is the Radio  Frequency Vacuum (RFV) kiln dryer. It uses radio waves to heat the wood and a vacuum to remove excess water from lumber. It is a very fast drying process and can dry very large timbers (greater than 200 mm), but it is a capital-intensive technology. The third type of drying technology is the Superheated Steam Vacuum (SSV) kiln dryer. It uses circulating superheated-steam in a vacuum environment to dry lumber. It is intermediate between conventional and RFV kilns both in terms of speed of drying and capital costs.  The first hypothesis of this thesis is: All three technologies are appropriate for drying 114 mm-square hemlock destined for the Japanese market, by considering moisture content distributions and degrade.  4  1.2.2  Post-drying Equalization to Japanese EMC  All three drying technologies have proven themselves successful by being able to remove water from wood. It is unknown whether or not the drying properties of each technology influence the physical properties of the dried wood when it is exposed to Japanese weather conditions. Wood is an anisotropic and orthotropic material; it has different physical and mechanical properties in the three principle material directions. In addition, it is also a hygroscopic material, which means it can easily pick up or lose moisture from the surrounding environment. The moisture content of wood depends on the temperature and the humidity of the air surrounding it. Wood will adsorb or desorb water if the temperature and humidity conditions are constant, until the wood reaches equilibrium determined by the surrounding environment. The equilibrium moisture content (Me) is the moisture content of wood that allows it to be in a dynamic equilibrium with its environment. The following equation links temperature, humidity and equilibrium moisture content: J  14  M„ =  1 8 0 0  W  Kzs]Kis2Ht r 100+ KlK2H  TS TT \ K 2H \00-K2Hj  (1.1)  Where H is the relative humidity in percent, W, Ki and K2 are coefficients derived from the Hailwood-Horrobin adsorption model (Siau, 1984).  Other factors such as  adsorption/desorption hysteresis, temperature during drying, previous drying history, stress, species, and wood extractives also impact the M e (Ahmet et al., 1999; Chaffe, 1991; Skaar, 1988).  Since Me value is affected by the changes of the ambient temperature and humidity it is therefore a seasonally affected value.  For example, in Tokyo, the highest outdoor 5  average Me is 15% and occurs in the summer months. Conversely, in Vancouver, the highest average Me reaches 18% and occurs in the winter months.  The second hypothesis in this thesis is: the difference of winter Me values between Tokyo, Japan and Vancouver, Canada has an impact on the stability of kiln-dried 114 mm-square hemlock for post and beam housing in Japan.  1.3  Overview of Thesis  Chapter 2 reviews the microscopic and ultrastructure anatomy of wood, the wood-water relationships, physical properties of hemlock, and industrial drying practices. Chapter 3 discusses the materials and methods for both the drying and the equalizing components of this thesis. Chapters 4 and 5 introduce and discuss the results from kiln drying and equalizing, respectively.  Chapter 6 makes recommendations for each of three  technologies, where it is expected their strengths are best capitalized for the challenge of drying hemlock for the Japanese market.  6  Chapter 2  Literature Review  2.1  Cellular Structure of Wood  2.1.1  Wood Chemistry  All softwood species are composed of holocellulose (cellulose and hemi-celluloses), lignin, extractives and ash (Siau, 1971). Holocellulose is described as the main building block of wood, lignin as the glue that keeps it all together, and extractives and ash being minor constituents, which impart the different character traits of individual species.  Cellulose accounts for approximately 40 to 50% of the cell wall material. It is a polymer composed of cellobiose, often containing 5000 repeating units. Cellobiose in turn is composed of two P-glucose molecules bonded together. Thus cellulose is a long strand of 10,000 glucose molecules bonded together into a chain. P-Glucose, shown in Figure 1, is a six-carbon sugar arranged in a central flat plane. Attached to each carbon atom are a hydroxyl group and a hydrogen atom. The hydroxyl groups always lie in the equatorial plane and the hydrogen atom always lie in the axial plane. This configuration allows glucose molecules to move in close association and form hydrogen bonds between them. This tight bonding forms a straight chain or crystalline structure (Haygreen and Bowyer, 1989; Siau, 1971).  7  I H  OH  CHgOH  R-frlnmse CH^OH  H  Cellobiose unit  OH  •  1  Figure 1. The molecular structure of cellulose Hemicelluloses are the other major constituent of wood. They account for approximately 20 to 35% of the cell wall material. They are similar to cellulose in nature, but have other sugars in addition to glucose. monomers bound together.  A hemicellulose polymer has between 150 and 200 Monomers (sugars) such as mannose, galactose, and  arabinose do not have all their hydroxyl groups in the equatorial plane, which allows for side chains to be formed. These side chains do not allow for close association with other molecules, and therefore when bound, form non-crystalline structures (Haygreen and Bowyer, 1989).  Lignin is wood's adhesive, which keeps the cellulose and hemicelluloses together. In softwoods, it is composed of guaicyl units, (shown in Figure 2) polymerized in a chain of 50-500 monomers. Due to the structure of guaicyl, with its aromatic "head" and aliphatic "tail" it can form head-to-tail bonds, head-to-head bonds or tail-to-tail bonds, therefore making a compound, which readily bonds to other compounds (Haygreen and Bowyer, 1989).  8  o—e  I  I  I  I  I  I  v—c—c—c-  0CH3  Figure 2. The structure of softwood lignin  Ash is the term given to all inorganic compounds found in wood. They usually act to catalyze reactions, which would occur naturally. Ash is also attributed with dulling saw blades during lumber breakdown operations.  Extractives are described as the "personality chemicals" in wood, giving each species different attributes, depending on the composition of their extractives. These chemicals can contribute to wood's colour, decay resistance, gluing properties, and finishing properties (Panshin and deZeeuw, 1970).  2.1.2  Wood infrastructure  Microfibrils form the backbone of wood's ultrastructure. They are long thin strands composed of a cellulose core, a layer mixed with cellulose and hemicellulose, and an outer layer of lignin. Sheets or layers of microfibril are termed lamellae. It is these lamellae that will form the cell wall. Forty to fifty lamellae are usually required to form a cell wall, but vary from species to species, cell function to cell function as well as the time of year that the cell grew (Walker, 1993).  9  The cell wall is actually comprised of two different structures; the compound middle lamellae, and the secondary wall, Figure 3. The compound middle lamellae is in turn composed of the true middle lamellae and the primary cell wall.  The true middle  lamellae is a lignin rich region which adds integrity to the cell wall. The primary cell wall is a thin layer, which has a high concentration of lignin as well but also has microfibrils organized in a randomized fashion. The primary cell wall is the original cell wall.  The secondary cell wall, comprised of the SI, S2 and S3 layers is usually  considered to be the most important component of the cell wall. The SI layer, or outer layer, is a small region accounting for approximately 5% of the cell wall thickness. It is usually comprised of 4-6 lamellae in depth, which are aligned on an angle of 50-70 degrees from the cells' longitudinal axis. The S2 layer, or the middle layer is the largest and most important component of the cell wall and comprises roughly 85% of the cell wall. The number of lamellae varies from 30-150, depending on whether the cells form in the spring or summer. The S2 layer's lamellae are aligned 10-30 degrees from the longitudinal axis of the cell.  The S2 layer is also very structured due to the high  proportion of cellulose. The S3, or inner layer, comprises roughly one percent of the cell wall, and may be only six lamellae thick. The microfibrils are aligned at 60-90 degrees from the longitudinal axis of the cell. The inner space of the cell, or void is called the lumen (Haygreen and Bowyer, 1989; Panshin and deZeeuw, 1970; Walker, 1993).  10  Warty layer (W) Sj(0-1 (im)  Earlywood (1 - 2jim) Latewood (3 - 5 (im) S, (0-1-0-3)  ~\  _  I •§ f i W  Primary wall (P) (0-1-0-2 Jim) Middle lamella(ML)  Figure 3. The cell wall (Adapted from Walker, 1993) Approximately 95% of softwood cells are longitudinal tracheids. These cells vary from 2.0 to 5.0 mm in length, 30 to 70 /u.m in width and have tapered ends. Longitudinal tracheids are aligned along the longitudinal axis of a tree. Each spring, the cells produced are called earlywood, and have relatively thin cell walls with numerous natural openings called pits. Earlywood serves primarily for conducting water and nutrients throughout the tree, but also acts for mechanical support. Early each summer longitudinal tracheids cells are formed and called latewood. These cells differ from earlywood because they have much thicker cell walls with fewer openings. The primary function of these cells is for mechanical support of the tree. The annual growth of earlywood and latewood each year is called a growth ring. Transition from earlywood to latewood is a character trait that can be used to identify certain species (Walker, 1993). Western hemlock has a gradual transition from earlywood to latewood. The openings in softwood longitudinal tracheids, called bordered-pits, allow for communication between cells. The pits are described as being donut shaped and are composed of four parts, the margo, torus, aperture and border. The margo, a membrane, spans the opening, or aperture and the  11  torus is located in the centre of the membrane. If the torus bonds to one side of the opening, effectively closing the aperture, it is referred as an aspirated pit. This aspiration phenomenon occurs commonly during drying (Haygreen and Bowyer, 1989).  Panshin and deZeeuw (1970) also note the presence of longitudinal parenchyma cells, epithelial cells and ray cells in softwood trees. Longitudinal parenchyma cells are usually 50 to 100 jura in length, 20 to 30 /um in diameter and act as storage vessels for nutrients. They are usually found in single column stacks throughout the tree. Epithelial cells are found in association with canals or voids, which transport water and nutrients in longitudinal and/or radial directions.  Ray cells are conduits that run in the radial  direction and are responsible for moving nutrients radially. Ray cells are also important in the event of damage to the tree. If the normal nutrient conduction pathway is blocked, they can move laterally and then continue in a longitudinal direction once it is away from the damaged area. In softwoods, ray cells will always have ray parenchyma cells present and will sometimes have ray tracheids present. Western hemlock can be identified by the presence of ray tracheids on the margins of most rays. Western hemlock does not usually have any resin canals, but it may produce traumatic resin canals if the tree is damaged in some manner.  A tree grows in height and diameter each year. Growth in diameter occurs through periclinal and anticlinal division of the cambium layer while growth in height occurs in the apical meristem. For the first twenty years of growth, trees produce what is called crown wood. It differs from normal wood by having microfibril angles much greater  12  than 20 degrees in the S2 layer of the secondary wall. Although this angle gives the young tree better protection from swaying in heavy winds, it causes greater shrinkage of wood in the longitudinal direction. The S2 layers larger microfibril angle renders the cell less rigid than normal, protecting the young tree from blow down.  Lumber that is  produced with juvenile wood present will experience longitudinal shrinkage, and if the juvenile wood is not equally distributed in a piece of lumber, the partial longitudinal shrinkage can cause the lumber to warp (Jozsa and Middleton, 1994).  As trees grow in diameter, the actively dividing cells, the cambium layer, is located just under the bark. Next to this layer is a region which is termed Sapwood. These cells are the pathways for nutrients and water moving through the tree. The older parts of the tree no longer serve as a pathway for water but they act as reservoirs for extractives and ash as well as acting as mechanical support of the tree. This portion of the tree is called heartwood (Walker, 1993).  When an older living tree is exposed to strong winds, or if it is growing on a steep incline, the longitudinal tracheids produced change from their usual rectangular form to a shorter, rounder form and are termed reaction wood or compression wood.  The  microfibril angle of the S2 layer in reaction wood is between 40-45 degrees. This causes the tree to be more rigid and less susceptible to blow down (Haygreen and Bowyer, 1989).  13  Specific gravity (G) is essentially a ratio of the density of oven-dry wood to the density of water. Specific gravity can be calculated three different ways. Using the green volume of wood and the ovendry weight is called basic, or green specific gravity (Gb). Using the oven-dry volume and the oven-dry weight us called oven-dry specific gravity (G0) and using the volume at the time of test and the oven-dry weight is called nominal specific gravity (Gn). Since it is calculated by dividing a density by the density of water it is therefore unitless and provides a measure of the amount of wood material in a given volume of wood. Higher G values indicate more potential bonding.sites for water, but less voids for water to easily flow through (Panshin and deZeeuw, 1970). Permeability is a measure of how easily liquids can pass through the wood above the fibre saturation point. Highly permeable woods will usually lose free water quickly as the pit aperture is relatively large compared to slightly permeable woods, which will usually lose free water slowly due to the small size of the pit aperture (Panshin and de Zeeuw, 1970; Siau, 1971).  2.2  Water and wood  Water is a polar molecule. Its oxygen atom has a slight negative charge and its two hydrogen atoms have a slightly positive charge. These slight charges and its triangular shape mean that water molecules are attracted to and bond easily with other molecules with a free negative or positive ion.  Water is naturally found in wood. There are generally two types of water in wood; free water and bound water. A third type of water is water vapour, but it does not contribute to the M in any significant manner. Free water is the water that is contained in the cell  14  lumen, and can act like liquid water. Bound water is the water associated to bonding sites, hydroxyl groups of holocellulose, within the cell wall. It bonds in the spaces between the non-crystalline regions of microfibrils (amorphous areas). Water does not bond to the crystalline structures of the cell wall because the hydrogen bonds keep the chains tight, decreasing potential bonding sites available to water (Wengert, 1988). Drying refers to the removal of both free water and bound water.  When wood is at a constant temperature, it will adsorb or desorb water depending on the humidity of the air, until it reaches Me. The sigmoid curve that relates humidity of the air and equilibrium moisture content at a constant temperature is called a sorption isotherm (Skaar, 1988; Time, 1998). The initial loss of moisture as the humidity is decreased from 100% to 0% creates the initial desorption isotherm. If that lumber is then • gradually exposed to increasing humidity, it creates the adsorption isotherm.  The adsoption  isotherm is always lower than the desorption isotherm. The difference between the two curves is called hystersis.  A second desorption cycle will produce a curve lying  intermediate between the initial desorption and the initial resorption, up to 50% partial humidity. This is a cyclic pattern of continuous desorption and adsorption of water in wood. This phenomenon is due to the fact the Me is determined by past history of the wood, e.g. exposure to high temperatures or different pressures. Exposure of wood to high temperatures will lower the Me of wood by reducing the number of potential bonding sites of water to microfibrils (Haygreen and Bowyer, 1989; Panshin and deZeeuw, 1970; Skaar, 1988; Walker, 1993).  15  Capillary forces and diffusion are the two main driving mechanisms which naturally impact moisture movement in wood. Capillary forces control the removal of. free water and diffusion controls the removal of bound water. Free water is able to move from one longitudinal tracheid cell to an adjacent longitudinal tracheids cell through the bordered pits. Removal of free water from wood follows Jurin's and Darcy's laws (Siau, 1971). When free water is located close to the drying surface, it can be evaporated into the environment. Ideally, this evaporation creates a pressure difference within the wood, and more free water is continuously drawn towards the drying surface to replace the water lost to the environment. If some of the bordered-pits are aspirated, or there are numerous extractives and ash stored in the cell, the flow of water can be interrupted. This reduction in the number of unobstructed bordered-pits decreases the permeability of wood and the removal of water is slowed.  In any given cell, there is a point during drying when all the free water has been removed from the lumen and the only water remaining is the bound water of a saturated cell wall and water vapour in the cell lumen. This point is called the fibre saturation point (Mfsp). A piece of lumber with an average moisture content around 25-30% is considered to be at MfSp. Once the cell is below Mfsp, the second driving force, diffusion, becomes the ratecontrolling step and capillary forces become negligible.  Diffusion is the natural movement of water from a region of high concentration to a region of low concentration. As water is removed from the cells walls close to the drying surface, water diffuses from the centre of the board to the drying face. Not only can  16  diffusion move water through cell walls but also through the lumen by evaporating water at one cell wall surface and condensing it on the opposite side of cell, in the direction of lower water concentration (Siau, 1971).  Mass flow is an additional phenomenon that can move water through wood.  It is  typically found when wood is subjected to temperatures above the boiling point of water, where steam pressure creates the mass flow. It can also be encountered when wood is dried in a vacuum environment. At pressures lower than 1 atm (101 kPa) the boiling point of water is decreased. Mass flow is created by steam pressure while the wood is still below 100 C. The steam pressure created by lowering the boiling point of water in a vacuum can produce very rapid water removal through the longitudinal ends of lumber (Walker, 1993).  The physical and mechanical properties of wood begin to change as the moisture content is reduced below MfSp. Shrinkage of wood only starts when bound water is beginning to be removed from the cell wall.  As water migrates from the spaces between the  microfibrils, the vacated bonding sites then bond with other adjacent microfibrils. The water molecules had been pushing the microfibrils apart, which are now free to return to their close association, causing the wood cell wall to contract or shrink. Shrinkage is different in each of the three principle directions of wood. Shrinkage is greatest in the tangential direction due to its anatomical features. Having a uniform distribution of cells in the tangential direction typically produces uniform shrinkage across a tangential face. Tangential shrinkage is approximately twice as large as the radial shrinkage. The lower  17  shrinkage in the radial direction is usually attributed to the presence of latewood and earlywood and rays in the radial plane, arranged in series. Latewood, having a thicker cell wall than earlywood, has more potential bonding sites for water, so it is subject to more shrinkage than earlywood cells. The presence of earlywood cells in every growth ring contributes to the radial face having less shrinkage than the tangential face. Shrinkage in the longitudinal direction is small due to the bonding nature of water, which pushes microfibrils apart, but does not lengthen its overall structure significantly. However, the high microfibril angle found in juvenile wood allows for shrinkage in all three planes and longitudinal shrinkage is detected, along with tangential and radial shrinkage (Haygreen and Bowyer, 1989; Walker, 1993).  Below MfSp, water is removed from the spaces between the microfibrils, allowing more and more microfibrils to return to the close associations. Hydrogen bonds can form between the adjacent microfibrils, which in turn strengthens the whole structure. This increase in strength of microfibrils is also realized in a piece of lumber. Additionally, by removing excess water from wood, to levels below Mfsp, one of three conditions necessary for decay fungi to grow is eliminated, and decay will be limited unless water, oxygen and heat are all reintroduced to the wood.  Drying of lumber  There are three phases of drying lumber: the constant rate period, the first falling rate period and the second falling rate period (Figure 4). The constant rate period is found at  18  the beginning of the drying cycle. It involves free water movement because at this stage the shell and core regions of the lumber are above the fibre saturation point. Capillary force is the rate-determining factor and the drying rate is constant. The next phase of drying is called the first falling rate period. At this stage in drying, the shell is below Mfsp but the core is still wet. In the shell region, diffusion takes over from capillary force as the rate-determining step while capillary forces are still present in the core region. It is during this phase of drying where drying stresses become a concern as the shell is starting to shrink while the wet core resists shrinking. Due to the presence of two factors driving drying, with diffusion being slower than capillary flow, this phase is characterized by an increase in the drying rate. The last phase of drying is called the second falling rate period. In this phase, both the shell and the core are below Mfsp and diffusion is the ratedetermining factor in the full cross-section thereby creating a lower drying rate than the first falling rate period. It is during the second falling rate period where the greatest shrinkage takes place resulting in large internal stresses in the lumber (Simpson, 1999).  Constant Rate Period \i  1st Falling Rate Period ^^^-^_^  2nd Falling Rate Period  Time (hrs)  Figure 4. The three phases of the drying curve  19  Kiln drying lumber is an endothermic process. Energy, in the form of heat is supplied to the lumber to speed the natural process along. The warm air used to heat the wood also absorbs the water on the wood surface, creating the pressure and concentration gradients necessary to drive more water from the wood; breaking the hydrogen bonds between the wood and water molecules in the process. However, at high temperatures, the lignin in the cell walls begin to break down.  Lignin is a thermoplastic compound; at high  temperatures it loses its ability to hold the cell structures rigid, resulting in the wood becoming pliable itself (Mitchell and Bigbee, 1989).  When high temperature and unequal shrinkage are encountered in wood, the impact can be severe as well as undesirable. Unequal shrinkage can occur if a board has both tangential and radial faces, or if juvenile wood is present. If a board is cut with one surface being primarily tangential while the other face is primarily radial, the resulting unequal shrinkage can cause the board to warp. If these differences are located in the transverse section, the unequal shrinkage, in conjunction with high temperatures, can cause the board to cup (Figure 5). When cooled, the lignin stiffens and the board becomes rigid in this cupped shape. If there are differences in the distribution of planes along the length of a board, differential shrinkage, coupled with high temperatures, can cause warp along the longitudinal axis, producing either crook or bow, depending on which direction the board moves. Crook denotes a deflection in thickness and bow denotes a deflection in width (Figure 5). Another particularly undesirable form of warp, especially in the Japanese market, is called twist. It results when not all four corners of a board are in the same plane. It is caused by the irregular growth of trees, causing  20  differential shrinkage in the board produced from that tree (Figure 5). In large dimension lumber, such as those destined for the Japanese market, diamonding is another undesirable form of distortion. Large timbers have the potential for mixed grains. Being neither completely flat grained or edge grained causes unequal shrinkage and diamonding results (Figure 5). Using weight restraint during drying will help alleviate warp during drying by not allowing the timber to move when the lignin becomes pliable. The weight restraint must remain on the wood until the timber cools and the lignin has once again become rigid (Mackay and Oliveira, 1989).  0val  Diamond  Cup  Figure 5. Warping of Timbers (adapted from Simpson, 1999)  21  Another undesirable effect of shrinkage is rupture of wood. These ruptures or splits are caused when the stresses that develop during drying exceed the strength of the wood. Fractures can occur either on the surface or in the centre of the board depending on the internal stress state. These stresses are caused by the fact that wood dries from the surface of the board inwards towards the core. This means that the surface region will reach MfSp before the core region. As the shell starts to shrink, experiencing tension forces, constrained by the still swollen core, which is now experiencing compression forces from the shrinking shell. As the drying boundary reaches further into the core, the core starts to shrink and tries to pull the shell with it, reversing the previous internal forces. Either the tension force or the compression force can be stronger than the strength of the wood, resulting in splits of the wood. When wood splits across the grain, it is called a check. If the split is exposed to the surface it is a surface check, otherwise is an internal check, sometimes referred to as honeycomb. When there is a separation of the annual growth ring, the split is called shake (Mackay and Oliveira, 1989).  Casehardening; a condition that results unless these internal shrinkage forces are relieved at the end of the drying process results in further movement of wood when it is cut longitudinally. During secondary manufacturing, a board may be ripped, turning the core into an exposed surface. The core and the shell will try to relieve the residual drying stresses and cause further warp in the board. Casehardening can usually be relieved after drying with a conditioning step. Conditioning uses low-pressure, saturated steam as a means of introducing moisture to the surface of the lumber in an effort to relieve the residual stresses (Walker, 1993).  22  2.3  Industrial Drying  The aim of industrial drying is to accelerate the natural drying process, to take advantage of dry wood's superior attributes, while minimizing some of the negative impacts that can be associated with drying. Moisture content, a ratio of the weight of water to the ovendry weight of wood, expressed as a percent, is a classical measure of the amount of moisture is in wood. For dimension lumber products, lumber is considered 'kiln-dry' when the moisture content is 19% or less. It is recommend that wood destined for secondary manufacturing be dried to 6-12% moisture content, or as close as possible to the Me of the service area (Simpson, 1998; Walker, 1993; Skaar, 1988). To meet North American dimension lumber grading standards the lumber must have a moisture content of 19% or less, but in order to minimize the amount of degrade induced by drying, the lumber should also have a tight distribution around the target final moisture content. If the distribution is large, there will be some lumber, which has been over-dried, while others still remain wet.  The over-dried lumber may have splits or warp, making it a less  valuable product. The wet lumber will be dressed to the standard finished size, and then will shrink to a smaller size less than the standard size when it equalizes with the environment. Either situation means lost potential earnings from over-dried lumber or an undersize product which may impact product performance in service.  Industrial drying is usually achieved by controlling the temperature and humidity of the air circulating over the exposed surface of wood to continuously reduce the Me in the kiln. Large kilns hold lumber while it is being dried. With the notable exception of  23  Radio Frequency Vacuum kilns, lumber is stacked in layers, separated by "stickers" to facilitate the warm air circulating between adjacent layers. Stickers can vary in width and height depending on the dimensions of the lumber being produced. When drying 114 mm by 114 mm hemlock, stickers 13 mm in height and at least 64 mm wide are employed. The air passing through these voids, pushed by several large fans, causes a turbulent flow to ensure good interaction between the air and the drying face of the lumber. In addition to controlling heat, the; humidity in the kiln is also controlled. In the early stages of drying, the temperatures are kept relatively low and the humidity relatively high (high Me) to prevent the surface of the lumber from drying too quickly. Once the lumber is heated uniformly, the temperature will be successively raised and the humidity will be successively dropped to maintain drying (lower and lower Me). The air circulating through the lumber picks up moisture from the surface of the wood as well as heating it. This saturated air must either be vented to the outside, in exchange for cooler drier air, or be condensed and the water removed from the kiln. Conventional drying is an energy intensive process because the warm moist air is vented and replaced with cooler, drier air, which must be reheated.  Drying will continue by adjusting the  temperature and humidity of the air until the desired moisture content is achieved (Smith, 1984).  In order to meet kiln-dried lumber specifications, dimension lumber must be dried to a moisture content of 19% or less. However, most end-users would argue that lumber must be dried to moisture contents that it will experience where it is installed (Milota and Wengert 1995; Simpson, 1999). In doing so, movement of wood as it equalizes from  24  delivery Me to its end-use Me will be decreased and issues such as floor and wall buckling and dry wall staining will be reduced. Kiln operator experience and hot-checks towards the end of the drying are used to guess the moisture content during drying. Using a hand-held resistance type moisture meter, a kiln operator will take approximately 200 measurements and when the lumber meets the final moisture content, with an acceptable standard deviation, they will stop drying (Smith, 2000).  Kiln-drying hemlock has several inherent difficulties including brown stain, wet-pockets and sinker heartwood. Brown stain is produced in the longitudinal tracheids of sapwood and is particularly prevalent after kiln drying.  It is characterized by a brown  discolouration that is often superficial but can also occur at varying depths beneath the surface of the wood (Ellis and Avramidis, 1993). Brown stain is caused by an enzyme catalyzed oxidative reaction of wood extractives followed by condensation of the reaction products upon drying. The main chemical thought to be involved is catechin [2-(3,4dihydroxyphenol)-3,4-dihydro-2H-l-benzopyran-3,5,7-triol].  The concentration  of  catechin varies according to when the tree was felled. Catechin concentration is highest between April and May, and these high concentrations often lead to high levels of staining. An increase in the severity of the kiln schedule also caused an increase in the severity of staining, (Ellis and Avramidis, 1993). There are chemical treatments that may be applied to the wood before drying which have had some successes. The downside to this type of solution is the introduction of yet another topical chemical being applied to lumber. These chemicals work as either an anti-oxidant or act to lower the pH of wood and make the reaction slower (Avramidis et al. 1993). Avramidis also concludes that  25  pre-steaming the lumber at 100°C for 8 hours can help to alleviate the occurrence of brown stain.  The presence of wet-pockets in heartwood is another problem associated with kiln-drying hemlock. Wet-pockets are described as regions of kiln-dried wood that remain above 19%, even after kiln drying. In Western hemlock, wet-pockets are caused by extractives closing the pit membrane. This blockage disrupts the removal of the free water from the cell by capillary forces (Lin and Kozlik, 1971). The high concentration of extractives in the cell may also decrease the diffusion rate, leaving a water-laden region of wood after drying. If the pockets are small and infrequent the water contained in these pockets can equalize through the piece of wood without too many problems, but if the pockets are large and frequent, quality issues due to retarded shrinkage may develop.  The presence of sinker heartwood in hemlock also creates drying problems.  Sinker  heartwood is much more dense than normal heartwood, has higher moisture content and takes longer to dry than normal heartwood. The presence of sinker heartwood in a kiln charge can cause the distribution of the final moisture content around the average to be large, thereby increasing the amount of potential degrade induced during drying (Kozlik and Ward, 1981).  2.3.1  Conventional Kilns  The most common type of dry kiln in BC is the conventional heat-and-vent kiln. Conventional drying builds on the water removal systems of natural drying, but does so  26  in the closed environment of a kiln, where the temperature, humidity and airflow are all controlled to increase the speed of drying lumber, while trying to minimize degrade (Mitchell and Bigbee, 1989). Heat is usually supplied in one of three ways: direct gasfired heating, electric heating or thermal oil heating. Gas-fired kilns most commonly combust natural gas to heat the air which circulates between layers of wood. Electrically heated kilns use electricity to heat the air circulating in the kiln. Thermal oil heaters circulate thermal oil through finned pipes to heat the circulating air. The oil is usually heated with either a natural gas fired burner or a wood residue burner.  The physical plant can be manufactured out of several materials, with aluminium and stainless steel reported to be the most durable, but other materials such as concrete, blocks, mild steel, timber and clay blocks have been utilized.  Through mauipulation of dry-bulb temperature, wet-bulb temperature and airflow, kiln operators have been able to adjust how quickly, or how slowly wood dries and can therefore influence the amount of defect that may occur. The extent to which the kiln operators accelerate drying depends on the ultimate end-use of the wood. If the lumber is going to be used for structural purposes, the kiln operator usually tries to maximize the throughput and therefore, dries the lumber quite quickly without too much concern for final quality. Most industrial operations drying hemlock target an average final moisture content (Mf) of 19%, with a standard deviation less than 5%; when drying for structural lumber end-uses. Whereas, if the wood is being dried for remanufacturing, it is usually dried more slowly, with the emphasis placed on quality, not quantity, and will dry the  27  lumber down to a Mf range of 7 to 10% (Oliveira and Mackay, 1989). Oliveira (1998) reports that the average drying time for 105 mm by 105 mm Pacific Coast Hemlock (PCH), a mix of Western hemlock and Amabilis fir, is 11 days, which includes a short equalizing period. Most drying operations use air velocities between 2 and 3.5 m/sec but some operations are now using velocities above 4.5 m/sec. Some kilns are equipped with variable frequency drives for the fan motors. These drives, which allow the velocity to vary during drying in order to realize potential energy savings by slowing fan speed during start-up and when water removal is governed by diffusion. Most operations use o  maximum temperatures around 85 C.  General advancements in conventional drying have been numerous and well documented because it is the oldest, and most common drying method. Small scale studies have reported that stickers which are less than 63.5 mm in width can fall onto the load in the wrong way and cause serious degrade during drying as the load is not sitting properly (Oliveira, 1998). While more rigorous studies have involved the introduction of vertical air gaps, in addition to the horizontal gaps created by horizontal stickering. Li et al. (1997) reported decreased drying time using vertical air gaps but there was no associated decrease in shell and core moisture content differences, shrinkage or final quality of the 105 mm by 105 mm PCH, so this innovation has not been implemented because the advantages of faster drying are not greater than the loss of kiln volume.  Dedrick and Ziegler (1984) report a 10% reduction of degrade due to warping when weight restraint is utilized. They also report that as the Mf is reduced from a target of  28  12% to 9%, the amount of degrade due to warping increases substantially. Mackay (1995) reports that weight restraint at 220 kg/m2 showed the most promising results. It is worth mentioning that the decision to use weight restraint depends on the end-use of the lumber. If drying for high quality is required, for example, a remanufacturable product, it might be possible to sacrifice decreased kiln volume for the increase in quality when using weight restraint. Kiln specific cost benefit analysis should be completed to see if the increased quality could outweigh the decreased kiln volume.  Research has also been conducted to determine if presteaming has any benefit to the overall final quality of the dried lumber. Avramidis and Oliveira (1993) report that although the final moisture content distribution was more uniform, and> there was no change in the incidence of defects, the increase in kiln residency time and increase in volumetric shrinkage did not warrant pre-steaming, unless for the additional purpose of reducing brown stain. Cost benefit analysis should be completed to determine if the increased costs could be recovered by the decrease in downgrade.  Another advancement in conventional drying has been to add a conditioning step at the end of drying. Avramidis and Mackay (1988) report that conditioning 105 mm by 105 mm PCH for 24 hours at a dry-bulb temperature of 65.5 C and a wet-bulb temperature of 63.3 C will relieve casehardening stresses and close surface and end-checks. Casehardening is a condition where the shell of the board is in tension and the core is in compression. This condition is produced by steep moisture content gradients during drying around the fibre saturation point. Conditioning the wood at this high humidity  29  allows the shell of the lumber to pick up water and relieve some of the internal stresses. As was previously mentioned, alleviating casehardening is particularly important when the lumber will be remanufactured.  One of the remaining areas where a large amount of research is being conducted is sorting PCH based on the basic density prior to drying. Milota and Wengert (1993) argues for sorting when drying multiple species in one kiln charge, as is the case for PCH. It was observed that in order to meet the final target moisture content of multispecies loads, timbers in the load with lower initial moisture contents will be over-dried and subject to the associated degrade. Zhang et al. (1994) found that there was little benefit to sorting based on species separation, with regards to improving uniformity of the final moisture content, and that sorting based on basic density must be conducted before there is any improvement in the distribution of the final moisture contents. Sorting based on basic density also produced a shorter drying time for the low basic density kiln charges, and a decrease in the volumetric shrinkage. There was no reported improvement of drying time for high basic density lumber.  Warren and Johnson (1997), Milota et al. (1993), Jamroz (1997), Garrahan (1992) and Zhang et al. (1994) have all investigated the impact of moisture sorting lumber prior to drying. Uniform lumber entering the kiln should maintain its uniformity unless there are temperature and humidity problems with the kiln. Lumber of the same species will have relatively the same drying rate, such that if one board enters the kiln at 70% and another at 55%, in all probability they will not have the same moisture content after drying.  30  However, if lumber is sorted according to moisture content, groups of lumber which enter the kiln within a specified range, will probably exit the kiln within that specified range. Most of these studies have been carried out as a means to decrease the amount of overdrying as well as decreasing energy consumption by shortening drying time.  The  challenge with sorting based on initial moisture content is confidently estimating initial moisture content. Jamroz (1997) developed an infrared laser detection system. Northern Milltech Inc., (NMI) developed a moisture content index system by measuring the weight and capacitance of timbers.  2.3.2  Superheated Steam Vacuum Kilns  Superheated Steam Vacuum (SSV) drying uses a low pressure environment, 200 kPa, to lower the boiling point of water to 55 C. Superheated steam surrounds the wood to transfers its heat to the wood and water.  The Superheated Steam Vacuum kiln,  manufactured by IWT Denmark, uses air, pushed by 11 fans, to circulate at 16 m/s, through a stickered load. In SSV drying, the main transport phenomenon is mass flow, produced by the boiling of the water in the wood. Oliveira et al. (1998) reported decreases in drying times of thick dimension PCH of up to 60%, decreasing drying time from 16 days to 6 days. They were also able to reduce the shell to core moisture content variation, and the incidence of warping and checking. Pang and Dakin (1999) have conducted experiments to test the optimum circulation rate. In vacuum kilns, the density heat capacity of steam is decreased. In order to compensate for lowered heat transfer,they recommend that the circulation speeds of superheated steam be greater than 10 m/s. Removal of water from the vacuum chamber is usually done through condensing plates  31  or condensing chambers. SSV kiln control can be based one of three methods, time, moisture content or time and moisture content. Time-based kiln schedules are time increments; certain conditions (temperature and humidity) will be applied in the kiln for specified periods of time. Moisture content schedules are based on certain conditions (temperature and humidity) at specified moisture contents. Moisture content can be estimated by one of two methods. Insulated stainless steel pins can be inserted into the lumber, and an applied current between the pins can be measured and related to moisture content. Alternatively, the volume of water removed can be used to estimate the volume of water remaining in the wood, assuming the weight, specific gravity, and initial moisture content are known.  2.3.3  Radio Frequency Vacuum Kilns  Since the dielectric constant for water is much larger than it is for dry cell wall material, radio frequency heating preferentially heats the water at a much more rapid rate than it does the cell wall material. This is opposite to the heat transfer process of conventional kiln drying and superheated steam vacuum kiln drying. Radio Frequency Vacuum (RFV) drying produces moisture flow from the centre of each board to its ends. The water then evaporates into the low pressure, approximately 2.7 kPa environment. For this reason, stickers and airflow are not required for this drying technology.  Past (United States) and present (Japan) RFV kiln manufacturers use radio-frequency oscillators to produce the radio wave due to their perceived simplicity. The oscillators use vacuum tubes in a tuned plate-tuned grid configuration. This configuration results in  32  the wood being part of the electrical circuit. Some of the past problems with RFV kiln driers have been due to the fact that as wood dries, its capacitance changes and unless this is accounted for in the radio electric circuit tuning, serious inefficiencies can result. Additionally, with an RF oscillator the radio waves inside the chamber are generated in an uncontrolled fashion.  Avramidis and Zwick (1996) have attributed some of the  documented problems with RFV drying to the RF oscillator technology.  HeatWave  Drying Systems Ltd. (Heatwave) uses a 50 ohm RF amplifier to magnify a low power output from a frequency stabilized oscillator.  An amplitude modulator controls the  voltage that is applied to the load and an active matching network compensates for the changing capacitance of the wood.  Heatwave states that this chain-component  configuration allows for easy diagnosis and repair if one of the components fails. With either approach in RF generation, the overall result is water being driven from the wood very rapidly in both a gas and liquid phase (Avramidis et al. 1994).  Avramidis et al. (1994) report very favourable results in terms of degrade.  The  difference between shell and core moisture contents did not vary by more than 1.5 percentage points, and the variation along the length of the specimens did not vary from the final target moisture content by more than 1.5 percentage points. The width and thickness shrinkage values were smaller than those of conventional drying, and also that in 50% of the timbers, the core was drier than the shell, and no longitudinal or transverse casehardening stresses were present. Avramidis et al. (1994) also report drying times that are ten times faster than conventional drying. RFV drying shows great potential and optimized drying schedules are being investigated in order to minimize drying time and  33  degrade. Avramidis and Zwick (1996) report that during full-scale RFV drying runs, PCH was dried successfully with no lumber staining or internal stresses, reduced surface checking and good final moisture content distribution. With proper schedules, RFV drying also exhibited less shrinkage than conventional drying. Some modifications to the electrodes were deemed necessary to increase the uniformity of the drying and when the modifications were made, the increase in uniformity was appreciable. RFV drying also produced timbers that had a core moisture content which was lower than the corresponding shell moisture content, indicating uniform heating throughout the piece of wood.  Heatwave has developed an in-kiln weight restraint system that applies 112 kN (144 lb/ft2) of pressure to the lumber during drying.  This pressure prevents wood from  moving due to differential shrinkage and helps decrease the twist, bow and crook.  Heatwave has also developed a protocol called Q-sift (Zwick, 1999). This process involves drying hemlock in a conventional kiln until the average moisture content is approximately between 22 and 26%. Before planing, the wet timbers, detected by an inline moisture content meter, are selected to be redded in an RFV kiln. This process takes advantage of conventional kiln drying's high volume capability to remove free water and RFV kilns to quickly remove the remaining bound water from hard to dry, or slow to dry timbers without invoking warp.  34  2.5  Physical Properties of Western hemlock  Western hemlock accounts for roughly 1.1 billion m3 of standing tree volume and represents 15.7% of the growing stock in BC (COFI, 1998). This shade tolerant species grows primarily in the coastal biogeoclimatic zone and is predominantly found in shady areas with well-drained soils. It grows to heights of 35 to 55 meters with breast height diameters ranging from 900 to 1200 mm. Its sweeping apical tip and small cones at the end of its branches characterize the standing tree. It is usually found in stands with Amabilis fir, Sitka spruce, Western redcedar and Douglas-fir (Gonzalez, 1995).  Once sawn into lumber, it is described as a light coloured wood with a pinkish-toreddish-brown tinge, straight and even grained, with a medium-to-fine texture. There is little difference in colour between sapwood and heartwood and the annual rings are distinct (Mullins and McKnight, 1981).  Nielson et al. (1985) report the average specific gravity (Gb) as 0.423, placing it between Douglas-fir and the spruce, pine and fir mix.  The green moisture content varies  substantially between sapwood and heartwood, which are reported as 143% and 55%, respectively. Its volumetric shrinkage is reported as 12.8%.  When dry, hemlock is considered a relatively strong species. Haygreen and Bowyer (1989) report its modulus of elasticity in bending (MOE) as 11,238 MPa. MOE is an index of the stiffness of a beam and is measured through resistance to bending deformation. Hemlock has an average modulus of rupture (MOR) of 77.9 MPa. The  35  MOR is the maximum stress that a beam will carry at failure. Compression strength parallel to the grain is 50 MPa. The compression strength perpendicular to the grain is 3.8 MPa. Compression strength perpendicular to the grain is an important mechanical property in connection design.  Mullins and McKnight (1981) describe the durability of hemlock as slightly resistant and recommend treatment with a wood preservative. Chromated-copper-arsenate (CCA) is the most widely used chemical, but with Japan moving to ban CCA due to its heavy metal content, new studies are being conducted with borate treatments.  Typical end-uses of hemlock include: framing lumber, joinery, millwork, and window, door or cabinetry components. It is also widely used in the Japanese post and beam house building. Japan imported 0.6 million m3 in 1997, or 21% of BC's wood products production for their housing industry (COFI, 1998).  Presently, the only kiln-dried  products manufactured for this market are posts, joists, studs, lateral bracings and roof rafters.  36  Chapter 3 3.1  General Methods and Materials  Materials  The Western hemlock {Tsuga heterophylla [Raf.][Sarg.y) for this research was harvested from the Queen Charlotte Islands, British Columbia.  This region forms part of the  coastal biogeoclimatic zone on North America's west coast. The logs were shipped to Port Alberni, BC, by water, where they were processed into 8 meter posts measuring 114 mm by 114 mm by Weyerhaeuser Co. These posts were then wrapped for shipment to Vancouver and the Western Research Lab of Forintek Canada Corp.  The one hundred and fifty 8-meter long posts were cut into three 2.43 metre timbers, and separated into three groups. Matched samples were produced by designating Group A for the conventional kiln, Group B for the SSV kiln and Group C for the RFV kiln. Disks of size 25 mm were cut at the 305 mm, 2743 mm, 5.2 m and 7.6 m marks (shown in Figure 6). The initial moisture content (M;) and specific gravity (Gg) were calculated using the oven-dry method, using Equations 3.1 and 3.2, (Simpson ,1999) fur M(%)  -W mc  \ TOO  ovendry  w ovendry  W Gb=-f±Zgreen  I PHlP  (3.1)  J  (3-2)  /  37  MC Sample  Waste  MC Sample  Conventional  MC Sample  SSV  MC Sample  RFV  Waste  Figure 6. Cutting pattern of green lumber  Each group was sorted into two sub-groups based on the median density of the population. Matched samples were verified to ensure that natural variation of moisture content and basic density along the longitudinal axis did not affect the sampling. Five samples were discarded due to variation along the longitudinal axis. The samples were stored outside under lumber wrap and tarp to reduce any natural water loss.  Green weight, dimensions, warp and visible defects were measured prior to drying. All weight measurements were collected using an ANG digital scale with an accuracy of 0.02 kg. All dimensional measurements were taken as width and thickness measurements, and collected with Mitutoyo digital calipers, accurate to 0.025 mm. Grain orientation was randomized.  Green dimensions were recorded at 610 mm and 1830 mm.  Warp  measurements were collected by measuring deflection from a straight-edge using a tapered gauge accurate to 0.2 mm. Green defects were marked with a black lumber crayon and the lengths were recorded.  Statistical analysis was conducted using NCSS 2000 software package.  38  3.2  Phase I: Drying  3.2.1  Conventional drying:  The conventional kiln used in this research was a 5.5 m3 electric heat kiln equipped with a low pressure steam spray. Air velocity was constant at 4 m/s. The low density samples and high density samples were dried separately using the same schedule. Stickers of cross section 50 mm by 19 mm stickers were used to separate the drying layers. Drying was terminated based on weight, measured by a load cell in the kiln. Table 1. Conventional kiln schedule Phase  3.2.2  Heat-up  Time (Hrs) 6  Ramp (Hrs) 0  Temperature (C) 49  Drying 1  2  2  Drying 2  22  Drying 3  Humidity ' M e (%)  (%)  100  25.5  54  92  18.9  22  60  87  16.0  72  72  65.5  76  11.9  Drying 4  41  29  71  77  11.7  Drying 5  31  0  71  65  9.0  Drying 6  180  0  71  60  7.7  Conditioning  12  0  60  87  16.0  Superheated Steam Vacuum Drying:  The SSV kiln used in this research was a 1 m3 thermal oil kiln manufactured by IWT of Denmark. The operating pressure of this kiln varied from 200 mbar to 500 mbar. The 39  velocity of the steam in the kiln was 16 m/sec. Two runs were required for each density class, requiring four runs for the experiment employing the schedule in Table 2. Stickers measuring 6.5 mm x 6.5 mm stickers were used to separate the drying layers. Drying was terminated on time.  Table 2. Superheated Steam Vacuum kiln schedule Phase  3.3.3  Temperature (C) 60  Humidity  Me  (%)  (%)  Heat-up  Time (Hrs) 8  95  20.5  Drying 1  112  60  90  17.5  Drying 2  48  71  80  12.4  Drying 3  24  75  61  8.0  Drying 4  24  80  49  6.0  Conditioning  8  71  90  17.0  Radio Frequency Vacuum Drying:  The RFV kiln used in this research was a 30 m kiln manufactured by HeatWave Drying System Ltd., located in Cresent Valley, BC. The radiowaves are produced by a 50 ohm RF generator and operates between 4 to 8 kPa of pressure. No stickers are used with this technology. The power density used was 5 kW/m3. The lumber was dried mixed with other Hem-fir lumber to fill the kiln. Three runs were required to dry all the lumber.  40  3.3.4  Kiln Dry Measurements  After drying the following measurements were collected: weight, dimensions, warp, defects and final moisture content (Mf). Moisture content data was collected using a Delmhorst RDM-2S pin-type moisture meter, calibrated for temperature and species with insulated pins. Measurements were collected at 610 mm and 1830 mm locations. The shell measurements were collected at a depth of 25 mm and the core measurements were collected at a depth of 52 mm. Four 25 mm disks were cut 305 mm from the numbered end of each specimen. One disk was used to measure the moisture content of the whole cross section (Mf), one disk was further cut into shell and core components to measure the shell and core moisture contents (Ms and Mc), and two disks were cut to produce casehardening samples parallel and perpendicular to the drying face. Casehardening was quantified using a ratio of the deflection of two prongs to the length of the prong, according to Figure 7.  Prong position=  a +b  Figure 7. Casehardening measurements  41  After the kiln-dried timbers had been sampled they were planed to 105 mm to 105 mm using a single face planer. The timbers were end-sealed, stacked in solid piles, packaged in lumber wrap and stored outside until all samples were ready for the humidity chamber.  3.3  Phase II: Equalizing to Winter Japanese Equilibrium Moisture Content  Once all the timbers had been dried and sampled, tests were run to select timbers for the humidity chamber. The timbers had to meet two criteria: the moisture content had to be within 5% of the target 19% moisture content and had to have similar moisture contents between all three technologies.  Thirty-six timbers from each density class were  statistically selected for the humidity chamber.  Prior to equalizing, instrumentation was conducted by inserting two insulated moisture content pins into the wood 1016 mm from the numbered end. The shell pins were inserted to a depth of 25 mm and the core pins were inserted to a depth of 52 mm. Measurements were taken by touching the end of the pins to a modified Delmhorst plunger and collecting the readings. Width and thickness dimensions were taken 890 mm from the numbered end. Defects were marked and dated with a coloured crayon and measured for length. Warp was measured using a warp table constructed from medium density fibreboard using a tapered gauge for measuring. Weight was measured using an ANG digital scale. The lumber was stickered on 13 mm by 25 mm plywood stickers which had been equalizing in the humidity chamber. The stickered lumber was placed on carts so they could be easily removed from the humidity chamber.  42  The humidity chamber was a 14 m3 room equipped with a heating, ventilation, airconditioning (HVAC) system and mobile dehumidifier. The temperature varied from a maximum of 20°C to a minimum of 8°C over the course of the study with an average temperature of 8.8°C. The relative humidity was varied from a maximum of 80% to a minimum of 62% over the course of the study, with an average humidity of 71%. Weight, dimensions, moisture contents and defects were recorded weekly until the wood equalized to approximately 12% moisture content. Position of the density sorted groups was randomized weekly.  43  Chapter 4 4.1  Drying Analysis and Discussion  Density Sorting  The results from the density sort produced two statistically different density classes. The low-density samples (LBD) in all three matched sample groups had a basic density average of 389 kg/m3 with a standard deviation of 19 kg/m3. The high-density samples (HBD) in all three-matched sample groups had a basic density of 455 kg/m3 with a standard deviation of 24 kg/m .  4.2  Conventional kiln drying  Drying for both the LBD and HBD lumber was controlled by a time based schedule, and a load cell in the kiln controlled shutdown. The LBD lumber dried to the final moisture content target in 340 hours, including a 12-hour conditioning period. The HBD lumber dried to the final moisture content target in 354 hours, including a 12-hour conditioning period. The moisture content results are located in Table 3. Even though the initial moisture content of the HBD lumber was lower than the LBD group, it took longer to dry to the target moisture content. This indicates that although there are more potential bonding sites in the HBD lumber group, the void space in the cell lumen is smaller and therefore cannot hold as much free water as the LBD group. As the LBD groups are more permeable, they were able to lose most of their free water quickly, by diffusion. Below Mfsp, the LBD group had less bound water and was able to reach the target moisture content 14 hours faster than the HBD group. Even with a 12-hour conditioning step at the end of drying, there was a significant difference in Ms and Mc. This gradient  44  was the same for both the LBD and the HBD groups. Although the 6% moisture content gradient may seem large, as the core had an average (Mc) of 19%, when the moisture equalized throughout the board, the moisture content would meet lumber grading requirements.  Table 3. Moisture content results from conventional drying Drying Time (Hrs) 340 LBD 354 HBD  Mi (%)  Mf(%)  Mc(%)  Ms(%)  AM  [St. Dev.]  [St. Dev.]  [St. Dev.]  [St. Dev.]  (Mc-Ms)  75.3  16.4  19.5  13.6  5.9  [26.1]  [2.3]  [3.6]  [1.3]  59.5  15.5  19.2  12.8  [18.1]  [1.8]  [4.4]  [2.1]  6.4  Since the HBD lumber groups have more potential water bonding sites, it would have been expected that the HBD lumber group would have experienced more shrinkage than the LBD group. For the conventionally dried lumber, these results are inconclusive (Table 4). Averages between the two regions measured, shows the LBD group had greater shrinkage in width than the HBD group, while the LBD groups experienced less shrinkage in thickness than the HBD group.  Figure 9 shows the initial and final moisture content distributions for the conventional samples. From these graphs the effect of sorting green lumber on specific gravity is  45  observed. Two statistically different initial moisture content distributions become two statistically similar distributions after drying.  3 A LBD M i d A LBD Mf  50 45 40 35 30 25 20 15 10 5 0  n in CM  t  pji o  I- d o  1^  EAHBDMi OAHBDMf  ID + o  50 45 40 35 30 25 20 15 10 5 0  B_  [IQQC in +  M o  oo  o o  MC Class MC Class  (8a.) Conventional LBD M; and Mf (8b.) Conventional HBD Mj and Mf distributions distributions Figure 8. Initial and final moisture content distribution of conventional samples  Table 4. Shrinkage results from conventional drying  LBD  HBD  Width (% shrinkage)  Thickness (% shrinkage)  [St. Dev.]  [St. Dev.]  2.95  2.45  [1.1]  [1.1]  2.80  2.95  [0.8]  [1.1]  Warp results (twist, crook and bow) are presented in Figure 9. These graphs show the negative issues found with kiln-drying lumber. In the green condition, there is no twist in 46  either of the LBD and HBD groups, however after drying, there is twist present in almost half of the dried timbers. Japanese pre-cutting facilities can tolerate up to 3 mm of warp in a 2 m post, represented by the dashed line in Figure 9, so most of these timbers would be considered acceptable (Doucette, 2000). Crook is present in most of the timbers in both density groups. The green crook is usually associated with sweep of the log or poor sawing control during manufacture. In the dry condition, the high incidence of crook in the LBD and HBD groups may be attributable to the large amount of shrinkage in width; the same plane where crook is measured. The degree of crook in the HBD timbers would meet most grading rules with the exception of 10 pieces. Also, the incidence of bow in the LBD group can be considered high, although the severity is not great enough to seriously affect the grade. The HBD lumber showed favourable results with very few timbers having bow, with the one notable exception where the bow was 10 mm.  Casehardening results are presented in Table 5. HBD lumber should experience greater shrinkage than LBD and this should result in more residual stress since the drying schedule and time are relatively similar.  The LBD lumber not only has smaller  casehardening values, but they also have a tighter distribution around the mean.  Drying curve and drying rate results are located in Figure 10. The drying curves show both the raw data and a third degree polynomial fit to the curve. Differentiating the polynomial equation and conducting a piecewise linear-linear regression results in the drying rate graphs. From this graph, it is possible to estimate the moisture content range where the rate-limiting step changes from capillary flow to diffusion. In the LBD group,  47  this change takes place at approximately 35% Mf and approximately 30% Mf for the HBD group. It is also observed from the differentiated equation, that the LBD group dried faster than the HBD groups  48  0 Green Twist O Dry Twist  S Green Twis O Dry Twist  s—|  64  64  55  55  F  % 46  1  37  *  28  !_,! I I J  46  E|  19  <S 28 19  10  10  ins^>  P«i.  1  1 8  4  0  4  8  4  12  0  4  Twist (mm)  Twist (mm)  (9a) LBD Twist  (9b) HBD Twist  Q  Green Crook O Dry Crook  Q Green Crook O Dry Crook  64  64  55  55  *  46  f  37  f  &  28  <5S 2 8  a  46  3 7  FF—  ""• * mi c:—  19  "T-S&*  10  10  1  1 4 - 0  «-+=  4  4  Crook (mm)  (9d) HBD Crook  EJ Green Bow O Dry Bow  ^ Green Bow ODry Bow  64  64  55 46  t  37  4  Crook (mm)  (9c) LBD Crook  *  0  »  55 * CD  t  (S 28  &  19 10  46  37 28 19  L i^_ -fS= :  10  1  1 •  12  8  4  0  4  8  12  12  8  4  Bow (mm)  (9e) LBD Bow  0  4  8  12  Bow (mm)  (9f) HBD Bow  Figure 9. Warp results for conventional HBD and LBD drying  49  Table 5. Casehardening results from conventional drying Parallel  Perpendicular  [St. Dev.]  [St. Dev.]  0.10  0.10  [0.03]  [0.03]  0.13  0.14  [0.05]  [0.05]  LBD  HBD  —•—experimental 100  model  —•—experimental 100  y =2E-07x 3 +0.0005x 2 • 0.3498x +78.377  90  y = 1E-07X3 +0.0004X 2 • 0.283x + 6 3 . 9 3 9 R2 = 0 . 9 9 7 2  90  2  R =0.9979  80  80  70  70  „  60.  ~  60"  -  50  I  50  40  40  30  30  20  20  10  10 0  50  100  150  200  250  300  350 .  0  50  100  Time(hrs)  200  250  300  350  (10b.) HBD Drying Curve • Drying Rale (L-L)  •Drying Rate (L-L)  0.5 -  0.5 0.4 -  0.4 -  ^  3  § °- " 0.2  /  T3  f  0.3-  §  0.20.1 -  / 20  ^S  T3  0.1  0  150  Time (hrs)  (10a.) LBD Drying Curve  f  rrodel  40  60  80  100  M (%)  (10c.) LBD Linear-linear drying rate  0  / 20  40  60  80  100  M(%)  (10d.) HBD Linear-linear drying rate  Figure 10. Drying curves and drying rates for conventional drying  50  4.2  Superheated Steam Vacuum kiln drying  Both the LBD and the HBD timber were controlled by a time based schedule. Drying was terminated when a predetermined volume of water had been removed from the lumber. The target moisture content was 19% at the core. The LBD lumber dried to the final moisture content target in 227 hours including an 8-hour conditioning period, removing approximately 240 litres of water. The HBD lumber dried to the final moisture content target in 230 hours, including an 8-hour conditioning period, removing approximately 200 litres of water. The small difference in drying time is largely due to. a poorly optimized drying schedule, such that the drying rate during the 1st falling rate period was not maximized for the fastest drying times. This is discussed in detail later in this section. The maximum temperature reached in the SSV kiln was 80°C.  Determining the end point of drying based on the volume of water to be removed and weight loss from the lumber is a (or the) reliable method of control. Both the LBD and HBD loads were close to the target in the core region. The high standard deviation of the core moisture contents would not be considered acceptable for an industrial operation, as too many of the timbers were 'wet' in the core, even after the moisture had equalized in the board. This situation can probably be improved with a better drying schedule. The shell-core moisture content gradient is large, where a difference of 10% and 8% for the LBD and HBD respectively (Table 6) was observed. This gradient means that the board will be able to equalize to approximately the Mf and may meet most grading rules, but the lumber could be subject to further shrinkage or warp as the moisture migrates from the core region to the shell region.  51  In terms of drying time, sorting lumber based on density appears to be beneficial as well when drying in a SSV kiln. The LBD lumber lost 40 more litres of water in almost the same amount of time required for the HBD lumber to reach the target moisture content. These differences in drying times between the density groups could become an economic consideration if total kiln utilization were to be considered in a technology review.  Table 6. Moisture content results from SSV drying Drying Time (Hrs) 227 LBD 230 HBD  Mc(%)  Ms(%)  Mi (%)  Mf(%)  [St. Dev.]  [St. Dev.]  .79.6  16.1  21.9  11.9  [31.4]  [4.1] .  [6.5]  [2.9]  59.4  14.8  20.5  12.3  [17.2]  [3.2]  [8.0]  [1.7]  [St. Dev.] [St. Dev.]  AM (Mc-Ms) 10.0  8.2  Shrinkage results are reported in Table 7. As expected the HBD timbers experienced more shrinkage than the LBD timbers. The standard deviation is higher for the LBD timbers (width) than it is for the HBD timbers even though the shell moisture content is lower for the LBD timbers. The total higher Mf could be a factor in this result.  Figure 11 shows the initial and final moisture content distributions for theSSV samples. Similarly to the conventional samples, the effect of sorting green lumber on specific  52  gravity is observed.  Two statistically different initial moisture content distributions  become two statistically similar distributions after drying.  DBLBDMiHBLBDMf  0 B HBD Mi H B HBD Mf  t: 30  50 45 40 5- 35 X. 30  o  I 25  50 45 40 Co"  35  a- 20 a) LL  15  10 5 0  I I t=lr- Fl P [IIP in  lH|  o  CM  + o  MC Qass.  g- 20 £ 15 10 5 0  IB  I p p l E a p E3 + o o  o  MC Qass  (1 la.) SSV LBD M; and Mf distributions (1 lb.) SSV HBD Mj and Mf distributions Figure 11. Initial and final moisture content distribution of SSV samples  Table 7. Shrinkage results from SSV drying  LBD  HBD  Width (% shrinkage)  Thickness (% shrinkage)  [St. Dev.]  [St. Dev.]  3.0  2.65  [1.3]  [1.1]  2.99  3.13  [0.8]  [1.2]  Warp results (twist, bow and crook) are presented in Figure 12. The severity and high incidence of twist in both the LBD and HBD timbers do not reflect the severity of shrinkage or high drying stresses induced by this kiln, but are better explained by the  53  loading procedures of the kiln used in this experiment. Only four layers of 114 mm by 114 mm timbers can fit in this kiln, so there is no 'built-in' weight restraint. In most other kilns, the lumber may be layered 20 timbers high, effectively acting as a weight restraint for the lower timbers while the last few upper rows will have very little weight restraint. It is for this reason that the timbers are free to move according to the drying stresses and therefore these results should be viewed as the worst case, or potential of wood to warp during drying. There is very little difference in twist between the LBD and HBD groups. There is little difference in the incidence of crook between the green condition and the dry condition, for both the LBD and the HBD groups. Again, there is considerably less bow in both the LBD and the HBD groups, compared to crook, although there are 3 timbers where the bow is too high to meet most grading rules.  Casehardening results are presented in Table 8. Statistically, the results are not different, which can again be referred to the poorly optimized schedule or with respect to casehardening results, the gentle drying schedule. With such a high Me early in the schedule there was very little difference between the LBD and HBD drying, so the results would be expected to be similar. Even though the final difference between shell and core moisture contents was very pronounced, the kiln conditions that caused this gradient was short enough that residual stresses were not set in the lumber.  Drying curves and drying rates are presented in Figure 13. The drying curves have both the raw data and a fitted 3rd-degree polynomial model. Differentiating the polynomial and performing a linear piecewise regression produces the drying rates. The poorly  54  optimized schedule can be easily seen here. During the early stages of drying, when permeability is important, the drying rate is rapidly decreasing only to increase later in the 1st and 2nd falling rate periods. Ideally, the highest drying rates should be observed during the early stages of drying when the free water is being removed from the cell lumen. However, in this schedule, the lowest drying rates are seen in the early stages of drying, which demonstrates that it was the kiln schedule, and not the permeability of the lumber which was retarding faster drying. The high humidity in the kiln meant that the water being removed from the lumber had nowhere to go, and as a result, remained below the evaporating surface. If the humidity had been lower the moisture would have to be evaporated from the surface and drying would have been much faster. This would have been advantageous as instead of spending so much time removing the free water, that time could have been better spent diffusing water from the core to the shell region in the later stages of drying, decreasing the gradient and possibly the standard deviation around the core moisture content.  55  QGreenTwist QDryTwist  EP—  64 55  »  OGreen Twist O Dry Twist  CT  64  46  FgT;  a 16  "PI  0)  a. 37  | 3 7  «  3> 28  28 19  CI  10  10  —Eij  1  1  4  8  4  0  4  8  i' ; |  1 B  19  12  0  4  Twist (mm)  Twist (rrm)  (12a.) LBD Twist  (12b.) HBD Twist SGreenCrook QDry Crook  jGreen Crook D Dry Crook  64  64  55  55 46  » 46  n  •5. 37 E » 28  "a. 37  e  c=  & 28  19  19  10  10  1  r^fi >  §fei l„,'J i ^  1  12  8  12  8  4  0  4  8  12  Crook (rrm)  (12c.) LBD Crook  (12d.) HBD Crook  OGreen Bow Q Dry Bow  ©Green Bow' o D r y Bow  • ™L—!  64 55 » 46 -a. 37 « 28 19 10 1  m  64  -  55  -EH 4  "S. 37  4  ..  1 28  =  19  ^4  ^^^  10  S  1 8  4  Bow (rrm)  (12e.) LBD Bow  \  <f  El 0  <=a_  « 46  -3  0  !  4  8  Bow (rrm)  (12f.) HBD Bow  Figure 12. Warp results for LBD and HBD SSV drying  56  Table 8. Casehardening results from SSV drying Parallel  Perpendicular  [St. Dev.]  [St. Dev.]  0.13  0.12  [0.06]  [0.07]  0.14  0.14  [0.06]  [0.07]  LBD  HBD  -•—experiemental •  -•—experimental • y =• lE-05x 3 +0.0029x 2 • 0.4449x + 8 1 . 5 3 9 R2 = 0 . 9 9 6 4  100  200  y = -5E-06x 3 +0.0014x^ • 0.2591x +60.147 R2 = 0 . 9 9 8  300  100  Time (hrs)  Drying Rate (poly)  0  200 Time (hrs)  (13a.) LBD SSV drying curve •  -Model  (13b.) HBD SSV drying curve  H  •  Drying Rate (poly) — o - H  0.7  100  0.6 80 0.5 60  0.4 0.3 -  & O  0.2 0.1 0.0 ^ 20  40  60  M (%)  (13c.) LBD SSV drying rate  (13d.) HBD SSV drying rate  Figure 13. Drying curves and drying rates for SSV drying  57  4.3  Radio Frequency Vacuum kiln drying  Drying of both the LBD and HBD runs was controlled by HeatWave's patented "Endpoint Detection" control system. Both the LBD and HBD groups were dried in 29 hours. Since water absorbs the radio-frequency waves more rapidly than wood material, the LBD timbers will heat faster than the HBD resulting in a faster drying rate.  The moisture content results are presented in Table 9. Even though the LBD group had a higher initial moisture content, it dried to a lower final moisture content in the 29 hour drying time. The standard deviation around the Mf is quite impressive considering the drying time was only 29 hours. There is the possibility that the standard deviation could be reduced if the drying rate is reduced, allowing the wood time to equalize to the conditions in the kiln before drying termination and the unloading of the kiln. The moisture content gradient between the shell and core is an interesting result. In some instances, due to dielectric heating pattern of radio-frequency waves, it has been reported that the core is drier than the shell (Avramidis, 1996), but this was not the case with this project. The core retained more moisture than the shell in both the LBD and HBD groups. The gradient between the shell and core was very small and when examined in context with the Mf, the timbers should equalize well around the Mf with little continued shrinkage or warp1.  V  The moisture content results were collected two weeks after drying due to storage and transportation delays. Therefore, the moisture content gradients results should not be compared with the other two drying technologies where the moisture content gradients were measured immediately after drying.  58  Shrinkage results are presented in Table 10. In this case, the LBD lumber experienced greater shrinkage than the HBD lumber groups.  This observation is contrary to  conventional theory, with the only plausible explanation being the small difference in average moisture contents between the two density groups and the large standard deviation of the HBD timbers allowing for wetter timbers with less shrinkage.  Table 9. Moisture content results from RFV drying Drying Time (Hrs) 29 LBD 29 HBD  Mi (%)  Mf(%)  Mc(%)  Ms(%)  AM  [St. Dev.]  [St. Dev.]  [St. Dev.]  [St. Dev.]  (Mc-Ms)  78.5  16.1  17.8  14.8  3.0  [25.5]  [4.0]  [5.1]  [2.9]  59.2  18.0  20.7  16.1  [16.1]  [4.5]  [6.8]  [3.0]  4.6  Figure 14 shows the initial and final moisture content distributions for the RFV samples. Similarly to the conventional and SSV samples, the effect of sorting green lumber on specific gravity is observed. Prior to drying, the two density classes had statistically different initial moisture content distributions. However, because the RFV kiln was controlled on time, both the LBD and HBD samples were dried in 29 hours. This  59  produced two statistically different populations after drying, with the LBD have a lower Mf.  Table 10. Shrinkage results from RFV drying  LBD  HBD  Width (% shrinkage)  Thickness (% shrinkage)  [St. Dev.]  [St. Dev.]  2.81  2.91  [1.2]  [1.3]  2.47  2.44  [1.0]  [1.1]  Warp results (twist, bow and crook) are presented in Figure 15. The immediate impact of these results is the near absence of twist in both density classes. The effectiveness of HeatWave's weight restraint system could have a large impact on the ability of the  60  lumber to meet the grade standards of the lumber destined for the Japanese market. The results for crook would be considered acceptable and the drying technology does not appear to induce any more crook than what was present in the green condition, with a few exceptions in the HBD group, which can be attributed to the cellular structure and not the drying technology. Bow results are also considered acceptable with most timbers being under the acceptable limit of 3 mm of bow.  Casehardening results are presented in Table 11. Even though the LBD lumber had lower moisture contents and more shrinkage the casehardening was more severe in the HBD lumber. This means that the drying stresses were higher in the HBD lumber as proportionately more of the water was removed from the cell wall, inducing high stress between the shell and core as shrinkage occurs. It appears that the rate of drying is not always linked to casehardening as long as the gradient between shell and core is minimized, as with RFV drying.  61  0 Green Twist Q Dry Twist  0Green Twist Opry Twist  64  64  55  55  » 46  46  *  •5. 37  f  v, 28  <o 28  19  19  10  10  3 7  1  1 4  0  4  4  Tw ist (mm)  (15a.) LBD Twist  4  8 .  12  (15b.) HBD Twist  fjGreenCrook QDryCrook  ©GreenCrook DDry Crook  64  64 55 n  0 Twist (mm)  55 «  «  46  •§. 37 E » 28  •£ 37  19  19  10  10  1  1  t  «  28  4  0  4  Crook (rrm)  (15c.) LBD Crook  (15d.) HBD Crook BGreen Bow DDry Bow  H Green Bow DDry Bow  64 55 » 46 a. 37  ^B  64 55 « 46 -£ 37  I 28  w 28  19 10 1  19 10 12  8  4  0 Bow (rrm)  (15e.) LBD Bow  4  8  12  1 12  8  (15f.) HBD Bow  Figure 15. Warp results for LBD and HBD RFV drying  62  Table 11. Casehardening results from RFV drying  LBD  HBD  Parallel  Perpendicular  [St. Dev.]  [St. Dev.]  0.10  0.10  [0.04]  [0.04]  0.14  0.15.  [0.07]  [0.07]  Drying curves and drying rates are presented in Figure 16. Because of the constant power density used throughout these drying trials, the falling rate period is almost nonexistent (Avramidis and Zwick, 1996). This is because of the continuous thermal energy being supplied and then liberated to the lumber sets up continuous pressure and temperature gradients. The bulk of the water is removed along the longitudinal axis of the board where the diffusion coefficient is greatest.  Differentiating the 2n degree  polynomial fit to the drying curves produces the drying rate. The drying rate curves provide a clear example how the presence of more water in the LBD lumber allows faster heating and therefore faster drying.  63  • " . experimental  g S  90-] 804 ^ \ ^ 70^ ^ • ^ ^ 60 50 " 40 3020 " 10 0  Model  *~~experimental  y = 0.0039X2 - 2.2639X + 78.606 R2 = 0.9999  * 2  10  20  30  90 "i 80 70 60 i 50 ^->*»^_^  Model  y = 0.0019X2 - 1.4708X + 59.171 R2 = 0.9999  40 30 20 10  40  0  10  20  Time (hrs)  30  40  Time (hrs)  (16a.) LBD RFV drying curves •  .  (16b.) HBD RFV drying curve  Drying Rate  •  2.5  Drying Rate  2.5  2  2 -  Cs 1.5  I 1-51•o I"  I1 •o  0.5  0.5"  0  20  40  60  80  100  0  20  M (%)  (16c.) LBD RFV drying rate  40  60  80  M (%)  (16d.) HBD RFV drying rate  Figure 16. Drying curves and drying rates for RFV drying  4.4  Moisture Content Summary  Analysis of variance (ANOVA, a=0.05) results on Mf conclude that only the RFV dried HBD timbers were significantly higher than the other samples. However, as Mf is only an indicator of the potential after-drying Me, this result will not affect the stability of the timbers and does not reflect the performance of the RFV kiln. ANOVA results (a=0.05) on Mc conclude that only the LBD RFV timbers were significantly drier than the other timbers.  As the drying end-point was Mc=19%, it can be inferred that the control 64  mechanism for each technology is an effective method of control. ANOVA results (a=0.05) on Ms conclude that there is greater variation in the Ms than there was in the Mc. The LBD and HBD timbers dried in the conventional kiln were statistically not different as were the LBD and HBD timbers dried in the SSV kiln. Both the LBD and HBD timbers dried in the RPV kiln were significantly different from each other as well as the other timbers. However, examining the ANOVA results (a=0.05) from AM an important result is observed. The gradient between Ms and Mc can drive continued migration of water from the core region to the shell region, which can result in a decrease in stability. The LBD and HBD timbers dried in the conventional kiln were not statistically different from the HBD RPV dried timbers. The LBD RPV dried timbers were statistically different from the other groups, but had a smaller gradient, meaning those timbers were less likely to move as the two regions equalized. The LBD SSV dried timbers had the highest gradient and were statistically different from the HBD SSV timbers as well. This large gradient between the shell and core regions means that there is greater potential for these boards to warp as they equalize to a value similar to Me. An improved drying schedule should improve this potential problem.  4.5  Shrinkage Summary  As shrinkage is a function of moisture content below Mfsp, it is hard to draw comparisons of shrinkage results when the moisture contents are different. However, for the LBD timbers, the Mf showed no significant difference between the drying technologies, so their shrinkage results can be compared.  The ANOVA results (a=0.05) show no  significant difference in percent shrinkage for the LBD groups, both parallel and  65  perpendicular to their horizontal position in the kiln. For the HBD timbers, shrinkage results can only be compared between the conventionally dried timbers and the SSV dried timbers because the RFV dried timbers have a significantly higher Mf. There was no significant difference in the shrinkage results between the timbers dried in the conventional and SSV kilns. It cannot be determined if the significantly lower shrinkage of the HBD RFV dried timbers is a function of the Mf or the drying process.  66  Chapter 5 5.1  Equalizing Analysis and Discussion  Equalizing Conditions  The wood in the humidity chamber was exposed to the range of conditions that it would experience in the Tokyo region of Japan. The Me conditions in the chamber ranged from a peak of 19% to a low of 10% (Figure 17).  25 20 ~ CD  5  15 .  10 5 0 1  2  3  4  5  6  7 8 Week#  9  10  11  12  13  14  Figure 17. Equilibrium moisture content conditions in the humidity chamber  67  5.2  Conventionally dried samples  The average board weights are presented in Figure 18.  The LBD samples are  consistently 2 kg lighter than the HBD samples but neither group experienced any major changes in weight throughout the exposure to the Tokyo M^.  Shell/core moisture  contents and dimensional changes are presented in Figure 19. The moisture content response of the samples is presented in Figures 19a and 19b for LBD and HBD samples, respectively.  In both the LBD and the HBD samples, the shell moisture contents  responded very quickly to the changes in the Me and the magnitude of the response was a. maximum of 5%. The core moisture content response shows a slow decreasing trend to the final Me and only experienced small changes in the range of 2% M. The shrinkage and swelling results are presented in Figure 19d and 19e. Even though there were some large fluctuations in the moisture content, the corresponding results in shrinkage and swelling still remained in the target size range of 105 mm. The largest change for both LBD and HBD samples was only 0.4 mm. Figures 20a-f presents the warp results between the kiln-dry condition and the Mf condition. Overall, there is no significant increase in warp as the timbers equalized to the equilibrium moisture content. In the LBD group, there were no timbers that went from having less than 2 mm or twist to having greater than 2 mm of twist after equalizing. In the HBD group, there was only one board that experienced an increase in twist from less than 2 mm to more than 2 mm after equalizing. Most timbers had some degree of crook and bow both in the kiln-dry condition and the equalized condition. Again, there were no significant changes between the kiln-dry condition and the equalized condition for both crook and bow.  68  -•—LBD Weight - « — H B D Weight 13.0 12.5 „  12.0  o  11.0 10.5 10.0 9.5  —I  1  2  1  j  1  ;  i  1  1  3  4  5  6  7  8  9  1  1  ;  I—  10 11 12 13 14  Week*  Figure 18. Weight changes during equalization  69  "Shell — * — C o r e  -•—Shell —•—Core  20  10 1  2  3  4  5  6  7  8  9 10 11 12 13 14  1  2  3 4  5  6  Week#  (19a.)  7  8  9 10 11 12 13 14  Week#  LBD moisture content response (19b.) HBD moisture content response during equalizing. during equalizing "Thick —m— width  - • — T h i c k —®~"Width  106.0  105.8  1  2  3  4  5  6  7  8  9 10 11 12 13 14  1  2  3  4  5  6  7  8  9 10 11 12 .13 14  Week # Week #  (19c.) LBD dimensional response during (19d.) HBD dimensional response during ' equalizing. equalizing Figure 19. Moisture content and dimensional response of conventionally dried samples during equalization. ..  70  O Kiln dry Twist ® Equalized Twist  112 *  97  |  87  w  0 Kiln dry Twist O Equalized Twist  134  T5B i  96 71  • ' { ' ' " Y f | '"""'i''''i''''''' i '*'-*  67  46  44  28 5  1 4  2  I 0  2  4  6  4  2  0  Twist (mm)  2  4  6  8  Twist (mm)  (20a.) LBD twist response between kiln (20b.) HBD twist response between kilndry and equalized conditions. dry and equalized conditions. O Kiln dry Crook B Equalized Crook  te^  112  »  £gj  97  gM  134  j  96  TB3  (U  87  •3 Kiln dry Crook S Equalized Crook  71  67  46  44  28  1  •  8  6  4  ;  '";  I""T-"I  2  0  2  I"*"""  4  5 6  6  4  2  0  2  4  Crook (mm)  Crook (mm)  (20c.) LBD crook response between kiln (20d.) HBD crook response between kiln dry and equalized conditions. dry and equalized conditions. ^KilndryBow ^ Equalized Bow  OKiln dry Bow @ Equalized Bow  112  134  a  97  96  §•  87  71  67  46  CO  <°  44  _I • 8  6  4  2  28  | , .  " j |—Lm  1  0  2  Bow (mm)  | _i_ 4  5 6  6  4  2  0  2  4  6  Bow (mm)  (20e.) LBD bow response between kiln- (20f.) HBD bow response between kilndry and equalized conditions. dry and equalized conditions. Figure 20. Warp results between conventionally dried and equalized conditions  71  5.3  Superheated Steam Vacuum dried samples  The average board weights during equalization are presented in Figure 21. As was found with the conventionally dried timbers, the LBD samples are consistently lighter than the HBD samples. However, the difference between density groups was not as great for the SSV dried timbers, having only a 1.5 kg difference.. Neither group experienced any major changes in weight throughout the exposure to Tokyo's equilibrium moisture content. Shell/core moisture contents and dimensional changes during equalization are presented in Figure 22. Figures 22a and 22b report the moisture content responses of LBD and HBD samples respectively. In both the LBD and the HBD samples, the shell moisture contents responded very quickly to the changes in the Me. However, the shell region of the HBD timbers absorbed significantly more moisture than the LBD samples. The magnitude of the response was a maximum of 7% for the HBD samples and 4% for the LBD samples. This finding follows that there are more potential bonding sites for water in the HBD timbers than in the LBD timbers.  The core moisture content  experiences a similar response to the conventionally dried timbers and shows a slow decreasing trend to the Mf and only experienced small fluctuations of 2%. The shrinkage and swelling results are presented in Figures 22c and 22d. Even though there were some large fluctuations in moisture content, especially in the shell HBD timbers, the corresponding results in shrinkage and swelling still remained within the target size range of 105 mm.  In the HBD group, the increase of 7% in the shell moisture content  corresponds to a 0.2 mm increase over a 1-week period. Figures 23 a-f present the warp results between the kiln-dry condition and the final Me condition. Overall, there is no significant increase in warp as the timbers equalized to the Me. There was a general  72  decrease in the amount of twist in both the LBD and the HBD samples. This finding is probably a factor of the load acting as weight restraint during equalizing, but this improvement in the severity of twist could be factored into increasing the amount of acceptable twist in a grade category. Most timbers had some degree of crook and bow both in the kiln-dry condition and the equalized condition.  Again, there were no  significant changes between the kiln-dry condition and the equalized condition for both crook and bow.  —•— LBD Weight ~"*~ HBD Weight 13.0 :  12.5 12.0  10.5  9 . 5 -I  1  1  1  i  1  1  1  1  i—-i  2  3  4  5  6  7  8  9  1  1  1  10 11 12 13 14  Week #  Figure 21. Weight changes during equalization  73  -Shell —®—Core  "Shell — ® — Core  24  10-1 1  10 2  3  4  5  6  7  8  9  10 11 12 13 14  1  2  3  4  5  6  Week#  (22a.)  7  8  9  10 11 12 13 14  Week#  LBD moisture content response (22b.) HBD moisture content response during equalizing. during equalizing - • — T h i c k —<•—Width  - • — T h i c k —®—Width  105.0 1  2  3  4  5  6  7  8  Week#  9 10 11 12 13 14  1  2  3  4  5  6  7  8  9 10 11 12 13 14  Week #  (22c.) LBD dimensional response during (22d.) HBD dimensional response during equalizing. equalizing Figure 22. Moisture content and dimensional response of SSV dried samples during equalization.  74  Q Kiln dry Twist Q Equalized Twist  Q Kiln dry Twist ^ Equalized Twist  V—f-^— 112  134 *  96  I  86  1  71  I  67  <°  46  97  44  28  1  5 6  4  2  0  2  4  {  I'i  I  . .  |  i = ^  6  6  4  2  Twist (mm)  • 0  2  4  6  Twist (mm)  (23a.) LBD twist response between kiln (23b.) HBD twist response between kiln dry and equalized conditions. dry and equalized conditions. Q Kiln dry Crook 0 Equalized Crook  Oftln dry Crook 'Q Equalized Crook  112  134  97  96  86  71  67  46  44  28 -'•  1 6  4  2  '  ' »  0  2 '  5 4  6  8  Crook (mm)  6  4  2  0  2  4  Crook (mm)  (23c.) LBD crook response between kiln (23d.) HBD crook response between kiln dry and equalized conditions. dry and equalized conditions. D  Kiln dry Bow B Equalized Bow  ^Kiln dry Bow E Equalized Bow  112  134  97  96  86  71  67  46  44  28  1  5 6  4  2  0  2  4  Bow (mm)  (23e.) LBD bow response between kiln (23f.) HBD bow response between kiln dry and equalized conditions. dry and equalized conditions. Figure 23. Warp results between SSVdried and equalized conditions  75  5.4  Radio Frequency Vacuum dried samples  The average board weights during equalization are presented in Figure 24. As was found with the conventional and SSV dried timbers, the LBD samples are consistently lighter than the HBD samples. However, the difference between density groups was greatest for the RFV dried timbers, having a 1.7 kg difference. Neither group experienced any major changes in weight throughout the exposure to Tokyo's equilibrium moisture content. Shell/core moisture contents and dimensional changes during equalization are presented in Figure 25. Figures 25a and 25b report the moisture content responses of LBD and HBD samples respectively. In both the LBD and the HBD samples, the shell moisture contents responded very quickly to the changes in the Me. However, the shell region of the HBD timbers absorbed significantly more moisture than the LBD samples. The magnitude of the response was a maximum of 7% for the HBD samples and 4% for the LBD samples. This finding follows that there are more potential bonding sites for water in the HBD timbers than in the LBD timbers. The core moisture content experienced a similar response to the conventionally dried timbers and shows a slow decreasing trend to the final Me and only experienced small fluctuations in the range of 2%. The shrinkage and swelling results are presented in Figure 25c and 25d. Even though there were some large fluctuations in moisture content, especially in the shell HBD timbers, the corresponding results in shrinkage and swelling still remained within the target size range of 105 mm.  In the HBD group, the increase of 7% in the shell moisture content  corresponds to a 0.2 mm increase over a 1 -week period. Figures 26a-f present the warp results between the kiln-dry condition and the final Me condition. Overall, there is no significant increase in warp as the timbers equalized to the Me. There was a general  76  decrease in the amount of twist in both the LBD and the HBD samples. This finding is probably a factor of the load acting as weight restraint during equalizing, but the improvement in the severity of twist could be factored into increasing the amount of acceptable twist in a grade category. Most timbers had some degree of crook and bow both in the kiln-dry condition and the equalized condition.  Again, there were no  significant changes between the kiln-dry condition and the equalized condition for both crook and bow. "~H^"~ LBD Weight —w-" HBD Weight 13.0  ui  12.0 "  b  Weight (kg)  12.5 "  >  »—_ H > - .... &  •$——-O-  ffi  O-  0  0~  5  7  8  9  10 11 12  O"—&*~»—Q—^  10.5 10.0 " 9.5 2  ' 3  4  6  13 14  Week #  Figure 24. Weight changes during equalization  11  "Shell — • — C o r e  "Shell — • — C o r e  20  1  1  2  3  4  5  6  7  8  9  10 11 12 13 14  1  2  3  4  5  6  (25a.)  7  8  9  10 11 12 13 14  Week#  Week#  LBD moisture content response (25b.) HBD moisture content response during equalizing. during equalizing -Thick —®—Width 106.0  -•—Thick —•—Width 106.0  105.8  E E —  105.6  1  2  3  4  5  6  7  8  Week#  9 10 11 12 13 14  1  2  3  4  5  6  7  8  9 10 11 12 13 14  Week*  (25c.) LBD dimensional response during (25d.) HBD dimensional response during equalizing. equalizing Figure 25. Moisture content and dimensional response of RFV dried samples during equalization.  78  O Kiln dry Twist ^ Equalized Twist  I  134  Q Kiln dry Twist B Equalized Twist  112  I  96  97  71  84  46  67  28  44  5  1 6  4  2  0  2  4  4  2  Twist (mm)  0  2  4  6  Twist (mm)  (26a.) LBD twist response between kiln (26b.) HBD twist response between kiln dry and equalized conditions. dry and equalized conditions. Q Kiln dry Crook Q Equalized Crook  I  134 *  Q Kiln dry Crook Q Equalized Crook  112  ' 1—4'  96  97  '  84  71  F <° 46  67  IIH-4V  28  '—I—""""J  5 6  4  2  0  44  I^ 2  1 4  Crook (mm)  (26c.) LBD crook response between kiln (26d.) HBD crook response between kiln dry and equalized conditions. dry and equalized.conditions. ^Kiln dry Bow O Equalized Bow  Q«iln dry Bow Q Equalized Bow  134  112  96  97  71  84  46  67  28  44  5  1 8  6  4  2  0 Bow (mm)  2  4  6  6  4  2  0  2  4  6  Bow (mm)  (26e.)  LBD bow response between kiln (26f.) HBD bow response between kiln dry and equalized conditions. dry and equalized conditions. Figure 26. Warp results between RFV dried and equalized conditions  79  5.5  Stability Summary  The stability of the kiln-dried specimens, represented by the coefficient of variation for twist, crook and bow are reported in Figures 27-29.  The coefficient of variation  expresses sample variability relative to the mean but is not influenced by the actual magnitude of variation.  The coefficient of variation for twist was lowest for the conventionally dried LBD and HBD samples, compared to the SSV and RFV dried samples. The LBD samples had a higher coefficient of variation than the HBD samples, however both of these coefficients were the least variable over the test period (Figure 27a). The coefficient of variation for twist in the SSV and RFV dried samples shows the LBD having a lower coefficient than the HBD samples. The results show much more variation over the test period in both density groups (Figures 28a and 29a respectively) compared to the conventionally dried samples.  The coefficient of variation for crook was similar for both density groups and between all three drying technologies over the testing period. There was very little variation in crook over the testing period suggesting that the stability of these boards, with respect to crook, is good. Additionally, the coefficient for crook was smaller than it was for twist (Figures 27b, 28b and 29b).  80  The coefficient of variation for bow was higher for the LBD samples than it was for the HBD samples. technologies.  However, the coefficient was similar between the three drying The coefficient for bow was intermediary between twist and crook.  Therefore, twist is the primary concern to be addressed for the improved stability of these samples, followed by bow then crook.  81  f  -LBD — ® - H B D 3.5 3.0  10  15  Week#  (27a.) Coefficient of variation for twist of conventionally dried samples "LBD  "HBD  3.5 3.0 „ 2.5 £ 2.0 o> 1 5 1.0 0.5 0.0 10  15  Week#  (27b.) Coefficient of variation for crook of conventionally dried samples -LBD —a—HBD  3.5 3.0 „ 2.5 £ 2.0 > 1.5 1.0 0.5 0.0 10  15  Week #  (27c.) Coefficient of variation for bow of conventionally dried samples Figure 2 7. Coefficient of variation for stability of conventionally dried samples  82  -•—LBD —®~HBD 3.5 3.0 „ 2.5 ° i 2.0 > 1.51.0 0.5 -  It"""*--  fls.-ra  9>+- -t^r  *a  r \  ^^if  ^  •  u.o •  '  0  '  5  10  15  Week#  (28a.) Coefficient of variation for twist of SSV dried samples —•— LBD —«— HBD 3.5 "i  •  1  3.0 „ 2.5g. 2.0 > 15-  o  Iktt^i^^pJ^^^ 1.0 0.5 •o.o (3  5  10  15  Week #  (28b.) Coefficient of variation for crook of SSV dried samples •  *  |_gQ _ p _ |_|gQ  3.5 3.0 „ 2.5 g. 2.0 > 15  o  g = -g ==jiB _^0 1.0 0.0 0.5 0  'D-tt-i^HK*^  '  '  5  10  '  15  Week#  (28c.) Coefficient of variation for bow of SSV dried samples Figure 28. Coefficient of variation for stability of SSV dried samples  83  •LBD - * - H B D 3.5 3.0 -? 2 - 5 b 2.0  -=®s  d1-5 °  1.0 0.5 0.0  10  15  Week#  (29a.) Coefficient of variation for twist of RFV dried samples -LBD — « — HBD  O  3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0  L  10  15  Week#  (29b.) Coefficient of variation for crook of RFV dried samples -•—LBD —®~HBD  3.5 3.0 „ 2.5 g. 2.0 >  1.5 1.0 0.5 0.0  10  15  Week#  (29c.) Coefficient of variation for bow of RFV dried samples Figure 29. Coefficient of variation for stability of RFV dried samples  84  >  Chapter 6  Conclusions and Recommendations  Wood is a naturally variable material, and as such when it is used as a building material, these variations must be expected. Building codes and grading standards such as MSR lumber grading, take into consideration these variations and allow for variation by using the lowest 5th percentile property to establish the standard strengths used in design stress development. This way, an end-user will know that a piece of lumber will perform to a certain level at least, if not better.  The results of this research show that lumber dried to approximately 19% moisture content at the core, does respond to a change in the equilibrium moisture content. The simulated Tokyo equilibrium moisture content in the humidity chamber caused changes in moisture content, dimensions and warp but the changes would be considered acceptable when taken in context that wood is a naturally variable material.  The stability of this unrestrained lumber in the humidity chamber should be considered acceptable. In most instances, the increases seen in warp were in the magnitude of 2-3 mm and over a 2 m board these changes are relatively small.  This scenario of  unrestrained lumber should be considered the worst case situation and that lumber used in post and beam housing will have the added weight and restricted movement due to connector and adjacent members to minimize movement. Addressing the variation in twist during equalization will have the most beneficial impact.  85  The shell moisture content was susceptible to variation in moisture content and as such is prone to climb into the moisture content range where decay is possible. However, as long as the lumber is not subjected to extend periods of wetting, these findings also show that the lumber has the ability to react quickly to drying conditions and return to moisture content levels where the possibility of decay is reduced. As decay was cited as one reason why the post and beam homes performed so poorly during the Hyogoken-Nanbu earthquake, further consideration should be given to drying the lumber to moisture content levels below 19%. Knowing the difference in the equilibrium moisture content between Tokyo and Vancouver is 7%; it would be logical to calculate the average equilibrium moisture content between the two regions and target 15% moisture content at the core. This may not only decrease the potential of decay but also decease the amount of movement seen in use as the wood equalizes to ambient conditions. A business case would have to be developed to determine the impact of drying to 15% to see if the economics make sense when shrinkage, drying time and drying degrade are considered.  The shrinkage and swelling changes during equalization are encouraging because even though there were some large variations in the shell moisture contents, the corresponding swelling and shrinking changes remained in the 105 mm range. This result means that there should not be any destructive forces created due to shrinkage or swelling unless there is unequal shrinkage due to the natural variation of lumber.  Although there are obvious differences between the low basic density and high basic density materials, they are small enough that sorting lumber based on performance during  86  equalization is not a strong argument. However, as can be seen from the drying results, there is a strong.argument for sorting lumber by basic density and a business case should be analyzed especially if drying to 15% at the core is to be considered.  There was no direct comparison between drying technologies in this study, although it was determined there were no differences between the final equilibrium moisture content that the wood will reach by drying in any one of the different technologies.  The  temperature that the wood was exposed to varied from a low of 30 C to approximately 90 C and these differences did not affect the wood enough to impart any changes in the final equilibrium moisture content.  All three drying technologies dried the lumber to meet the standards, however the extraordinary good warp results from the RFV weight restraint system strongly supports the possibilities of reducing warp during drying and the resulting implications of movement during equalization. If a board has very little warp, the addition of 2 to 3 mm of warp from drying with RFV will not be a negative result.  As this study measured unrestrained lumber during equalization, it is recommended that another study repeat this research with the addition of using restrained lumber, using a scale-model of a post and beam house to determine if the results from this study can be replicated and whether the movement of individual timbers contribute to the overall movement of the whole structure.  87.  References Ahmet, K., Dai, G., Jazayeri, S., Tamlin, R., Kaczmar, P and Riddiough, S. 1999. Experimental procedures for determining equilibrium moisture content of twenty timber species. Forest Prod. J. 49(l):88-93 Avramidis, S. and Mackay, J. 1988. Development of kiln schedules for 4x4 Pacific Coast hemlock. Forest Prod. J. 38(9):45-48 Avramidis, S. and Oliveira, L.C. 1988. Influence of presteaming on kiln drying of thick Hem-fir lumber. Forest Prod. J. 43(11/12):7-12 Avramidis, S. and Zwick, R. 1992. Exploratory RF/V drying of 3 BC coastal softwoods. Forest Prod. J. 42(7/8): 17-24 Avramidis, S. and Zwick,R. 1996. Commercial scale Radio-frequency Vacuum Drying. Forest Prod. J. 46(6):27-36 Avramidis,S., Ellis, S., Liu, J. 1993. The alleviation of brown stain in Hem-fir through the manipulation of kiln drying schedules. Forest Prod. J. 43(10):65-69 Avramidis, S. 1988. The effect of presteaming on drying quality of thick Pacific Coast Hemlock. SCBC Po# 12-71-k-474 Chaffe, S.C. 1991. A relationship between equilibrium moisture content and specific gravity in wood.* J. Inst. Wood Sci. 12(3): 119-122 Council of Forest Industries. 1998. BC Forest Industry Fact Book, 1997. Dedrick, D. and Ziegler, G. 1984. Restraint application during drying of Hem-fir lumber. In Proc. Western Dry Kiln Club.: pp 12-24 Doucette, G. 2000. Personal Communication. Ellis, S. and Avramidis, S. 1993. Brown stain in Pacific Coast Hemlock. IAWA J. 14(l):23-28 Garrahan, P. 1992. Energy Reduction through the drying of moisture sorted lumber. Ontario Ministry of Energy. Toronto, ON. Gaston, C , Cohen, D. and Fell, D. 2000. Wood Market Trends in Japan. Forintek Canada Corp. Vancouver, BC. Sp-43  88  Gonzalez, J. 1995. Vancouver, BC.  The Properties of Western hemlock.  Forintek Canada Corp.  Haygreen, J. and Bowyer, J. 1989. Forest Products and Wood Science: An Introduction. Iowa State University Press. Jamroz. 1997. Method and apparatus for non-contact and rapid identification of wood species. U.S. Patent 5406378 Jozsa, L. and Middleton, G. 1994. A discussion of wood quality attributes and their practical implications. Forintek Canada Corp. S.P. No.-34. Vancouver, BC. Kozlik, C. and Ward. 1981. Properties and kiln drying characteristics of young growth Western hemlock dimension lumber. Forest Prod. J. 31(6):45-53 Li, M., Avramidis, S., Oliveira, L.C, and Hartley, I.D. 1997. Effect of vertical air gap, air velocity and fan revolution on drying characteristics of thick Pacific Coast Hemlock. Holzforschung 51:381-387 Lin, R. and Kozlik, C. 1971. Permeability and drying behavior of Western hemlock. Proc. 22nd Annual Meeting Western Dry Kiln Association, pp 44-50. Oregon State University Press. Mackay, J. 1995. A review of Forintek studies on drying characteristics of Hem-fir and methods to improve efficiency of kiln drying. CFS Report No. 31C Mackay, J. and Oliveira, L.C. 1989. Kiln Operators's Handbook for Western Species. Forintek Canada Corp. Vancouver, BC. SP-31 Milota, M., Morrell, J., Lebow, S.T. 1993. Reducing moisture content variability in kiln dried Hem-fir lumber though sorting: A simulation. Forest. Prod. J. 43(6):6-12 Milota, M. and Wengert, G. 1993. Applied Drying Technology 1988-1993. Forest Prod. J. 45(5):33-41 Mitchell, P. and Bigbee, K. 1989. Drying variables affecting drying rate and moisture content distribution on Western hemlock. Proc. Western Dry Kiln Association. Mullins, E. and McKnight, T. 1981. Canadian Woods: Their properties and uses. University of Toronto Press. Toronto. Neilson,R., Dobie, J. and Wright, D. 1985. Conversion factors for the Forest Products Industry in Western Canada. Forintek Canada Corp. Vancouver, BC. Sp-24R Oliveira, L.C. 1998. Best Drying Practices for Coastal Lumber Products Exported to the Japanese Market. CONFIDENTIAL REPORT. Forintek Canada Corp. Vancouver, BC. 89  Oliveira, L.C. and Mackay, J.F.G. 1998. SSV Drying of BC Softwood Lumber. Science Council of British Columbia. Burnaby BC. SCBC No. FR-96/97-195. FRBC No. 96074-RE Pang and Dakin. 1999. Drying rate and temperature profile for SSV drying and moist air drying of softwood lumber. Drying Technol. 17(6):1135-1147 Panshin, A.J. and DeZeeuw, C. 1970. Textbook of Wood Technology. Volume 1. McGraw-Hill. Toronto. Siau, J. 1971. Flow in Wood. Syracuse University Press. Syracuse. Simpson, W.T. 1973. Predicting equilibrium moisture content of wood by mathematical models. Wood and Fiber. 5(l):41-49 Simpson, W.T. 1999. Drying and control of moisture content and dimensional changes. Wood handbook- Wood as an engineering material. Gen. Tech. Rep. FPL-GTR-113. Forest Products Lab, Madison, WI. Skaar, C. 1988. Wood-water relations. Springer-Verlag. New York. Smith, R. 1984. Energy Consumption in Lumber Drying Operations. In Proc. Western Dry Kiln Club. Smith, R.G. 2000. Personal communication Time, B. 1998. Hygroscopic Moisture Transport in Wood. Ph.D. Thesis. Norwegian University of Science and Technology. Trusty, W.B. and Meil, J.K. 1997. ATHENA: An LCA Decision Support ToolApplication, Results and Issues. Proc: Second International Conference on Buildings and the Environment. Paris, France. 12pp Walker, J.C.F. 1993. Primary Wood Processing. Principles and Practice. Chapman and Hall. New York. Warren, S. and Johnson. 1997. The economic benefits of sorting SPF lumber to be kilndried on the basis of initial moisture content. Forest Prod. J. 47(3):58-61 Wengert, G. 1988. The Wood doctor's Rx. Virginia Polytechnic Institute and State University. Blacksburg. Zhang, Y., Oliveira, .L.C, Avramidis, S. 1994. Drying Characteristics of Hem-fir squares as affected by species and basic density presorting. Forest Prod. J. 46(2):44-50 Zwick, R. 1999. Q-Sift: An integrated RFV Solution to drying hemlock baby-squares. http://www.heatwave.com  90  Appendix A Photographs  Preparing to cut the 8 meter boards into matched samples  Cutting the 8 meter boards into matched samples  91  Measuring weights and volume for Mj and G„ calculations  RFV samples (LBD and HBD) prior to drying  92  Data collection tools  Conventional kiln with LBD load  93  SSV Kiln with LBD Load  RFV kiln at HeatWave Drying Systems Ltd, Cresent Valley, BC  94  Weight measurements during equalization  Dimensional measurements during equalization  95  Shell moisture content measurements during equalization  Core moisture content measurements during equalization  96  Cart being loaded for the humidity chamber  t 1  4  ... ..,  -  1  *.-. Mft  Fully loaded cart being pushed into the humidity chamber  97  Fully loaded humidity chamber  98  Appendix B Page/Date/Time Database Response  ANOVA Table  1 04-12-2001 09:02:35 C:\Program Files\ncss97\IMC.S0 A_LBD,B_LBD,C_LBD  Tests of Assumptions Section Assumption Skewness Normality of Residuals Kurtosis Normality of Residuals Omnibus Normality of Residuals Modified-Levene Equal-Variance Test  Test Value 5.6511 1.9578 35.7677 0.7964  Prob Level 0.000000 0.050256 0.000000 0.452305  Decision (0.05) Reject Accept Reject Accept  Box Plot Section Mi  140.00  o 100.00  < 60.00  20.00  V T T A_LBD  B_LBD Variables  C_LBD  Expected Mean Squares Section Source Term Denominator Term DF Fixed? Term A(...) 2 Yes S(A) S(A) 207 No Note: Expected Mean Squares are for the balanced cell-frequency case. Analysis of Variance Table Source Sum of Term DF Squares (Alpha=0.05) A(...) 2 711.5143 S(A) 207 159887.1 209 160598.7 Total (Adjusted) Total 210 Term significant at alpha = 0.05  Mean Square 355.7571 772.4017  F-Ratio 0.46  Expected Mean Square S+sA S(A)  Prob Level  Power  0.631559  99  Page/Date/Time Database Response  Analysis of Variance Report 2 04-12-2001 09:02:35 C:\Program Files\ncss97\IMC.S0 A_LBD,B_LBD,C_LBD  Kruskal-Wallis One-Way ANOVA on Ranks Hypotheses Ho: All medians are equal. Ha: At least two medians are different. Test Results Method Not Corrected for Ties Corrected for Ties  DF 2 2  Number Sets of Ties Multiplicity Factor  24 180  Chi-Square (H) 0.6767598 0.676773  Prob Level 0.712924 0.712920  Decision(0.05) Accept Ho Accept Ho  Mean Rank 100.89 106.44 109.18  Z-Value -0.7781 0.1578 0.6203  Median 70.3 71.5 69.65  Group Detail Group AJ.BD BJ.BD C LBD  Count 70 70 70  Sum of Ranks 7062.00 7450.50 7642.50  Means and Effects Section Term All A: A_LBD B_LBD C LBD  Count 210  Mean 77.78714  70 70 70  75.25857 79.58714 78.51572  Standard Error  3.321793 3.321793 3.321793  Effect 77.78714 -2.528571 1.8 0.7285714  Plots of Means Section Means of Mean Value 80.00  78.75-  77.50  76.25  B_LBD  C LBD  Variables  100  

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.831.1-0090424/manifest

Comment

Related Items