UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Dielectric characteristics of two BC coastal species during radio frequency heating Zhou, Bingning 1997

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

Item Metadata

Download

Media
831-ubc_1997-0375.pdf [ 3.86MB ]
Metadata
JSON: 831-1.0075272.json
JSON-LD: 831-1.0075272-ld.json
RDF/XML (Pretty): 831-1.0075272-rdf.xml
RDF/JSON: 831-1.0075272-rdf.json
Turtle: 831-1.0075272-turtle.txt
N-Triples: 831-1.0075272-rdf-ntriples.txt
Original Record: 831-1.0075272-source.json
Full Text
831-1.0075272-fulltext.txt
Citation
831-1.0075272.ris

Full Text

DIELECTRIC CHARACTERISTICS OF TWO BC COASTAL SPECIES DURING RADIO FREQUENCY HEATING by BINGNTNG ZHOU B.Sc, Northeast Forestry University, 1989 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENT S FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Department of Forestry) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA June, 1997 © Bingning Zhou, 1997 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree . that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. . Department of The University of British Columbia Vancouver, Canada Date Zw&fa/tffJ DE-6 (2/88) Abstract The purpose of this study was to evaluate the dielectric properties (loss factor) of two British Columbian softwoods (western hemlock [Tsuga heterophylla (Raf.) Sarg.] and western red cedar [Thujaplicata Donn]) at different moisture content, temperature, and electric field strength levels. This study is expected to provide useful information for the development of a mathematical model to describe heat and moisture transfer mechanisms in wood during radio frequency/vacuum (RF/V) drying. Such a model would predict and optimise the RF/V drying behaviour of wood on a commercial scale without the requirement of experimentation. In this study, the radial direction loss factor of full-sized western hemlock sapwood and heartwood, and western red cedar heartwood timbers was measured using the direct calorimetric data method with a laboratory-scale RF/V dryer at the frequency of 13.56 MHz, moisture content range between 10 and 80%, temperature range between 25 and 55 °C, and root mean square (rms) electrode voltages 0.8 and 1.1 kV, respectively. The results indicated that the moisture content, temperature, electric field strength and species significantly affected the loss factor. Empirical regression equations were derived based on the experimental data, that made possible the calculation of the loss factor and power density within the wood mass during RF heating. ii Table of Contents Abstract ii Table of Contents iii List of Tables v List of Figures vi Acknowledgements 1 X Chapter 1 Introduction 1 Chapter 2 Literature Review 4 2.1 Reasons for drying wood 4 2.2 Conventional and RF/V drying 5 2.3 Electromagnetic waves 11 2.4 Wood: a dielectric material 13 2.5 Types of polarization 14 2.6 Dielectric properties of wood 16 2.7 Power density 20 2.8 Time rate of temperature change 24 2.9 Influence of wood variables on its dielectric properties 25 2.10 Methods of measuring wood dielectric properties 33 Chapter 3 Experimental Procedure 37 3.1 Materials 37 3.2 RF/V dryer 37 3.3 Calibration of the RF electrode voltage 40 iii 3.4 Methods 44 3.4.1 Determination of initial moisture content of specimens (Mj) 44 3.4.2 Determination of oven dry weight of specimens (Wo) 46 3.4.3 Determination of average moisture content (MJ of specimens during each run 46 3.4.4 Calculation of average density (p) during each run 47 3.4.5 Temperature monitoring and calculation of rate change (3T/9t) 47 3.4.6 Determination of specific heat of wood (cp) 49 3.4.7 Dielectric runs 49 Chapter 4 Results and Discussion 51 4.1 Calibration of the RF electrode voltage 51 4.2 The experimental data 58 4.3 Analysis of the experimental results 68 4.3.1 Moisture effect 70 4.3.2 Temperature effect 76 4.3.3 Electric field strength effect 83 4.3.4 Species effect 92 4.4 Regression equations 98 4.5 Calculated power density 103 Chapter 5 Conclusions 109 Literature Cited Ill iv List of Tables Table 1: Time rate of temperature change for the calibration of the RF electrode voltage 55 Table 2: Relative error of calculated specific heat 59 Table 3: Mean loss factor for western hemlock sapwood at 0.8 kV 61 Table 4: Mean loss factor for western hemlock sapwood at 1.1 kV 62 Table 5: Mean loss factor for western hemlock heartwood at 0.8 kV 63 Table 6: Mean loss factor for western hemlock heartwood at 1.1 kV 64 Table 7: Mean loss factor for western red cedar heartwood at 0.8 kV 65 Table 8: Mean loss factor for western red cedar heartwood at 1.1 kV 66 Table 9: Fully factorial ANOVA — effects of the variables on the loss factor 69 Table 10: Quadratic regression parameters, a, b, and c, from the loss factor versus moisture content data 74 Table 11: Linear regression parameters, a and b, from loss factor versus temperature data 82 Table 12: Regression line slopes from the loss factor versus rms electrode voltage data 89 Table 13a: Calculated power density for western hemlock sapwood at 0.8 kV, various moisture contents and temperatures 104 Table 13b: Calculated power density for western hemlock sapwood at 1.1 kV, various moisture contents and temperatures 105 Table 14: Calculated power density for western hemlock heartwood at two rms electrode voltages, various moisture contents and temperatures 106 Table 15: Calculated power density for western red cedar heartwood at two rms electrode voltages, various moisture contents and temperatures 107 v List of Figures Figure 1: Electromagnetic waves 12 Figure 2: Current-voltage relation in an ideal capacitor 18 Figure 3: RC circuit for dielectrics (wood) 19 Figure 4: Vector diagram of dielectrics (wood) 21 Figure 5: Dispersion and absorption curves representing the Deby model 31 Figure 6: A typical Q-meter measuring circuit 34 Figure 7: Cross-section schematic of the RF/dryer 39 Figure 8: Location of plexiglas blocks during the RF voltage calibration 42 Figure 9: Location of the optic fibre temperature probes in plexiglas blocks during the RF voltage calibration 43 Figure 10: Location of sections (S) for the determination of the initial moisture content of each dielectric specimen 45 Figure 11: Location for optic fibre temperature probes in each dielectric specimen 48 Figure 12a: Time rate of temperature change for the voltage calibration 52 Figure 12b: Time rate of temperature change for the voltage calibration 53 Figure 12c: Time rate of temperature change for the voltage calibration 54 Figure 13: Calibration curve for rms voltage 57 Figure 14: Plot of loss factor against moisture content for western hemlock sapwood at 0.8 kV (top) and 1.1 kV (bottom) 71 Figure 15: Plot of loss factor against moisture content for western hemlock heartwood at 0.8 kV (top) and 1.1 kV (bottom) 72 Figure 16: Plot of loss factor against moisture content for western red cedar heartwood at 0.8 kV (top) and 1.1 kV (bottom) 73 vi Figure 17: Plot of published loss factor data (Torgovnikov, 1993) for various moisture contents, two temperatures and three densities 77 Figure 18: Plot of loss factor against temperature for western hemlock sapwood at 0.8 kV (top) and 1.1 kV (bottom) 79 Figure 19: Plot of loss factor against temperature for western hemlock heartwood at 0.8 kV (top) and 1.1 kV (bottom) 80 Figure 20: Plot of loss factor against temperature for western red cedar heartwood at 0.8 kV (top) and 1.1 kV (bottom) 81 Figure 21a: Plot of loss factor against voltage for western hemlock sapwood at 25 °C (top) and 35 °C (bottom) 84 Figure 21b: Plot of loss factor against voltage for western hemlock sapwood at 45 °C (top) and 55 °C (bottom) 85 Figure 22a: Plot of loss factor against voltage for western hemlock heartwood at 25 °C (top) and 35 °C (bottom) 86 Figure 22b: Plot of loss factor against voltage for western hemlock heartwood at 45 °C (top) and 55 "C (bottom) 87 Figure 23 a: Plot of loss factor against voltage for western red cedar heartwood at 25 °C (top) and 35 "C (bottom) 88 Figure 23b: Plot of loss factor against voltage for western red cedar heartwood at 45 °C (top) and 55 °C (bottom) 89 Figure 24a: Plot of loss factor against moisture content for the three species at 0.8 kV, 25 °C (top) and 35 °C (bottom) ...93 Figure 24b: Plot of loss factor against moisture content for the three species at 0.8 kV, 45 °C (top) and 55 °C (bottom) 94 Figure 25a: Plot of loss factor against moisture content for the three species at 1.1 kV, 25 °C (top) and 35 °C (bottom) 95 Figure 25b: Plot of loss factor against moisture content for the three species at 1.1 kV, 45 °C (top) and 55 °C (bottom) 96 Figure 26: 3-D plot of the loss factor for western hemlock sapwood at 0.8 kV 100 vii Figure 27: 3-D plot of the loss factor for western hemlock sapwood atl.lkV 100 Figure 28: 3-D plot of the loss factor for western hemlock heartwood at 0.8 kV 101 Figure 29: 3-D plot of the loss factor for western hemlock heartwood at 1.1 kV 101 Figure 30: 3-D plot of the loss factor for western red cedar heartwood at 0.8 kV 102 Figure 31: 3-D plot of the loss factor for western red cedar heartwood at 1.1 kV 102 viii Acknowledgements My sincere thanks goes to Dr. Stavros Avramidis, my supervisor, for his academic and research guidance during my graduate studies, and to my co-supervisor, Dr. Sawas Hatzikiriakos, for his valuable advice and help. I am grateful to Drs. Bruce Neilson and Simon Ellis for their helpful comments, suggestions and critical review of my thesis. I am also grateful to Mr. Jan Aune and MacMillian Bloedel Research for providing the wood used in this study and to the Natural Sciences and Engineering Research Council (NSERC) of Canada for providing financial support through a Strategic Grant (STR0167393). I owe a lot of thanks to my friends who are working and studying in the Wood Physics group, for their assistance during my experimental work and their helpful discussions. Finally, a special thanks is due to my wife, Manxia, whose encouragement, understanding and patience made the successful conclusion of my graduate studies at UBC, possible. I love her more than she can imagine. ix Chapter 1 1.0 Introduction It is well known that kiln dried lumber has a considerable number of advantages over green lumber. Generally speaking, the former is lighter, stronger, dimensionally more stable and more resistant to fungal degradation. From the economic point of view, drying lumber is a value-adding process. For example, in 1995 the price range for western hemlock green baby squares was about US$432 - 445/m3, whereas kiln dried baby squares would have an added value of 15% (Li, 1996). Therefore, lumber kiln drying is of key importance in the utilization of wood and for that reason, a large volume of research has been carried out to identify more efficient drying methods and technologies. Western hemlock [Tsuga heterophylla (Raf.) Sarg.] and western red cedar [Thujaplicata Dorm] are two of the most important and abundant timber-producing species in B.C. (Hosie, 1976), hence, kiln drying of these two species (as well as of the other local softwoods and hardwoods) is essential to the B.C. forest products industry. At present, conventional drying is the most common method used not only locally, but also globally. Conventional drying is carried out on the basis of convective and conductive heat transfer mechanisms, and it is widely considered as a time- and energy-consuming method because large amounts of heat are released through venting. Furthermore, due to the internal drying stresses developed during the conventional drying process, the drying degrade can become quite significant thus reducing the lumber value. Past research studies have indicated that the problems encountered in conventional drying 1 of western hemlock and western red cedar included uneven final moisture content between and within boards, long drying schedules, presence of "wet pockets" in hemlock, and collapse in cedar, which is the result of its extremely high content of extractives. (Kozlik, 1970; Meyer and Barton, 1971; Avramidis and Mackay, 1988; Li, 1996). These problems were more severe in the case of conventional drying of thick sections and high quality lumber, since there are no proper conventional methods to dry them (Zwick et al.,1995). Dielectric heating of wood, due to its unique features of quick and uniform internally produced heat, has played an important role in wood drying, gluing, impregnating, and pressing (Biryukov, 1961). One of the recent significant applications of dielectric heating in wood drying is the radio frequency/vacuum drying in which, both the pressure and temperature gradients in the lumber to be dried point toward the same direction, i.e., from inside to outside, consequently causing the moisture to be driven out from the lumber much faster when compared to conventional drying. Previous studies revealed that the RF/V drying seemed to be a promising method for the hard to dry species and for the thick lumber that is almost impossible to dry with conventional methods (Zwick et al.,1995; Avramidis and Zwick, 1996). In order to develop and optimize RF/V drying schedules, a better understanding of the dielectric properties of wood to be dried, as a function of various species characteristics or properties is required. Another benefit of investigating the dielectric properties of wood is that the knowledge could help us to further understand the micro- and macro-structure of wood and the mechanisms of interaction between water and wood (James and Hamill, 1965; Norimoto, 1976; Cao et al., 1986). Therefore, the study of dielectric properties of wood has both theoretical and practical significance. 2 Wood under a radio frequency electromagnetic field reveals its dielectric properties, which are characterized by three parameters: dielectric constant, loss tangent and loss factor. The loss factor is the product of the dielectric constant and loss tangent. A considerable amount of past research work has confirmed that the following wood related variables have major effects on its dielectric properties: moisture content, wood density, temperature, frequency, and grain direction (Skaar, 1948; Lin, 1967; Nanassy, 1970; James, 1975; Torgovnikov, 1993). Over the frequency range of 0 - 300 MHz, the measurement techniques of the dielectric properties of wood used by researchers can be classified into two categories: the Q-meter and the direct calorimetric data method. Most researchers have used the Q-meter method to measure the dielectric constant and loss tangent of the small wood specimen (Skaar, 1948; Wittkopf and Macdonald, 1949a, 1949b; Vermaas, 1971; James, 1975,1977; Cao et al., 1986; Dai et al., 1989; Avramidis and Dubois, 1993; Torgovnikov, 1993). According to Biryukov (1961) and Torgovnikov (1993), the necessary accuracy of this method is questionable due to the small sample size and volume (2-5 cm3), and due to the so-called "edge effect" which causes changes in the actual capacitance of the capacitor containing the wood sample. The direct calorimetric data method was found to be only used by Biryukov (1961). By using this method, the loss factor and power density can be directly obtained from calorimetric data. For the measurement of full-sized dielectric specimens, this method is more practical. The objective of this study was to investigate the dielectric characteristics of western hemlock and western red cedar, two of the most important local species, by using the direct calorimetric data method. The dielectric specimens were full-sized timbers, as far as thickness and width are concerned, which are being or will be dried in a commercial size RF/V dryer. 3 Chapter 2 2.0 Literature Review 2.1 Reasons for drying wood It has been well recognized that fresh green lumber cut from a log, has a moisture content (M) of more than 30% and therefore, must be dried to a lower moisture content before it can be successfully utilised (Bramhall and Wellwood, 1976). That is because, dried lumber has a considerable number of advantages over green lumber. These advantages include, (1) Dried lumber is dimensionally more stable. Because most shrinkage and distortion have taken place during the drying procedure aiming at the moisture content which lumber will ultimately attain in use, further dimensional changes will be inconspicuous if the environmental conditions are reasonably stable. (2) As lumber dries, most of its mechanical properties improve. (3) Dried lumber is less subject to fungal degradation and insect attack, since wood with moisture contents below 22%> and 10%> is immune to fungal and insect attack, respectively. (4) Dried lumber is lighter, which results in reduction of transportation and handling costs. (5) Further treatments like pressure impregnation with liquids or gases of dried wood will be more feasible. (6) Both finishing and gluing treatments require a low moisture content level in order that the bonding process takes place. From the business point of view, drying lumber is a value-adding process. Considering western hemlock as an example, in 1995 the prices ranged between US$432 - 445/m3 for green baby squares, that is the lumber with the dimension of 105 x 105 mm in cross section and of various lengths, whereas kiln dried baby squares would have an added value of 15%> (Li, 1996). 4 Indisputably, both technical and economic requirements make lumber drying a necessity. Western hemlock and western red cedar are two of the most important and abundant timber-producing species in B.C. (Hosie, 1976). Due to its light colour, straight grain and fairly high strength, western hemlock has been used in general construction, and the increasing popularity of hemlock baby squares used in Japanese traditional housing construction has increased demand (Li, 1996). Western red cedar wood is very light, soft, and relatively low in strength. Another unique characteristic is its high level of extractives, which result in its characteristic colour, odour and decay resistance (Holmes and Kozlik, 1989). It is thus recognized as a kind of valuable material for poles, posts, boat-building, green house construction, doors, window sashes, and interior finishing. Therefore, kiln drying of western hemlock and western red cedar is of key importance to the B.C. forest products industry. 2.2 Conventional and RF/V drying Presently, conventional drying, which is also referred to as "heat-and-vent" drying, is the most common method used to dry softwood and hardwood lumber. Conventional drying is carried out on the basis of convective and conductive heat and mass transfer mechanisms. The circulating hot air in the kiln through lumber banks separated by stickers provides sufficient thermal energy to the surface of lumber by convective heat transfer which is then transferred to its geometric centre by conduction. Water that evaporates from the lumber is removed by the circulating air. Conventional drying is widely considered as a time- and high energy-consuming method. Furthermore, due to the internal drying stresses that develop during this process, the drying degrade could be considerable, thus reducing the lumber value. Past research has shown that the most common problems encountered in conventional drying of western hemlock and western red cedar were uneven final moisture content between and within boards, long drying schedules due to the so-called "wet pocket" existence, and collapse, which is result of the extremely high content of extractives. (Kozlik, 1970; Meyer and Barton, 1971; Avramidis and Mackay, 1988; Li, 1996;). Attempts have been made, using upgraded kiln drying schedules such as high temperature schedules, to alleviate the uneven final moisture content problem. However, the results were unsatisfactory, although the drying rates increased slightly (Guernsey, 1957; Kozlik, 1970). Presteaming and conditioning after conventional drying seemed to be effective methods for reducing the large variation of final moisture content, but obviously, they prolonged the total drying time (Avramidis and Mackay, 1988; Avramidis and Oliveira, 1993). These problems appeared to be more pronounced during conventional drying of thick and high quality western hemlock and western red cedar timbers. Today, there are not appropriate conventional methods to dry them (Zwick et al., 1995). Russian scientists were the first to suggest and implement dielectric heating of wood in industrial processes. Due to its unique features of quick and uniform internally produced heat, dielectric heating has played an important role in wood drying, gluing, impregnation, and pressing (Biryukov 1961). Especially in the field of wood drying, research works have been carried out on the dielectric properties of wood, the technical feasibility of dielectric heating for drying and the effects of dielectric drying on its physical and mechanical properties. One of the recent significant applications of dielectric heating is the RF/V wood drying, which is carried out in a high vacuum kiln, where energy, in the form of RF waves is transmitted to lumber through electrode plates (Harris and Lee, 1985; Trofatter et al., 1986; Avramidis and Zwick, 1992; Avramidis et al., 1994). In RF/V drying of wood, both the pressure and temperature gradients in the lumber to be dried point toward the same direction, i.e., from inside to outside, consequently causing the moisture to be driven out from the lumber much faster when compared to conventional drying. Furthermore, the final moisture content distribution between and within lumber pieces is more uniform because of the unique characteristic of selective dielectric heating of the wetter areas of wood (Miller, 1966; Miller, 1969; Schiffmann, 1987). Although some practical problems exist at present, thus requiring further studies, RF/V drying seems to be a promising method for the hard-to-dry species and for the thick lumber that is almost impossible to dry with conventional methods. Pound (1966) conducted a study of drying green obeche to 22% final moisture content in an RF dryer connected to a generator with the output of 24 kW. The final moisture content distribution was found to be uniform both in the longitudinal direction and throughout the thickness of the lumber. Miller (1971, 1972) dried wood with an RF process combined with kiln drying. The dry-bulb, wet-bulb and the wood surface temperatures were measured in order to control the conditions in the kiln. In this process, the drying rate was increased considerably without degrade. Simpson (1980) reported drying of short lengths of northern red oak lumber (25 mm thick, 100 mm wide, and 600 mm long) by a 5 kW RF kiln. The results of this study revealed that severe honeycomb occurred after a 15-minute drying cycle. Harris and Lee (1985) conducted a study to evaluate selected physical and mechanical properties of white pine lumber (100 mm thick and 2.4 m long) dried by a RF/V dryer and 7 compared them to lumber dried in a conventional kiln. Conventional drying of white pine from a moisture content of 146% down to 8% required 6 weeks whereas RF/V drying required only 54 hours. Static bending, shear, and hardness showed no significant differences between lumber dried by conventional and RF/V processes. A similar study was also carried out by Lee and Harris (1984). Matched pairs of red oak lumber, 100 mm thick and 2.1m long, were dried by RF/V and dehumidification processes. It took about 35 days to dry the lumber from green to 8% moisture content in the dehumidification kiln and only 60 hours in the RF/V kiln. The RF/V drying resulted in red oak lumber with slightly but significant lower specific gravity, equilibrium moisture content, and compressive strength than the lumber dried by the dehumidification process. Red oak lumber dried by RF/V and dehumidification showed no significant differences in static bending, shear, and hardness. This was in good agreement with the results reported by Harris and Lee (1985). Harris and Taras (1984) reported drying of 100 mm thick, 200 mm wide, and 2.4 m long red oak lumber in a RF/V kiln at an ambient pressure of 20 mmHg, and a conventional kiln. The ratio of the drying rates by RF/V and conventional was 1/17. Although the moisture distribution and stress patterns were similar throughout drying for both methods, the conventional dry kiln process resulted in the lowest moisture gradient across the board thickness, and the lowest stress level. Total shrinkage of lumber dried in the RF/V kiln was about 30% less than the shrinkage of lumber dried in the conventional kiln. In order to evaluate if subsequent humidity cycling of the red oak lumber would result in different swelling and shrinking patterns for lumber dried by the above two methods, Harris (1988), dried red oak and eastern white pine in another study. The results revealed that no 8 significant differences could be found in swelling and shrinking of red oak or white pine when exposed to subsequent humidity cycling. Avramidis and Zwick (1992) reported drying Pacific coast hemlock, western red cedar and Douglas-fir of different sizes and grades at a pressure of 20 to 25 mmHg with a 23 m3 stainless steel commercial RF/V kiln, which was connected to an RF generator that had a power output up to 260 kW at a frequency of 3 MHz, and a maximum power density of 11.6 kW/m3. Evaluation of the dried lumber showed that the three species could be dried in short time, and the amount of degrade was low with the exception of clear cedar, in which severe honeycomb developed. This study also showed an advantage of no discolouration of lumber dried by RF/V. Two years later, six drying runs with 91 by 91 mm in cross section and 2.24 m long western red cedar heartwood lumber were carried out in a laboratory RF/V dryer by Avramidis et al. (1994). They concluded that thick western red cedar could be dried in an RF/V dryer and at a constant electrode voltage of 0.8 kV in 23 hours without checking, collapse and discolouration. Recently, Avramidis et al. (1996a, 1996b, 1997) reported the results of an in-depth study addressing the philosophical approach in designing a commercial RF/V kiln, the investigation of RF/V kiln-drying of three B.C. softwood lumber of large sections, namely, Douglas-fir, Pacific coast hemlock and western red cedar, and the pertinent capital and operating costs. The results revealed that with proper schedules, RF/V drying resulted in no lumber stain, reduced surface checking, no internal stresses, and a uniform final moisture content distribution when compared to conventional drying. Furthermore, RF/V technology allowed for the drying of lumber sizes that were not dried in conventional kilns due to excessive drying times and degrade. Also, RF/V drying costs were more competitive to those for conventional kilns for lumber 101 mm and 9 thicker, with at least a 15 to 25 % cost advantage depending on species and grades. All the studies mentioned above concluded that further research is required in order to develop optimum RF/V drying schedules, which will maximize drying rates and minimize drying defects. This can be achieved in minimum time and effort by developing mathematical models that can predict the drying history of a particular size and species lumber. These models will include the typical heat and mass fluxes as in conventional drying with the addition of an extra factor for volumetric heat generation. The latter is a function of the wood's dielectric properties and the RF field's characteristics. Therefore, the importance of better understanding the dielectric properties of wood to be dried, as a function of various variables is also obvious, since it is necessary to tune the RF generator's circuit to the continuously changing dielectric properties of the wood load, thus maintaining a high power transmission and conversion efficiency (James, 1983; Zwick et al., 1995). As a matter of fact, there has been an increasing interest in the study of wood dielectric properties as a result of the increased utilization of radio frequency and microwave techniques in wood industry processes. Another benefit of investigating the wood dielectric properties is that such knowledge could allow us to better understand the micro- and macro- structure of wood and the mechanisms of interaction between water and wood (James and Hamill, 1965; Norimoto, 1976; Cao et al., 1986). Therefore, the study of dielectric properties of wood has both theoretical and practical significance. 10 2.3 Electromagnetic waves It is generally accepted that dielectric heating is performed at the frequency of 1 to 100 MHz, whereas microwave heating occurs between 300 MHz and 300 GHz. The industrial, scientific, and medical (ISM) bands established by international regulations allow a number of frequencies to be used for industrial heating. The most common ones are 13.56, 27.12, 40.68, 915, and 2450 MHz (Metaxas and Meredith, 1983). According to SchifFmann (1987), all electromagnetic waves are characterized by their wavelength and frequency, and consist of two components, namely, electric (E) and magnetic fields (H) as shown in Figure (1). Note that the electric and magnetic fields are perpendicular to each other and are both perpendicular to the travelling direction (X). The velocity (C) of the wave travelling in the X-direction in air or vacuum will be reduced as it passes through another medium, as indicated in Eq. (1): C' where; vp is the velocity of propagation, m/sec; C is the velocity of the wave in air or vacuum, 3 x 108 m/sec; and e' is the dielectric constant of the material through which the wave is propagated, as will be seen later. The relationship of the wavelength (k) and the frequency (f) could be described as: f=^ (2) 11 z Figure 1. Electromagnetic wave. 12 where, f is the frequency, Hz; and k is the wavelength, m. As an electromagnetic wave passes through a material, its frequency remains the same; therefore its wavelength changes, and this affects the depth of penetration (D). Although not a property of a material, penetration depth is of utmost importance, since electromagnetic heating such as dielectric heating, is bulk heating. Hence it is important that the energy penetrates as deeply as possible. The penetration depth is expressed as (Schiffmann, 1987): D=Ml ( 3 ) 27T6" V ' where, e"is the loss factor of the material, as will be seen later. An electromagnetic wave is an energy wave that changes its energy content and amplitude as it travels through a medium. As indicated in Figure (1), the electric or magnetic component is zero at some point, then it builds up to a maximum value, decays to zero, and again builds up to a maximum value with the opposite polarity before again decaying to zero. It is this periodic flip-plopping of the wave's polarity and its decay through zero that causes the stress upon ions, atoms and molecules, which is converted to heat. 2.4 Wood: a dielectric material From the practical standpoint, materials may be classified according to their electrical conductivities into three categories (Zaky and Hawley, 1970): (1) conductors - conductivity 106 to 108 ohm'W1; 13 (2) semiconductors - conductivity 10s to 10"7 ohm^ m"1; (3) dielectrics or insulators - conductivity 10"8 to 10"20 ohm^ m"1. The conductors are characterized as materials, such as metals, with a high electronic and ionic conductivity. Under the influence of an electric field, one or more electrons in a metal atom can readily be removed from or added to the outermost shell of orbiting electrons, therefore, metals are considered as good conductors of electricity. The dielectrics differ from the conductors by their relatively low electrical conductivity. Unlike conductors, dielectrics are materials with such an atomic and molecular structure that their outer shells are more stable, and as a consequence, there are very few free electrons to form an electric current. Although conduction might take place in dielectrics, it is on such a small scale that these materials are still recognized as dielectrics or insulators (Vermaas, 1971). According to Torgovnikov (1993), oven-dry wood with an electric conductivity in the range from 10"13 to 10"15 ohm^ m"1 is classified as a polar dielectric. Wood is also a hygroscopic material, and the electric conductivity of wood increases with increasing moisture content (Skaar, 1948; Hearmon and Burcham, 1954; Brown et al., 1963; Lin, 1965). When moisture content is above the fibre saturation point, ionic conductance may also take place within wood. 2.5 Types of polarization Under a high frequency alternating electric field, there are two types of heating processes within a material - inductive and dielectric heating. In inductive heating, when a coil surrounding the material which is good electrical conductor carries an alternating electric current, the material 14 inside the coil will become hot due to the development of eddy currents on the surface of the material. In dielectric heating, the heat is produced within the materials itself, when a dielectric is placed in a high frequency alternating electric field, which usually is formed by two parallel electrodes connected to the source of high frequency energy produced by a high frequency generator. The mechanism of heat generation in a dielectric material is mainly based on polarization effects (Tinga and Nelson, 1973). When an electric field is applied to a dielectric material such as wood, the positive and negative charges in the atoms and molecules of the material are displaced. This phenomenon is called polarization. According to the nature of the charges displaced, Zaky and Hawley (1970) considered the total polarization in any material being made up of four different components. These are, (1) Electronic polarization. This polarization is due to the displacement of electrons with respect to positive nuclei in the atoms under the influence of an external electric field, and it is connected to the heat dissipation at only very high frequencies corresponding to light waves. (2) Ionic or atomic polarization. Ionic polarization is only found in ionic substances, and it is characterized by the shift of the relative positions of the positive and negative ions of a molecule. Like electronic polarization, ionic polarization may result in heat dissipation only at very high frequencies. (3) Dipole or orientational polarization. This kind of polarization is associated with dipolar substances, in which the molecules possess dipole moment even in the absence of an external electric field. The total moment is zero as a result of random thermal agitation. Under the influence of the external electric field, the dipoles experience a torque that tends to orient them in 15 the direction of the field. Both water molecules and macromolecules such as cellulose, hemicellulose and lignin, which make up the cell wall substance in wood, are subject to dipole polarization (Torgonikov, 1993). (4) Interfacial polarization. Interfacial polarization is characterized by an accumulation of free ions at the interfaces of dissimilar compounds possessing different electrical conductivities. All four types of polarization require a certain period of time, known as relaxation time, to occur. As a result, the polarization never follows exactly an electric field. Electronic and ionic polarizations are of major importance only in the range of frequencies corresponding to the infrared radiation and visible light. At audio and radio frequencies, dipole and interfacial polarizations are the most important. 2.6 Dielectric properties of wood According to Torgovnikov (1993), wood under an RF electric field reveals its dielectric properties, which are characterized by three parameters: dielectric constant or relative permittivity (e'), loss tangent (tan5) and loss factor (e"). (1) Dielectric constant (e') It is well known, in the field of electrostatics, that the capacitance of a capacitor is increased if the space between the conductors is filled with a dielectric material. If C 0 is the capacitance of the capacitor with a vacuum present between the conductors, and C its capacitance when the medium is a dielectric material, then the ratio 16 (4) is called dielectric constant of the medium, and its value is independent of the dimensions of the conductors. This was first discovered by Faraday from his experimental results (Zaky and Hawley, 1970). Although the dielectric constant could also be defined as the ratio of field strengths between conductors with vacuum and dielectrics, in practical measurements the determination of the dielectric constant is based on Eq. (4) (Vermaas, 1971). According to Vermaas (1971), if the capacitor is connected to an alternating voltage source, the meaning of the dielectric constant may be explained as a measure of the electric displacement produced in a dielectric by a given electric stress or the energy stored in a dielectric when an external alternating electric field is applied to it. The amount of energy stored in the dielectric material is related to its dielectric constant, and the greater the polarization of the dielectric material, the greater will be the dielectric constant (Peyskens et al., 1984). (2) Loss tangent (tanb) Consider the simplest situation that a high frequency root-mean-square (rms) voltage V is applied to the two parallel electrodes between which a dielectric material is placed. If the material is a perfect dielectric, the polarization is instantaneous resulting in that the displacement or capacitive current Ic leads TT/2 radians out of phase with the voltage V (Figure 2). In this case, there is only a small power, called conduction loss, being produced in the dielectric. If the material is an imperfect dielectric, it can be represented electrically by a RC circuit, shown in Figure (3), consisting of capacitor C and resistor R in parallel, and there may appear a loss or conduction 17 Electrode Perfect dielectric Electrode Ic jt/2 Figure 2. Current - voltage relation in an ideal capacitor. 18 I I R Ic R Figure 3. RC circuit for dielectrics (wood). - 19 it current IR which is in phase with the voltage. Therefore, the total current I flowing through the system is inclined by an angle 0 (< 7r/2) against the rms voltage E and by an angle 8 against the displacement current. The angle 0 is called the dielectric phase angle and 8 is called the loss angle (Figure 4).The loss tangent, another important dielectric property parameter, is defined as the ratio of loss current to capacitive current, expressed as: tan8=^ (5) For practical purposes, the loss tangent can be regarded as an index of the fraction of the power dissipated by the dielectric material as heat (Lin, 1967). (3) Loss factor (e") In order to determine the power dissipated in the dielectric material that is under the influence of a high frequency electric field, the loss factor (e") is defined as the product of loss tangent and the dielectric constant. Mathematically, it is described as (Pound, 1973): e'Wtairi (6) It is noticed that these three important dielectric parameters, namely, e', tanS and e" are dimensionless since the former two are ratios of capacitances and currents, respectively. 2.7 Power density The power loss in unit volume of a dielectric material such as wood under the influence of 20 e I R Figure 4. Vector diagram of dielectrics (wood). 21 an external high frequency electric field is known as power density (PD). As indicated in Figures (3) and (4), wood as a dielectric material in an AC circuit could be represented as a capacitor C and resistor R in parallel. The current I consists of Ic and IR through the capacitance and resistance, respectively. Affected by the AC rms voltage V, the molecules in wood undergo cyclic motions that result in friction. It is this molecular friction that causes power loss within the wood, leading to a rise in temperature. According to Vermaas (1971), the power (P) released in the circuit can be described as: P=VIR-VIcos0=— (7) R where, V is the rms AC voltage, V; IR is the loss current, A; I is the resultant current, A; 8 is the phase angle; and R is the resistance, ohm. The reactance of the capacitor is given by: where, Xc is the capacitive reactance, ohm; f is the frequency, FIz; and C is the capacitance, F. The capacitance of the capacitor can be calculated from: „ z A C=e /e0- (9) where, e' is the dielectric constant; e0 is the permittivity of free space, 8.854 x 10"12 F/ m; A is the area of the electrode, m2; and d is the distance between two electrode plates, m. From Eqs.(8) and (9), reactance Xc can be expressed as: 22 (10) By considering I T * 5 <») we obtain 1 tan6 27rfe/e0Atan8 R X c The specific AC conductivity o is defined as: where, o is the AC conductivity, ohm -m . From Eqs. (7) and (13), P=V2a—=27Tfe/entan5V2— d 0 d 23 (12) a=-^-=27rfe/e„tan8 (13) RA 0 K ' (14) Therefore, the power density PD (W/m3) can be calculated as: PD=(5.56xl0"11)E2fe/ (15) where, E (V/d) is the field strength, V/m. 2.8 Time rate of temperature change If heat loss due to the moisture evaporation is not taken into account, and if it is assumed that there is no chemical reaction inside wood, the equation to calculate the time rate of temperature change (dT/dt) resulting by the conversion of high frequency energy from the AC electric field to heat can be derived. According to Gebhart (1993), the general conduction equation is obtained by equating the sum of the net rates of energy gain by conduction and generation to the time rate of increase of stored energy. The resulting equation is d dT s, d /, dT •. d 3TN /// dT where, T is the temperature, °C; k^, ky and kz are the thermal conductivity in the three principle directions of x, y and z, respectively, W/m °C; q'" is the rate of energy generation at x, y, z at time t per unit volume of the material, W/sec m3; p is the density, kg/m3; cp is the specific heat, J /kg °C; and dT/dt is the time rate of temperature change, °C/sec; The first three terms of the equation's left hand side are net rates of energy gain in the element (dxdydz) due to conduction in the three principle directions of x, y and z, respectively. Since the thermal energy, during dielectric heating, is generated inside the material, there is no heat conduction occurring. Therefore, these first three terms are reduced to zero. Moreover, q"' is the rate of energy generation at x, y, z at time t per unit volume of the material, in the case of dielectric heating, it can be considered as equal to the power density. Therefore, under the influence of an AC electric field the time rate of temperature change inside wood is 24 mathematically described as: 8T PD a pc 07) 2.9 Influence of wood variables on its dielectric properties A considerable amount of past research work has shown that the following variables have a major effect on the dielectric properties of wood. (I) Moisture content Wood is a hygroscopic material and nearly all its physical and mechanical properties are affected by its moisture content. Certainly, there is no exception to dielectric properties of wood. The moisture present in wood is in the form of bound water or/and free or capillary water. Bound water is physically adsorbed by the hygroscopic polymers like cellulose and hemicelluloses comprising the cell wall and free water is situated in the voids (i.e., lumens) within gross wood (Siau, 1995). It is understandable that under the external high frequency electric field, both bound and free water change their own dielectric properties and in turn affect the dielectric properties of moist wood. One of the early studies conducted by Skaar (1948) showed that the variation in the moisture content had an overwhelming effect on the dielectric constant and loss tangent of wood. In that study, conducted at room temperature and a frequency of 2 MHz, the dielectric constant of buckeye wood increased from 1.8 when oven-dry to 34.4 at a moisture content of 114%. Skaar 25 attributed this to the higher dielectric constant value of water, which is approximately 81, compared with the lower dielectric constant of oven-dry wood, which is in the range of 2.0 to 4.2, at the frequency of 2 MHz. Close inspection of the relationship between the dielectric constant and moisture content indicated that below the fibre saturation point the dielectric constant varied as an exponential ratio with the moisture content, but above the fibre saturation point a linear relationship was held. Similarly, the loss tangent of thirty wood species measured at 2.0 and 15.0 -MHz increased with moisture content, which was in the range of oven-dry to 16%. Venkateswaran and Tiwari (1964) studied the relationship between dielectric properties of ten species of indigenous wood and moisture content at the frequencies of 1, 2, 5 and 10 kHz. The results showed that while the dielectric constant linearly increased in all cases, the loss tangent displayed the existence of a maximum followed by a minimum in some cases. In order to quantitatively describe the dielectric properties, equations for dielectric constant and loss tangent were derived by the researchers assuming the dielectric to be a mixture of two components - dry wood and water. To be able to use the derived equations to calculate the dielectric constant and loss tangent, the static and optical dielectric constants of the individual components of the mixture and the relaxation time should be known. Because of the approximate nature of the equations, the experimental and calculated values of the dielectric constant and loss tangent of moist wood, though were not always in quantitative agreement, appeared to agree qualitatively. Hearmon and Burcham (1954) measured the dielectric constant and loss tangent of twelve species of wood at moisture contents up to 18% and over the frequency range from 2 kHz to 60 MHz. They also reported that the dielectric constant increased with the increasing moisture content. Whereas the loss tangent, while showing a steady increase generally, displayed a 26 maximum followed by a minimum under some conditions. In their two studies on the dielectric properties of Douglas-fir and ponderosa pine, at the frequencies between 2 and 40 MHz, Wittkopf and Macdonald (1949a, 1949b) summarised that the dielectric constant showed a large increase as the moisture content was increased from oven-dry to 15%, and the loss tangent in general showed a lesser increase as the moisture content increased in this range. James (1975) also agreed that the essential character of the dielectric properties of wood are strongly influenced by moisture content. In his comprehensive study aimed to investigate the dielectric properties of white oak and Douglas-fir at frequencies from 20 Hz to 50 MHz, moisture contents from oven-dry to complete saturation, and temperatures from -20 to +90 °C, the author found that the dielectric constant increased with increasing moisture content provided the other conditions being the same. The values ranged from 2.0 for cold, dry wood at high frequency to near 1 x 106 for warm, water-soaked wood at low frequency. Again, loss tangent showed maximum and minimum values under various circumstances. (2) Wood density Both theoretical and experimental work indicated that the dielectric properties of wood was influenced by wood density (Skaar, 1948; Yavorsky, 1951; Vermaas, 1971; Torgovnikov, 1993). Since the dielectric constant of air is approximately 1.0, whereas it is in the order of 4.5 for wood substance, there would be a definite relationship between dielectric constant and density of wood (Yavorsky, 1951). Peyskens et al. (1984) reported that a positive linear relationship existed between density and the dielectric constant, and that this relationship became more pronounced as the moisture 27 content increased. The influence of the wood density on the loss factor appeared to be dependent on the moisture content. The positive relationship was found at the lower moisture content, but at the highest moisture content (30 - 40%) a negative relationship was observed. According to Skaar (1948), the positive relationships could also be found between both dielectric constant and loss tangent, and the density of wood at frequencies of 2 to 15 MHz. On the other hand, similar results were obtained from the studies conducted by Wittkopf and Macdonald (1949a, 1949b) and Dai et al. (1989), but both of them pointed out that the relationship between the loss tangent and density was not so clear. Contrary to the studies mentioned above, Lin (1973) concluded that wood density had little effect on its dielectric properties. Since only one wood species, western hemlock, was used in his study, the author reasoned that the variability due to the different proportions of cell-wall substances could overshadow any possible effect of density on the dielectric properties. (3) Temperature Temperature is also considered as one of the variables affecting the dielectric properties of wood. Lin (1967) in his review of the dielectric properties of wood and cellulose summarised that the dielectric constant increased with increasing temperature, and the effects of temperature on dielectric properties were more pronounced when the moisture content of wood was high. The equation, for the calculation of the dielectric constant of moist wood that he quoted from his unpublished thesis showed this trend. Later, supporting Lin, Nanassy (1970) observed that the dielectric constant increased with an increase of temperature for end-grain orientation at frequencies of 30 Hz, 1 kHz, 100 kHz, 10 MHz and 1 GHz. He speculated that the dipolar groups were bound to the solid structure in such 28 a way so that the dipole was a structural element of the solid lattice, and that the rigidity of lattice hindered the reorientation of dipoles. Therefore, an activation energy was required to make the reorientation possible. With an increase of temperature the reorientation of dipoles became more feasible thus, the dielectric constant increased. The dependence of loss tangent on the temperature is more complex than that of the dielectric constant, since loss tangent can increase or decrease with a decrease in temperature at a given frequency depending on whether the frequency is below or above that corresponding to the relaxation time (Siau, 1995). Yavorsky (1951) stated that since an increase in temperature leaded to increased conductivity, which in turn gave rise to increased conduction losses so that the value of the loss tangent became larger, the higher temperature resulted in higher loss factor. This conclusion appears not to be true at all times, because the nature of dependence of the dielectric properties of materials is a function of the dielectric relaxation processes (Nelson and Kraszewski, 1990). Lack of consideration of the dielectric relaxation will result in unrealistic conclusions. (4) Frequency The relationship between dielectric properties and frequency is somewhat complex. Torgovnikov (1993) analysed the experimental data of the dielectric properties of oven-dry wood obtained by various researchers for more than 60 species. His results indicated that at room temperature, the dielectric constant of oven-dry wood decreased with increasing frequency that was in the range of 10 Hz to 105 MHz. He attributed this to a lesser part of the polar wood molecules having sufficient time to follow the external field changes. The loss tangent and loss factor increased with a increase in frequency, and after reaching the maximum at the 10 MHz they 29 started to decrease. These trends were qualitatively similar to the dispersion and absorption curves (Figure 5) representing the Debye mode for a polar substance with a single relaxation time, which is quoted by Nelson and Kraszewski (1990). In Figure (5), co = 27tf is the angular frequency, es and e„ represent the static dielectric constant, i.e. the value at zero frequency, and the dielectric constant at frequencies so high that molecular orientation does not have time to contribute to the polarization. Thus, the dielectric constant, at frequencies very low and very high with respect to the polar molecule relaxation process, has constant values, es and e„, respectively. For oven-dry wood, the very low and very high frequencies respectively are 10 Hz and 105 MHz. At intermediate frequencies, the dielectric constant undergoes a dispersion, and dielectric losses occur with the peak loss at the relaxation frequency, which is 10 MHz in the case of oven-dry wood. In the case of moist wood, the frequency effect is even more complex. The experimental data from the study carried out by James (1975) showed that there was a little decrease in dielectric constant when moisture content was below 12%. At higher moisture contents, the highest value of the dielectric constant of approximately 106 wasfound at 20 Hz for completely soaked wood. For the loss tangent, the maximum shifted to a lower frequency range (1 -10 kHz). It is evident that the relaxation time decreases with an increase in moisture content. (5) Grain direction Wood is not only a hygroscopic material, but also an anistropic one. When other conditions such as moisture content, density, temperature and frequency are the same, the dielectric properties in the three principle directions are different. If e'L, e'R, e'T, and tan8L, tan8R, tan8T represent dielectric constant and loss tangent when the alternating electric field strength E is 30 logo) Figure 5. Dispersion and absorption curves representing the Deby model. 31 parallel to the longitudinal, radial, and tangential directions of wood, respectively, the following relations apply: e'L > e'R > e'T, and tan8L > tan5R > tan5T. In these relationships, the difference of dielectric properties between the radial and tangential directions is less significant. A number of researchers (Skaar, 1948; Hearmon and Burcham, 1954; Vermaas, 1971; Lin, 1973; Torgovnikov, 1993) have agreed with the aforementioned conclusion, but there are still some controversial explanations for the mechanism of the effect of the grain direction on the dielectric properties. Skaar (1948) believed that the difference between parallel-to-grain and perpendicular-to-grain direction dielectric constants was attributable to the microstructure of cell wall. Torgovnikov (1993), agreed with Skaar and theorised that because the polar groups of cellulose and hemicellulose had more freedom of movement along the fibre direction (longitudinal direction) than in the transverse direction, the value of the dielectric properties would be larger when field strength was parallel to the longitudinal direction. The larger values of dielectric constant and loss tangent when the field strength oriented in the radial direction relative to those in the tangential direction was due to the presence of the rays along the radial direction. However, as Lin (1973) quoted, Kroner and Pungs (1953), Uyemura (I960), and Rafalski (1966) suggested that these differences resulted from cell wall orientation rather than from molecular structure of the cell wall. In summary, all of the described variables have considerable effects on the dielectric properties of wood and they have been investigated thoroughly by a lot of researchers. But there are still some factors, such as species and electric field strength, need to be further studied. 32 2.10 Methods of measuring wood dielectric properties Regarding the measurement techniques of the dielectric properties of wood, no standards for this purpose have been established (Torgovnikov, 1993). At the frequency range of 0 - 300 MHz, the measurement techniques used by researchers involved in studying the dielectric properties of wood can be classified into two categories: (1) Q-meter or bridge method Most researchers used this method to measure the dielectric constant and loss tangent under different temperatures and frequencies (Skaar, 1948; Wittkopf, 1949a, 1949b; Vermaas, 1971; James, 1975,1977; Cao et al., 1986; Dai et al., 1989; Avramidis and Dubois, 1993; Torgovnikov, 1993). According to Biryukov (1961) and Torgovnikov (1993), the necessary accuracy of this method is questionable due to the small sample size (2-5 cm3) and the so-called edge effect which causes changes in the actual capacitance of the capacitor containing the wood sample. A schematic of Q-meter measuring circuit as described by Brown et al. (1947) and Skaar (1948), is shown in Figure (6). This type of instrument contains a radio frequency generator (G) which produces a constant current into the resonate circuit through the resistor (r) and a voltage meter (V^ (for indicating resonance). The resistance is so small that it is negligible. In order to simplified the analysis of the circuit, the RF generator is neglected as well. When the connections (A and B) are open, a parallel resonate circuit consisting the inductor (L) and the calibrated capacitor (CjJ exists. When the connections are closed, it is necessary to decrease the capacitance of Cm, to retune the circuit to resonance. This decrease in the reading of the calibrated capacitor 33 -a o-A Figure 6. A typical Q-meter measuring circuit. 34 (C.J is then equal to the capacitance (CJ of the dielectric specimen. That is cx=crc2 (18) where, Cl and C 2 are the readings of the capacitor (C^ at resonance with the connections open and closed, respectively, F. Therefore, the dielectric constant (ej, according to Eq. (9), is given by: d~ <<VC2) (19) x A e m o where, dm is the thickness of the specimen, m; and A,,, is the surface area of the specimen, m2. By definition, the Q of the circuit is equivalent to the ratio of the parallel resistance to the reactance. Therefore, when the connections are open, the value of Q (Qx) is Q1=27TfRmC1 (20) In like manner, when the connections are closed, the measured Q value (Q2) is Q 2 = ^ L ( 2 ^ ( C 2 + C x ) (21) K + K m x Analysis of the circuit associated with the Q-meter indicates that the cotangent of the phase angle (0, see Figure 5) (QJ for the dielectric specimen is Q M c r c 2 ) x ^ ( ( V Q , ) ( 2 2 ) 35 if the value of Qx is greater than 10, the power factor (PF) can be given by: (23) Thus, the loss tangent of the dielectric specimen can be calculated as: tan8= PF Vl-PF 2 (24) (2) Direct calorimetric data method This method was found to be used only by Biryukov (1961). With this method, the loss factor and power density can be obtained directly from the calorimetric data based on Eqs. (15) and (17). For the purpose of practical application in wood drying, the loss factor is more useful for the calculation of the energy dissipated in wood. When the moisture content is above the fibre saturation point, it is difficult to obtain accurate measurements of the dielectric constant, thus, dielectric properties data measured at high moisture content are not so reliable (Skaar, 1948). It is the fact that Q-meter power is not high enough to measure the dielectric properties of wood with very high moisture content, such as the wood specimen saturated with distilled water (Avramidis and Dubois, 1993). However, it seems that there is no such problem in the direct calorimetric data method, because a high frequency generator with a high power can be used in this method, such as the one used in this study. 36 Chapter 3 3.0 Experimental Procedure 3.1 Materials Twenty green pieces of western hemlock (10 were all-heartwood and 10 all-sapwood), and ten green pieces of western red cedar (all-heartwood) timber were obtained from MacMillan Bloedel Research. The western hemlock timbers were 110 mm by 200 mm by 3 m long, and the western red cedar were 100 mm by 200 mm by 3 m long. The initial moisture contents for heartwood and sapwood were around 50% and 85%, respectively. All timbers were chosen based on the requirement that the thickness should be in the radial direction, and be free of surface checking and stain. In order to prevent moisture loss during storage, each timber was tightly wrapped with a waterproof plastic sheet and stored in a cold room where the temperature was set to about 2 °C. 3.2 RF/V dryer In this study, the direct calorimetric data method was used to measure the loss factor and power density for two important B.C. species, namely, western hemlock and western red cedar. The use of a laboratory size RF/V dryer eliminated the requirement of a complicated device such as the ones used by other researchers. The results of this method might be more suitable for a real-life situation and therefore, commercial scale applications. It was not only the size of the 37 equipment that made the difference, but the size of the specimens as well. A laboratory RF/Vdryer had already been fabricated and used in past drying studies. A cross-section schematic of the RF/V dryer is illustrated in Figure (7). The drying chamber consisted of a carbon steel cylinder 2.75 m in length and 0.76 m in diameter with removable bolted caps on both ends in order to facilitate loading and unloading of the lumber, and the interior was coated with an epoxy paint to resist corrosion. The contact surfaces between the cylinder and the caps were fitted with rubber O-rings to avoid air leakage. Two 300 x 2240 x 12.7 mm thick aluminum electrode plates (E) supported by polyethylene bolts (B), were horizontally fixed in the centre of the cylinder. The vertical space between the plates could be adjusted by lifting or lowering the upper electrode plate in order to accommodate the lumber (L) of different thicknesses. The ambient pressure inside the dryer was controlled by a liquid ring vacuum pump (VP) with an air injector attached to the main input line. The water vapour produced during drying or heating was condensed in a heat exchanger (EDE) and collected in a collection tank (CT) . Meanwhile, any condensation accumulated at the bottom of the dryer also was pumped (P) to the collection tank. Therefore, by measuring the water height in the collection tank via a pressure differential transducer, the relationship of the water height and the moisture evaporated from the lumber being dried or heated could be calculated as following: W 0 ( 1 + A M ) h = 7 (25) P w A where, h is the water height in tank, m; A M is the moisture content change, %; W Q is the oven-dry 38 HE Figure 7. Cross-section schematic of the RF/V dryer. 39 weight, kg; pw is the water density, 1000 kg/m3; and A is the area of the collection tank, 0.0850 m 2. During drying or heating, the temperature rise at various locations inside the timber specimen was monitored by fibre optic thermocouples. Al l process data were collected and recorded by a computer through a data acquisition system. The radio frequency generator (RFG) operated at a fixed frequency of 13.56 M H z , and had a maximum output of 10 kW at a maximum electrode voltage of 5 kV. In addition, the RF/V dryer could be automatically shut down as soon as the water height reached a pre-set value. 3.3 Calibration of the RF electrode voltage Due to the standing waves (Biryukov, 1961; Pound, 1973), when an alternating voltage is applied to the long electrodes, the RF voltage increases as the distance from the feed point becomes greater. In the case of this study, the errors of the RF voltage could result from two sources - standing waves and device errors of the rms voltage metre itself. Since the precise value of the rms voltage was essential in the direct calorimetric data method used in this study, the calibration of the RF electrode voltage of the laboratory dryer appeared to be necessary. Plexiglas is a homogeneous material both in physical and chemical properties, and it is also a low loss dielectric material. Due to its homogeneity and dielectric character, two plexiglas blocks were used in the calibration. Each of them was 300 x 200 x 90 mm thick. The whole calibration process was based on Eqs. (15) and (17). When one plexiglas block was placed between the RF/V electrode plates at a certain position, and at a fixed rms voltage 40 (Vr) that was read from the RF voltage metre attached to the dryer, the temperature rise could be monitored by the optic fibre thermocouple with the diameter of 2 mm. Therefore, the calibration of Vc, if the heat loss is neglected, was calculated from Eqs. (15) and (17) as follow: V =dx -=dx \\ 5.56xl0"nxfe"g \] — p c at p g (26) 5.56xl0_11xfe' (tanfi) Where, d is the vertical distance between two electrode plates, 0.09 m; f is the frequency, 13.56 x 106 Hz; dT/dt is the time rate of temperature change, °C/sec; p is the density of plexiglas, 1190 kg/m3 (Lever, 1966); cpg is the specific heat of plexiglas, 1465 J/kg-°C (Lever, 1966); e' g is the dielectric constant of plexiglas, 3.2 (Lever, 1966); and (tan5)g is the loss tangent of plexiglas, 0.025 (Lever, 1966). The calculated rms voltage (Vc) was the calibration rms voltage at the position where the plexiglas block was located. In this study, three positions were chosen to calibrate the rms voltage, namely, voltage input end, centre, and far end of the electrode plate (Figure 8). At each position, the calibration was carried out at six rms voltage levels, namely, 0.4, 0.6, 0.8, 1.0, 1.2 and 1.4 kV. In order to monitor the time rate of temperature change, two holes were drilled on each side of the plexiglas block. Of these four holes, two were 3 mm in diameter and 100 mm deep, and the other two had the same diameter, but were 50 mm in depth (Figure 9). By averaging the calibration rms voltages at these three different positions and six levels of voltages, a linear regression function describing the relationship between the calculated and read rms voltage was obtained. The calibration curves were then used for the dielectric experiment thus, any reference to rms voltage indicated the calibrated values. 41 Far end Centre Input end Figure 8. Location of plexiglas blocks during the RF voltage calibration. 42 Figure 9. Location of the optic fibre temperature probes in plexiglas blocks during the RF voltage calibration. 43 It should be mentioned here that the parameters of the plexiglas used in Eq. (26), such as p, cpg, e'g and (tan6)g were only usable at the room temperature (20 - 25 °C), therefore, the calibration conducted in this study was limited within the temperature range of 20 - 25 °C. In other words, at each calibration experiment, the plexiglas block was heated from 20 °C to 25 °C, which took 10 minutes, approximately. 3.4 Methods 3.4.1 Determination of initial moisture content of specimens (M:) At the very beginning of the study and from each green timber, two 20 mm thick initial moisture content determination sections (S) were cut at a distance of 470 mm from each end. This resulted in a dielectric specimen to be heated that was 2.0 m long (Figure 10). After cutting, these two sections were immediately weighed (w) on a digital balance to 0.01 g and dried in a oven at 103 ± 2°C until constant weight (oven-dry weight, w0). Hence, their initial moisture contents were calculated according to the following formula: w-w M.= -100% (27) wo where, Mj is the initial (green) moisture content of the sample, %; w is the green weight of the sample, g; and w0 is the oven-dry weight of the sample, g. The average Mj value of these two sections was taken as the initial moisture content of the 44 S Dielectric specimen § Figure 10. Location of sections (S) for the determination of the initial moisture content of each dielectric specimen. 45 green dielectric specimen. 3.4.2 Determination of oven dry weight of specimens (Wo) After cutting the moisture content sections, the specimen was weighed (Wj) in order to calculate its total oven dry weight based on the value of M ; : w i W o = - ^ — (28) —-+1 100 where, W Q is the oven dry weight of the specimen, kg; and Wj is the green weight of the specimen, kg. 3.4.3 Determination of average moisture content (Ma) of specimens during each run The dielectric specimen was weighed before and after each run, and the average moisture content (M a ) of the specimen during each run was obtained as: W +W. - 2 W n M =—2 b- ° 1 0 0 % (29) 2 W Q where, M a is the moisture content of the specimen during each run, %; W a is the specimen weight before each run, kg; and W b is the specimen weight after each run, kg. 46 3.4.4 Calculation of average density (p) during each run The average density of the dielectric specimen during each run was calculated from the following equation: W + W H P = ^ T ^ (30) 2v where, p is the density of the specimen during each run, kg/m3; and v is the volume of the specimen, m3. Here, assuming the shrinkage of the specimen during the experiment was negligible, the volumes of western hemlock and western red cedar dielectric specimens were: 0.11 x 0.2 x 2 = 0.044 m3 and 0.10 x 0.2 x 2.0 = 0.04 m3, respectively. 3.4.5 Temperature monitoring and calculation of rate change (dT/dt) Four holes, each 5 mm in diameter and 100 mm in depth, were drilled at equidistant points from the middle of the side face and all the way to the geometric centre of each dielectric specimen. During each run, four optic fibre thermocouples were inserted into these four holes in order to monitor the temperature rise. To make the thermocouples stable inside the holes, one plastic cap, which had a small hole fitting the thermocouple's diameter, was screwed into each hole (Figure 11). 47 Dielectric specimen T - probe T - probe T - probe T - probe Figure 11. Location for optic fibre temperature probes in each dielectric specimen. 48 3.4.6 Determination of specific heat of wood (cp) The specific heat of wood required for the determination of the dielectric parameters was calculated by the following equation (Simpson, 1991): 0.2393 +(0.0003T ) + m c=4187x 1 ?!—± ( 3 1 ) v 1+m v where, cp is the specific heat of wood, J/kg °C; and Ta is the average temperature range, °C; and ma is the fractional moisture content, Ma/100. It is worth noting that the specific heat of wood depends on the temperature and moisture, but is practically independent of density and species (Simpson, 1991). Since Eq. (31) is an empirical equation, its accuracy is acceptable in this study. 3.4.7 Dielectric runs In each run, one green dielectric specimen was heated at a constant electrode voltage from approximately 20 to 60 °C. The computer was set to record the four temperature points once every six seconds through a data acquisition system. By analysing the computer data record, the temperature was classified into four ranges, namely 20 - 30, 30 - 40, 40 - 50 and 50 - 60 °C. The average values (TJ of each temperature range was considered as the four temperature levels in this study, which were 25, 35, 45 and 55 °C. Then, the specimen was wrapped with a plastic sheet (1 mm thick) and was stored in a cold room to cool down until it was below 20 °C. Another run 49 was then repeated at the same procedure mentioned above, but at another voltage level. Two electrode voltage levels, 0.8 and 1.1 kV, were chosen in this study. After these two runs, the plastic sheet was removed and the specimen was dried to a lower moisture content level in order to obtain another set of calorimetric data. By this process, a series of dT/dt at different moisture content levels (10, 20, 30, 40, 50, 60, 70, and 80% for sapwood, and 10, 20, 30, 40 and 50% for heartwood) were obtained. In order to reach a lower moisture content level before another calorimetric run, the specimen was dried in the same RF/V dryer under the electrode voltage of 0.5 kV and ambient pressure of 25 torr. Furthermore, the RF/V dryer could automatically stop when the moisture content of the specimen reached a certain level corresponding to the water height pre-set into the computer before the drying. The pre-set value of the water height in the collection tank was calculated according to Eq. (25). By solving Eqs.(15) and (17), the loss factors and power densities of western hemlock heartwood and sapwood, and western red cedar heartwood were calculated at various moisture content, temperature and rms voltage levels. This study was conducted with three types of wood (western hemlock heartwood and sapwood, and western red cedar heartwood), ten replications, two electrode voltages, eight moisture levels for sapwood and five moisture content levels for heartwood, and four temperature levels. Therefore, the study was run as factorial 10x2x8x4 and 2x10x2x5x4 designs for sapwood and heartwood, respectively, resulting in a total of 1440 experimental runs. 50 Chapter 4 4.0 Results and Discussion 4.1 Calibration of the RF electrode voltage The temperature-time relationships for. calibration of the RF electrode voltage by using the plexiglas block at the three positions and the six voltage levels, namely, input end, centre and far end, and 0.4, 0.6, 0.8, 1.0, 1.2 and 1.4 kV, respectively, are plotted in the Figures (12a), (12b), and (12c). From the preliminary experiments, during the calibration heating the computer was set in such way that the temperature values inside the plexiglas block were saved every 30 and 20 seconds through a data acquisition system when the calibration was carried out at 0.4 to 1.0 kV and 1.2 to 1.4 kV, respectively. Since the relationship of the temperature and time (i.e., temperature rise), as it was expected, was linear for all calibration heating, a linear regression equation was chosen to be fitted in order to obtain the coefficient a, which was taken as the time rate of temperature change dT/dt if Y represented the temperature (T) and X the time (t). Y=aX+b (32) where, Y is the dependent variable; X is the independent variable; a is the slope; and b is the intercept. Correspondingly, the calibration rms voltage V c was calculated according to Eq. (26). Listed in Table (1) are the time rates of temperature change, coefficients of determination (R2) and the calibration rms voltages V c corresponding to the reading rms voltages V r read from the RF meter. Comparing the calibration rms voltages at the three calibration positions, it was not 51 52 53 54 Table 1. Time rate of temperature change for the calibration of the RF electrode voltage. Position 9T/dt (°C/s) R2 V r (kV) V c (kV) 0.0006 0.8423 0.4 0.4164 0.0014 0.9274 0.6 0.6360 Input end 0.0028 0.9952 0.8 0.8994 0.0046 0.9954 1.0 1.1529 0.0066 0.9971 1.2 1.3809 0.0093 0.9980 1.4 1.6392 0.0004 0.8583 0.4 0.3400 0.0017 0.9481 0.6 0.7008 Centre 0.0031 0.9942 0.8 0.9464 0.0047 0.9924 1.0 1.1653 0.0071 0.9976 1.2 1.4323 0.0099 0.9975 1.4 1.6913 0.0006 0.6757 0.4 0.4164 0.0016 0.9381 0.6 0.6799 Far end 0.0033 0.9786 0.8 0.9765 0.0050 0.9907 1.0 1.2019 0.0083 0.9973 1.2 1.5486 0.0110 0.9980 1.4 1.7828 55 surprising that the calibrition rms voltages were highest at the far end position, and lowest at the input end position corresponding to each reading rms voltage with the exception of the calibration rms voltages corresponding to the reading rms voltages of 0.4 kV. The source of this, according to Pound (1973), could be attributed to the standing waves associated with the long electrode plate, which are the result of the voltage's tendency to follow a sine-wave that leads to field distortion. Thus, when an RF voltage is applied to a long electrode plate, the voltage increases as the distance from the feed point becomes greater. This phenomenon explained why the calibration rms voltages increased along the electrode plates from the input end to the far end. Also, it is noticed in the Table (1) that most of the R2 values were in the range of 0.9274 to 0.9980. Only for reading voltages of 0.4 kV at the input end, centre and far end, were the R2 values 0.8423, 0.8583 and 0.6757, respectively.The regression lines in the form of linear models fitted through the temperature and time for each reading rms voltage at various positions (i.e., input end, centre and far end) are also plotted in Figures (12a), (12b) and (12c). It is obvious that the linear model fitted the temperature and time data (temperature rise) quite well. The errors of the calibration existed in the three positions for reading rms voltage of 0.4 kV, could be due to the small scale of the reading voltage because during calibration heating, the fixed reading rms voltages were achieved by continually adjusting the RF/V dryer's two control levers, and any changes of the reading voltage could result in considerable inaccuracy in the calibration results. Therefore, the results of calibration at 0.4 kV were not taken into account in this study. By averaging the calibration rms voltages at the three positions and five reading rms voltage levels (i.e., 0.6, 0.8, 1.0, 1.2 and 1.4 kV), the average calibration rms voltages against the reading rms voltages were plotted in Figure (13). Again, by fitting Eq. (32), a linear regression 56 Calibration (Average) 0.0 -I 1 1 1 1 1 1 1 1 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Vr (kV) Figure 13. Calibration curve for rms voltage. 57 equation in which the calibration rms voltage was the dependent variable and the reading rms voltage the independent variable, was obtained: V =1.2887V -0.0998 ( c r V where, Vr is the reading rms voltage, kV; and V c is the calibration rms voltage, kV. The regression equation, coefficient of determination, regression line and error bars are also shown in Figure (13).The error bars were one sample standard deviations of the mean. It is can be seen from Figure (13) that the systematic error was small and the liner regression line fell within all the error bars. That means that the result of the calibration was precise and acceptable. From Eq. (33), the two rms electrode voltage levels of 0.8 and 1.1 kV with the standard deviations of 0.04 and 0.088 kV, were practically fixed at 0.698 and 0.931 kV, respectively. 4.2 The experimental data Since the accuracy of the specific heat calculated according to Eq. (31) directly affected the results of this study, an error analysis of cp was necessary. The error of the calculated cp can be attributed to two sources, namely, wood temperature, and the its fractional moisture content. The relative errors (Sr) of the calculated specific heat due to temperature and fractional moisture content were calculated using the error propagation analysis procedure (Hargis, 1988).The results are listed in Table (2). To simplify the calculation, the maximum standard deviation values of 0.5 °C and 0.02 for the temperature and fractional moisture content measured in this study, respectively, were used. It can be seen from Table (2) that most relative errors were within 10% 58 Table 2. Relative error of calculated specific heat Ta(°C) ma cP(J/kg°C) sr(%) 0.1 1320.05 20.10 0.2 1558.96 10.20 0.3 1761.12 6.96 25 0.4 1934.39 5.39 0.5 2084.57 4.47 0.6 2215.97 3.89 0.7 2331.91 3.49 0.8 2434.97 3.20 0.1 1331.47 20.05 0.2 1569.43 10.10 0.3 1770.78 6.82 35 0.4 1943.37 5.20 0.5 2092.94 4.25 0.6 2223.82 3.63 0.7 2339.30 3.19 0.8 2441.95 2.88 0.1 1342.89 20.03 0.2 1579.89 10.06 0.3 1780.44 6.76 45 0.4 1952.34 5.12 0.5 2101.32 4.15 0.6 2231.67 3.51 0.7 2346.69 3.07 0.8 2448.93 2.74 0.1 1354.30 20.02 0.2 1590.36 10.04 0.3 1790.10 6.73 55 0.4 1961.31 5.08 0.5 2109.69 4.10 0.6 2239.52 3.46 0.7 2354.08 3.00 0.8 2455.91 2.66 Total mean Sr (%) 7.01 59 with the exception of the relative error values at 10% moisture content. Therefore, the accuracy of the calculated specific heat was considered acceptable in this study. The mean loss factors of western hemlock heartwood and sapwood, and western red cedar heartwood at various moisture content, temperature and rms electrode voltage levels are listed in Tables (3) to (8). Again, the accuracy of the final result (loss factor value) depended on the error accumulation from each individual measurement. In this study, the error of the calculated e" could be attributed to the following sources: the time rate of temperature change, rms electrode voltage, wood density, and its calculated specific heat. In order to estimate the accuracy of the loss factor values, the maximum relative errors of the time rate of temperature change, rms electrode voltage, wood density and calculated specific heat were used. Based on the experimental data obtained in this study, the maximum relative errors of the time rate of temperature change for wesetrn hemlock sapwood and heartwood, and western red cedar heartwood were 0.05, 0.051 and 0.048, respectively, and the maximum relative error due to the density was 0.1. The relative errors of the rms electrode voltage and specific heat had been determined above (Table 2). According to the error propagation analysis procedure (Hargis, 1988), the uncertainty or error (estimation standard deviation, S£.) and relative error (Se.r=100x S£./e") values of the loss factor were caculated and also listed in Tables (3) to (8). It is obvious that the error values were not hight since all of them were less than 1.0. All the total mean relative errors were in the range of 14.71 to 17.19%, with a mean value of 16.14%, and their majority was below 16% with the only exception of those at 10% moisture content. Also, it is noticed from Tables (3) to (8) that the relative errors decreased with increasing moisture content while the other conditions are the same. That is to say that the precision and accuracy of the calculated e" 60 Table 3. Mean loss factor for western hemlock sapwood at 0.8 kV. T(°C) M (%) e" Se- se..r (%) 10 0.94 0.22 23.54 20 1.89 0.30 15.94 30 2.63 0.37 14.09 25 40 3.12 0.42 13.38 50 3.54 0.46 13.04 60 3.86 0.50 12.85 70 4.24 0.54 12.73 80 4.58 0.58 12.66 10 1.11 0.26 23.50 20 1.92 0.30 15.88 30 2.85 0.40 14.02 35 40 3.49 0.46 13.31 50 4.00 0.52 12.96 60 4.48 0.57 12.77 70 4.88 0.62 12.66 80 5.39 0.68 12.58 10 1.16 0.27 23.48 20 2.02 0.32 15.85 30 2.77 0.39 13.99 45 40 3.59 0.48 13.28 50 4.14 0.54 12.93 60 4.61 0.59 12.74 70 5.10 0.64 12.63 80 5.75 0.72 12.55 10 1.39 0.33 23.47 20 2.62 0.42 15.84 30 3.38 0.47 13.97 • 55 40 4.05 0.54 13.26 50 4.70 0.61 12.92 60 5.29 0.67 12.73 70 5.90 0.74 12.61 80 6.48 0.81 12.53 Total mean Se»r (%) 14.71 61 Table 4. Mean loss factor for western hemlock sapwood at 1.1 kV. T(°C) M (%) e" se..r(%) 10 0.86 0.21 24.35 20 1.43 0.25 17.12 30 2.13 0.33 15.41 25 40 2.69 0.40 14.76 50 3.20 0.46 14.46 60 3.67 0.52 14.29 70 4.13 0.59 14.18 80 4.52 0.64 14.12 10 1.05 0.26 24.31 20 1.73 0.29 17.06 30 2.29 0.35 15.35 35 40 3.00 0.44 14.70 50 3.40 0.49 14.39 60 4.07 0.58 14.22 70 4.46 0.63 14.11 80 4.86 0.68 14.05 10 1.10 0.27 24.29 20 1.76 0.30 17.04 30 2.49 0.38 15.32 45 40 3.28 0.48 14.67 50 3.71 0.53 14.36 60 4.06 0.58 14.19 70 4.77 0.67 14.09 80 5.31 0.74 14.02 10 1.26 0.31 24.29 20 1.99 0.34 17.02 30 2.79 0.43 15.31 55 40 3.52 0.52 14.66 50 4.07 0.58 14.35 60 4.42 0.63 14.18 70 5.06 0.71 14.07 80 5.46 0.76 14.00 Total mean Ss»r (%) 16.02 62 Table 5. Mean loss factor for western hemlock heartwood at 0.8 kV. T(°C) MC (%) e" se» se..r(%) 10 1.31 0.31 23.56 20 2.86 0.46 15.97 25 30 3.90 0.55 14.12 40 4.80 0.64 13.42 50 5.88 0.77 13.08 10 1.43 0.34 23.52 20 2.88 0.46 15.91 35 30 4.06 0.57 14.05 40 5.01 0.67 13.34 50 6.18 0.80 13.00 10 1.50 0.35 23.50 20 3.00 0.48 15.88 45 30 4.11 0.58 14.02 40 5.18 0.69 13.31 50 6.25 0.81 12.97 10 1.63 0.38 23.49 20 3.19 0.51 15.87 55 30 4.50 0.63 14.01 40 5.67 0.75 13.30 50 6.81 0.88 12.96 Total mean Ss»r (%) 15.96 63 Table 6. Mean loss factor for western hemlock heartwood at 1.1 kV. T(°C) M (%) 8" sE..r(%) 10 1.11 0.27 24.37 20 1.93 0.33 17.15 25 30 2.56 0.40 15.44 40 3.25 0.48 14.80 50 3.84 0.56 14.49 10 1.18 0.29 24.33 20 2.10 0.36 17.09 35 30 2.66 0.41 15.38 40 3.34 0.49 14.73 50 4.01 0.58 14.42 10 1.21 0.30 24.32 20 2.06 0.35 17.07 45 30 2.79 0.43 15.35 40 3.45 0.51 14.71 50 4.18 0.60 14.40 10 1.35 0.33 24.31 20 2.14 0.36 17.05 55 30 2.93 0.45 15.34 40 3.67 0.54 14.69 50 4.38 0.63 14.38 Total mean Se»r (%) 17.19 64 Table 7. Mean loss factor for western red cedar heartwood at 0.8 kV. T(°C) M (%) 8" s6. ss»r(%) 10 1.51 0.36 23.50 20 3.17 0.50 15.88 25 30 4.42 0.62 14.02 40 5.44 0.72 13.31 50 6.37 0.83 12.96 10 1.59 0.37 23.45 20 3.17 0.50 15.81 35 30 4.53 0.63 13.95 40 5.70 0.75 13.23 50 6.80 0.88 12.89 10 1.63 0.38 23.44 20 3.37 0.53 15.79 45 30 4.63 0.64 13.92 40 5.96 0.79 13.20 50 7.25 0.93 12.86 10 1.85 0.43 23.43 20 3.62 0.57 15.78 55 30 4.97 0.69 13.90 40 6.26 0.83 13.19 50 7.70 0.99 12.84 Total mean SE"r (%) 15.87 65 Table 8. Mean loss factor for western red cedar heartwood at 1.1 kV. T(°C) M (%) S" Se- Se-r(%) 10 0.88 0.21 24.31 20 1.74 0.30 17.06 25 30 2.44 0.37 15.35 40 3.05 0.45 14.70 50 3.70 0.53 14.39 10 0.94 0.23 24.27 20 1.88 0.32 17.00 35 30 2.60 0.40 15.28 40 3.33 0.49 14.63 50 4.03 0.58 14.32 10 1.16 0.28 24.25 20 2.27 0.39 16.98 45 30 3.18 0.49 15.26 40 4.04 0.59 14.60 50 4.74 0.68 14.29 10 1.31 0.32 24.25 20 2.35 0.40 16.97 55 30 3.38 0.51 15.24 40 4.28 0.62 14.59 50 5.25 0.75 14.28 Total mean Se"r (%) 17.10 66 values were influenced by moisture content. Similar behaviour was reported by Skaar (1948), who used a Q-meter to measure the perpendicular-to-grain dielectric properties of thirty different kinds of wood, and plotted a curve of the percent standard deviation of the dielectric constant experimental data points at 2 MHz versus wood moisture content. From his plotted curve, it was found that the dispersion of the experimental data was influenced by the moisture content, and in the range of oven-dry to 15% moisture content. Contrary to this study, the percent standard deviation increased as the moisture content increased. He also pointed out that it was difficult to obtain accurate measurements of the dielectric constant at high moisture contents due to the high variation of the experimental data, thus data from tests on wood above the fibre saturation point were not as reliable as those obtained below this point. The reason for this might be attributed to the power generated by the Q-meter. Avramidis and Dubois (1993) found that the Q-meter's power was not sufficient enough to measure the dielectric properties of the dielectric specimens saturated with distilled water. In this study, the output of the RF/V dryer generator was so high that the dielectric properties of dielectric specimens with high moisture content (80%) were measured and results showed that the variance of the experimental data was low. That the coefficients of variance were higher at lower moisture contents could be explained by the fact that, at low moisture contents the specimens was considered as low loss dielectric materials, the loss factor was so low that any errors coming from the experimental work would result in larger data variance compared with those at higher moisture contents. James (1975), in his comprehensive study where he used a Q-meter to measure the dielectric properties of white oak, Douglas-fir, and four commercial hardboards, found that the data acquired in the study were generally constant and repeatable to within plus or minus 10% 67 of the mean data. 4.3 Analysis of the experimental results In order to find the effects of variables on the loss factor, the experimental data were analysed using the fully factorial ANOVA procedure. The resulted ANOVA table is listed in Table (9). It is obvious that the moisture content, temperature, rms electrode voltage and species had a significant effect on the loss factor of wood during this study, since all of them were highly significant (P = 0.000). However, the first- to fourth-order interactions were also found significant at high level. The reasons for this could be attributed to two sources: one is the experimental design, the other is the experimental errors. For the consideration of time and cost, it was almost impossible to conduct this study following a completely randomized factorial design process. For example, even only for western hemlock heartwood, in order to obtain one set of observations under the 40 conditions (five moisture content, four temperature and two rms electrode voltage levels, 5 x 4 x 2), 40 pieces lumber were required. The experimental errors might come from the temperature levels, because the four temperature levels, 25, 35, 45 and 55 °C, were the average values of the four temperature ranges, namely, 20 - 30, 30 - 40, 40 - 50 and 50 - 60 °C, respectively. That means that the four temperature levels were not truly fixed. Nevertheless, investigation of different factors affecting the loss factor of wood still can offer us meaningful information. 68 Table 9. Fully factorial ANOVA ~ effects of the variables on the loss factor. Source ss DF MS F P M 1367.661 5 273.532 12112.215 0.000 T 11.672 3 3.891 172.278 0.000 V 39.739 1 39.739 1759.663 0.000 S 69.787 2 34.893 1545.108 0.000 M*T 5.736 16 0.358 15.875 0.000 M*V 49.684 5 9.937 440.007 0.000 M*S 298.668 10 29.869 1322.612 0.000 T*V 0.326 3 0.109 4.812 0.002 T*S 0.473 6 0.079 3.491 0.002 V*S 17.741 2 8.871 392.794 0.000 M*T*V 0.755 16 0.047 2.091 0.007 M*T*S 2.452 29 0.085 3.744 0.000 M*V*S 13.894 10 1.385 61.323 0.000 T*V*S 0.545 5 0.109 4.824 0.000 M*T*V*S 1.256 30 0.042 1.853 0.004 Note: M in the moisture content %; T is the temperature °C; V is the rms electrode voltage, kV; and S is the species. 69 4.3.1 Moisture effect The effects of moisture content on the loss factor are graphically described in Figures (14) to (16). In these figures, the mean values of the ten dielectric specimens' loss factors for each species at various rms electrode voltages were plotted against moisture contents. For the shape of these data trends, a quadratic regression equation was fitted to the experimental data for each species at different rms electrode voltages using the nonlinear least square fit procedure e"=aM2+bM+c (34) where, a, b and c are the parameters of the quadratic regression equation. The quadratic regression lines and the error bars that indicating one sample standard deviations of the mean, are also shown on Figures (14), (15) and (16), and the parameters (a, b and c) and the coefficient of determination values were tabulated in Table (10). In Table (10), all the R2 values were very close to 1. Thus, the quadratic equation fitted the mean loss factor versus moisture content data quite well. The above results are in good agreement with the work conducted by Peyskens et al. (1984). They carried out the dielectric measurements at 3 GHz for three softwoods, i.e., European pine, spruce and hemlock within a moisture content range of 6 to 35%. By testing different equations as best fit to the experimental values, they concluded that the quadratic regression equation offered the best fit. The researchers considered two facts trying to explain this phenomenon: firstly, with increasing moisture content the amount of water within the wood matrix increases, which itself is characterized by high dielectric values; secondly, the polar 70 CO 7 6 5 4 3 2 1 0 Western hemlock sapwood, 0.8 kV + + + + 10 20 30 40 50 M (%) 60 70 80 o 25 C • 35 C A 45 C x 55 C 6 5 4 =co 3 + 2 1 0 0 Western hemlock sapwood, 1.1 kV + + + + + 10 20 30 40 50 M (%) 60 70 80 o 25 C • 35 C A 45 C x 55 C Figure 14. Plot of loss factor against moisture content for western hemlock sapwood at 0.8 kV (top) and 1.1 kV (bottom). 71 CO 7 6 + 5 4 3 2 + 1 0 0 Western hemlock heartwood, 0.8 kV 10 —I \— 20 30 M (%) 40 50 co 4.5 4 3.5 3 + 2.5 2 1.5 1 0.5 0 0 Western hemlock heartwood, 1.1 kV 10 —I 1— 20 30 M (%) 40 o25 C • 35 C A 45 C x 55 C 50 Figure 15. Plot of loss factor against moisture content for western hemlock heartwood at 0.8 kV (top) and 1.1 kV (bottom). 72 Western red cedar heartwood, 0.8 kV 8 7 + 6 5 = c o 4 3 2 + 1 0 0 + 10 20 30 M (%) 40 50 o25 C • 35 C A 45 C x 55 C Western red cedar heartwood, 1.1 kV Figure 16. Plot of loss factor against moisture content for western red cedar heartwood at 0.8 kV (top) and 1.1 kV (bottom). 73 Table 10. Quadratic regression parameters, a, b and c, from the loss factor versus moisture content data. Species V(kV) T(°C) a b c R2 25 -0.0005 0.0926 0.1622 0.9945 0.8 35 -0.0004 0.0980 0.1881 0.9977 45 -0.0003 0.0924 0.2951 0.9977 WHS 55 -0.0004 0.1024 0.5522 0.9966 25 -0.0002 0.0722 0.1331 0.9996 1.1 35 -0.0002 0.0736 0.3269 0.9983 45 -0.0002 0.0763 0.3628 0.9949 55 -0.0003 0.0892 0.3937 0.9977 25 -0.0008 0.1565 -0.1057 0.9963 0.8 35 -0.0006 0.1507 0.0236 0.9982 45 -0.0006 0.1553 0.0530 0.9986 Willi 55 -0.0007 0.1707 0.0150 0.9997 25 -0.0003 0.0857 0.2939 0.9993 1.1 35 -0.0003 0.0858 0.3928 0.9966 45 -0.0002 0.0864 0.3886 0.9993 55 -0.0001 0.0847 0.5150 1.0000 25 -0.0012 0.1918 -0.2556 0.9993 0.8 35 -0.0008 0.1784 -0.0954 0.9999 45 -0.0006 0.1741 0.0013 0.9987 RCH 55 -0.0005 0.1743 0.2187 0.9984 25 -0.0004 0.0913 0.0219 0.9991 1.1 35 -0.0003 0.0968 0.0273 0.9992 45 -0.0006 0.1268 -0.0389 0.9998 55 -0.0002 0.1090 0.2469 0.9998 Note: WHS, WHH and RCH stand for western hemlock sapwood, western hemlock heartwood and western red cedar heartwood, respectively. 74 components of cellulose obtained more freedom of rotation at higher moisture contents and in this way also contributed to a more pronounced dielectric behaviour. Therefore, when moisture was below the fibre saturation point, the combination of these two factors resulted in a rapid increase of the loss factor value. However, at higher moisture contents, the importance did not increase any more because their freedom of rotation reached a maximum when moisture content was at the fibre saturation point. Thus, this resulted in a lower slope curve. It is not surprising that the loss factor increased with increasing moisture content, because both dielectric constant and loss tangent have a positive relationship with the moisture content (Skaar, 1948; Hearmon and Burcham, 1954; Vermaas et al., 1974; James, 1975). Yavorsky (1951) also concluded that the loss factor of wood increased with moisture content when measured at a constant temperature. Similar "results were verified by past studies. It is evident from the curves of loss factor against moisture content plotted by Peterson (1951), that the loss factor increased regularly with moisture content up to 12%, and at higher moisture content (12 - 20%) this trend was more pronounced. Vermaas (1971) plotted the loss factor of Pinus pinaster, a common locally grown pine, as a function of moisture content for different densities and resin contents according to a polynomial regression equation.The curves were found to be a flat S-shape - the loss factor increased with increasing moisture content. It should be emphasised that the moisture contents tested in that study were in the range of 0 to 28%, which was below the fibre saturation point. Torgovnikov (1993) tabulated the dielectric constant and loss factor of wood at frequencies of 104 Hz to 104 MHz, moisture contents of 0 to 100%, temperatures between -20 to 90 °C and densities in oven-dry condition of 0.3 to 0.8 g/cm3. In order to compare Torgovnikov's 75 data with the result of this study, the loss factor (the production of dielectric constant and loss tangent) was plotted against moisture content for the frequency of 10 MHz, temperatures of 20 and 50 °C, and oven-dry densities of 0.3, 0.4, and 0.5 g/cm3. These curves could be seen in Figure (17). Again, the loss factor (Torgovnikov, 1993) had a positive relationship with moisture content, but its measured values were lower than those obtained in this study. However, most of them agreed within the the confidence interval at the 99% level with the exception of those at 10%) moisture content. This could be due to the high relative errors of the loss factor at this moisture content level. Avramidis and Dubois (1993) proposed that the loss factor was an exponential function of the wood moisture content, hence the higher the moisture content, the higher was the loss factor. In their study, only one moisture content data point was above the fibre saturation point. They cited Lin's (1967) conclusion that this trend was the result of the high dielectric constant of water compared with the relatively low dielectric constant of cellulose. In addition, the polar groups in the cell wall and in particular in cellulose, had increased freedom of rotation when the moisture content increased. 4.3.2 Temperature effect Shown in Figures (18), (19) and (20) are plots of the mean loss factor of the ten dielectric specimens for each species versus temperature at different moisture contents and rms electrode voltages. A linear regression model was fitted to the mean experimental data 76 Oven-dry density = 0.3 g/cm3 4 -, 3 -"co 2 -1 -0 « Si" 1 1 fr 1 | 1 i -o - 2 0 C - A - 5 0 C T 1 • 1 1 i 0 20 40 60 80 100 M(%) Oven-dry density = 0.4 g/cm3 M (%) Oven-dry density = 0.5 g/cm3 0 20 40 60 80 100 M (%) Figure 17. Plot of published loss factor data (Torgovnikov, 1993) for various moisture contents, two temperatures and three densities. 77 e"=a+bT (35) Where, a and b is the interception and slope of the regression line, respectively; and T is the temperature, °C. The linear regression lines plotted in Figures (18), (19) and (20) indicated how well the linear regression model has fitted the mean experimental data, and the values of R2 offered an idea about the spread of the data along the straight lines. As can be seen in Table (11), all slopes were positive and in most cases the slopes increased as the moisture content increased. In other words, the influence of the temperature on the loss factor was positively linear, and more pronounced at higher moisture contents. In addition, this trend was held irrespective of the different species and rms electrode voltages. Lin (1967) also found that the effects of temperature and frequency on wood dielectric properties were more pronounced when the moisture content was high. He theorised that when temperature was increased, dielectric constant and relaxation spectrum shifted towards higher frequency. This shift was caused by change in the mobility of polar molecules with temperature. Nelson and Kraszewski (1990) agreed with Lin (1967), and pointed out that as temperature increased, the relaxation time decreased, and the peak of the loss factor of materials illustrated in Figure (5) would shift to higher frequencies. Thus, in a region of dispersion, the dielectric constant increased with increasing temperature, whereas the loss factor might either increase or decrease, depending on whether the frequency used was higher or lower than the relaxation frequency. 78 Western hemlock sap wood, 0.8 kV 7 T 25 35 45 T(°C) Western hemlock sap wood, 1.1 kV 6 T 25 35 45 55 T(°C) Figure 18. Plot of loss factor against temperature for western hemlock sapwood at 0.8 kV (top) and 1.1 kV (bottom). 79 CO 7 6 5 4 3 2 1 0 Westerm hemlock heartwood, 0.8 kV 25 1 35 45 T(°C) 55 o 10% • 20% A 30% x 40% x 50% co 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 Western hemlock heartwood, 1.1 kV 25 35 45 T(°C) 55 o 10% • 20% A 30% x 40% x 50% Figure 19. Plot of loss factor against temperature for western hemlock heartwood at 0.8 kV (top) and 1.1 kV (bottom). 80 81 Table 11. Linear regression parameters, a and b, from loss factor versus temperature data. Species V(kV) M (%) a b R2 10 0.7929 0.1428 0.9506 20 1.5364 0.2301 0.7404 30 2.3675 0.2166 0.7291 0.8 40 2.8373 0.2897 0.9523 50 3.1914 0.3625 0.9576 60 3.4580 0.9465 0.9465 70 3.7311 0.5194 0.9602 WHS 80 4.0308 0.6072 0.9805 10 0.7536 0.1261 0.9537 20 1.2996 0.1707 0.9282 30 1.8792 0.2178 0.9783 1.1 40 2.4275 0.2782 0.9961 50 2.8657 0.2924 0.9842 60 3.4938 0.2238 0.8915 70 3.8256 0.3112 0.9994 80 4.2190 0.3268 0.9664 10 1.0617 0.0102 0.9845 20 2.5423 0.0110 0.9017 0.8 30 3.3942 0.0187 0.8765 40 4.0592 0.0277 0.9310 WHH 50 5.1328 0.0287 0.9063 10 1.0193 0.0775 0.9348 20 1.9105 0.0582 0.6985 1.1 30 2.4290 0.1230 0.9961 40 3.0813 0.1385 0.9506 50 3.6500 0.1800 0.9978 10 1.3826 0.1053 0.8810 "20 2.9401 0.1570 0.8799 0.8 30 4.1961 0.1762 0.9100 40 5.1607 0.2175 0.9984 RHC 50 5.9200 0.4440 0.9990 10 0.6905 0.1527 0.9612 20 1.5049 0.2229 0.9371 1.1 30 2.0485 0.3401 0.9482 40 2.5767 0.4393 0.9596 50 3.0900 0.5360 0.9829 Note: WHS, WHH and RHC stand for western hemlock sapwood, western hemlock heartwood and western red cedar neartwood, respectively. 82 According to Torgovnikov (1993), at 20 °C, the maximum value of the loss factor (region of dispersion) is at 10 MHz when the moisture content is 10%. Whereas, when the moisture content is above 20%, it is higher than 104 MHz. Thus, all the above theoretically explain why at 13.56 MHz used in this study, the loss factor increased with increasing temperature. Nanassy (1970) also confirmed this study's findings. In his study, from the relaxation spectra of oven-dry yellow birch over the frequency range from 20 Hz to 2 GHz and at the temperature range from 20 to 100 °C, it was found that at 10 - 100 MHz, the loss factor was increased as the temperature increased. Since the researcher did not carry out a data analysis, the exact relationship of the temperature and loss factor could not be found. 4.3.3 Electric field strength effect The mean values of the loss factors of the ten dielectric specimens for each species at various temperatures and moisture contents were plotted against the rms electrode voltage (Figures 21a, 21b, 22a, 22b, 23a and 23b). In order to find the effect of the electric field strength on the loss factor, the slopes of straight line formed by two mean experimental data points at 0.8 and 1.1 kV rms electrode voltage were calculated and listed in Table (12). It can be seen from Table (12) that all values of the slope were negative. That means that the loss factors were higher at 0.8 kV rms electrode voltage than those at 1.1 kV rms electrode voltage. In other words, the higher rms electrode voltage, e.g., electric field strength, resulted in a lower loss factor value. Furthermore, it seems that there was a trend: the higher the moisture content, the more clear the effect was, because the absolute values of the slopes were larger at higher moisture contents, 83 Western hemlock sapwood, 25 °C 5 j 4.5 --0.5 0.8 1.1 1.4 V(kV) Western hemlock sapwood, 35 °C 6 T 0.5 0.8 1.1 1.4 V(kV) Figure 21a. Plot of loss factor against voltage for western hemlock sapwood at 25 °C (top) and 35 °C (bottom). 84 6 5 4 =co 3 + 2 1 0 0.5 Western hemlock sapwood, 45 °C -a -© —I 1— 0.8 1.1 VfkV) o 10% • 20% A 30% x 40% x 50% o 60% + 70% - 80% 1.4 CO 7 6 5 + 4 3 2 1 0 0.5 Western hemlock sapwood, 55 °C 0.8 1.1 V(kV) o 10% • 20% A 30% x 40% x 50% o 60% + 70% - 80% 1.4 Figure 21b. Plot of loss factor against voltage for western hemlock sapwood at 45 °C (top) and 55 °C (bottom). 85 6 T 5 4 = c o 3 2 1 0 0.5 Western hemlock heartwood, 25 °C + + 0.8 1.1 V(kV) o 10% • 20% A 30% x 40% x 50% 1.4 CO 7 6 5 4 3 2 1 0 0.5 Western hemlock heartwood, 35 °C —I 1— 0.8 1.1 V(kV) o 10% • 20% A 30% x 40% x 50% 1.4 Figure 22a. Plot of loss factor against voltage for western hemlock heartwood at 25 °C (top) and 35 °C (bottom). 86 CO 7 6 5 4 3 2 1 0 0.5 Western hemlock heartwood, 45 °C 0.8 1.1 VfkV) 1.4 o 10% • 20% A 30% x 40% x 50% c o 7 6 5 4 3 2 1 0 0.5 Western hemlock heartwood, 55 °C + + 0.8 1.1 V(kV) 1.4 o 10% • 20% A 30% x 40% x 50% Figure 22b. Plot of loss factor against voltage for western hemlock heartwood at 45 °C (top) and 55 °C (bottom). 87 CO 7 6 5 4 3 2 + 1 -0 -0 Western red cedar heartwood, 25 °C + —I 0.8 1.1 V(kV) 1.4 o 10% • 20% A 30% x 40% x 50% CO 7 -r 6 5 + 4 3 2 -f 1 0 0.5 Western red cedar heartwood, 35 °C —I h-0.8 1.1 V(kV) 1.4 o 10% • 20% A 30% x 40% x 50% Figure 23a. Plot of loss factor against voltage for western red cedar heartwood at 25 °C (top) and 35 °C (bottom). 88 8 7 6 5 =co 4 3 2 1 0 0.5 Western red cedar heartwood, 45 °C + + 0.8 1.1 V(kV) 1.4 o 10% • 20% A 30% x 40% x 50% 8 7 6 5 =co 4 3 -2 --1 -0 -0 Western red cedar heartwood, 55 °C 0.8 1.1 V(kV) 1.4 o 10% • 20% A 30% x 40% x 50% Figure 23b. Plot of loss factor against voltage for western red cedar heartwood at 45 °C (top) and 55 °C (bottom). 89 Table 12. Regression line slopes from the loss factor versus rms electrode voltage data. Species T(°C) M (%) Slope Species T(°C) M (%) Slope 10 -0.2570 10 -0.6964 20 -1.5241 25 20 -3.1012 30 -1.6753 30 -4.4487 25 40 -1.4438 40 -5.1944 50 -1.1361 50 -6.8167 60 -0.6454 10 -0.8340 70 -0.3630 35 20 -2.6098 80 -0.1900 30 -4.6527 10 -0.1800 WHH 40 -5.5663 20 -0.6446 50 -7.2167 30 -1.8770 10 -0.9408 35 40 -1.6075 45 20 -3.1515 50 -2.0007 30 -4.3804 60 -1.3815 40 -5.7633 70 -1.4222 50 -6.9167 WHS 80 -1.7789 10 -0.9286 10 -0.2103 55 20 -3.4980 20 -0.8533 30 -5.2478 30 -0.9471 40 -6.6667 45 40 -1.0295 50 -8.1000 50 -1.4415 10 -2.1100 60 -1.8441 25 20 -4.7500 70 -1.1185 30 -6.5996 80 -1.4544 40 -7.9722 10 -0.4325 50 -8.9000 20 -2.1141 10 -2.1800 30 -1.9719 35 20 -4.3007 55 40 -1.7633 30 -6.4256 50 -2.1014 RCH 40 -7.8880 60 -2.9111 50 -9.2333 70 -2.7778 10 -1.5687 80 -3.4133 45 20 -3.6567 Note: WHS, WHH and RCH stand for 30 -4.8412 western hemlock sapwood, western hemlock 40 -6.3944 heartwood and western red cedar heartwood, 50 -8.3667 respectively. 10 -1.7862 55 20 -4.2327 30 -5.3074 40 -6.6056 50 -8.1667 90 particularly in western hemlock heartwood and western red cedar heartwood. Since the influence of the electric filed strength on the dielectric properties of wood had not been sufficiently studied yet, the research work about the effect of electric field strength on the water dielectric properties might be a good explanation of this study's results. Briggs (1928), after a literature review, cited a number of researchers' results where the dielectric constant of the liquid in the region of an interface was assumed to be equal to that of the liquid in bulk. The dielectric constant of water in the presence of an electric field could be lowered from a value of 80 to a value of 1 by placing the water in an electric field strength of the order of 5 x 10s V/cm. On the other hand, the electric field strength dependence of the dielectric constant of the solid dielectric material, such as barium-strontium titanate ceramic was also studied, and the results showed that when the temperature was above 350 °C, called Curie point for ceramics, the relationship between the dielectric constant and the electric field strength was negative (Von Hippel, 1954). All the electric field strengths used in the above mentioned studies were in the order of 105 - 106 V/cm. However, Nanassy (1972) carried out dielectric measurements on moist wood at 25 °C in A.C. fields from 4 to 500 V/cm and over a frequency range from 100 Hz to 100 kHz. He found that at any frequency, a change in the applied electric field strength left the calculated value of the dielectric constant unchanged. For the vacuum dried specimen there was no change of the loss tangent, but for the moist specimen (M < 30%), an increase in the electric field strength caused an increase in loss tangent only at frequencies lower than 104 Hz. According to Von Hippel (1954), a linear dielectric material that is not field-strength sensitive, may show field-strength sensitivity at high temperature (near 500 °C). Whether or not 91 wood has this characteristic, or if the critical temperature is lower than 500 °C in wood need to be further studied. As mentioned above, the effect of electric field strength on the dielectric properties of wood is not taken into account by the other researchers, with the except of Nanassy (1972), whose result seems to be opposite to this study's. However, the frequency of 13.56 MHz used in this study is much higher than those used by Nanassy (1972). Therefore, the interaction between the frequency and electric field strength should also be considered in the further study. Nevertheless, in the light of this limited study, a definite effect of the electric field strength on the loss factor was identified and an inversely proportional relation was derived. 4.3.4 Species effect The mean values of the loss factor of the ten dielectric specimens for the three species wood at different temperatures and rms electrode voltages were plotted against the moisture content and shown in Figures (24a), (24b), (25 a), and (25b). Also, the quadratic regression lines obtained by fitting the quadratic equation to the loss factor and moisture content data at different rms electrode voltages and temperatures are plotted in these figures. Obviously, most loss factors showed highest values for western red cedar heartwood and lowest ones for western hemlock sapwood, if the other testing conditions being the same. The only exceptions were those at 1.1 kV for 25 and 35 °C. By close inspection of these figures, it could be noticed that the differences between the loss factor values among these three species were larger at higher moisture contents than those at lower moisture content. In order to explain this, two factors should be taken into account - extractives content of the species and the actual weight of moisture that is present in the 92 93 94 1.1 kV, 25 °C 95 1.1 kV, 45 °C o WHS • win i A R C H 0 20 40 60 80 M (%) 1.1 kV, 55 ° C 96 wood. Considering the extractives content, western red cedar heartwood possesses a higher extractives content level than the western hemlock sapwood and heartwood. Vermaas et al. (1974) and Vermaas (1973) stated that although the loss tangent in the grain direction was not influenced by the amount of the extractives, the loss tangent increased linearly with increasing extractive content in the radial direction, and slightly decreased in the tangential direction. As the researchers pointed out, the mechanism of the difference in the loss tangent with the extractive content was difficult to explain, and must be correlated with the distribution of extractive in wood, chemical composition of the extractive and an orientational positioning or association of cellulose and extractive molecules. Based on this study's results, it is premature to reach a satisfying conclusion on the significant influence of the extractives content on the loss factor. The actual weight of moisture presented in the wood can be expressed as the ratio of the moisture weight and one unit oven-dry volume of the wood. Therefore, for a certain percent of moisture, a higher absolute moisture quantity is present in an equal volume of a heavier wood species. That means that the dielectric properties perhaps has been determined more by the moisture weight than by wood species (Peyskens et al., 1984). Also it can be concluded that increasing the same moisture content in two wood species will result in a larger increase of the actual moisture weight in the heavier wood species compared with that in a lighter wood species. In this study, the basic densities of western hemlock sapwood and heartwood, and western red cedar heartwood were calculated based on the measured green volume and calculated oven-dry weight of the specimens. Their mean values were 439.71 and 444.47, and 339.29 kg/m3, with a coefficient of variation of 12.86, 11.35 and 6.04%, respectively. Obviously, the basic density of 97 western hemlock heartwood was higher than that of sapwood. This could explain why the loss factor values of the former were higher than those of the latter, and why this trend was more pronounced at higher moisture contents. However, considering the actual moisture weight, we could not explain why the loss factor values of western red cedar were higher than those of both western hemlock sapwood and heartwood, since the western red cedar heartwood had the lowest basic density value. The higher extractive content of hemlock red cedar heartwood could have contributed to this discrepancy. 4.4 Regression equations Since two rms electrode voltage levels utilised in this study, that were too few to be used in developing a regression equation, and because the variable of species belonged to a categorical one, two regression equations for each species at two rms electrode voltage levels were developed. From the shape of the curves obtained and discussed in Chapter 4.3, six regression equations in the general form of Eq. (42) containing the terms of second power of moisture content (M2), moisture content (M), temperature (T) and the first order interaction of moisture content and temperature (MT) for each species at two rms electrode voltages, were obtained following the stepwise regression procedure, and only the independent variables which were significant at the 5% level in the analysis of variance were chosen. It should be also mentioned that in favour of the calculation process, the percent of moisture content data were transferred to the fractional ones. 98 e " =a+bM2 +cM+dT +eMT Where, a, b, c, d and e are the coefficients. The resulting six regression equations are (1) western hemlock sapwood at 0.8 kV: e" = -4.188M2+7.249M+0.006T+0.064MT at 1.1 kV: e" = -0.212-2.432M2+6.798M+0.013T+0.026MT (2) western hemlock heartwood at 0.8 kV: e" = -8.442M2 + 16.628M+0.017T at 1.1 kV: e"=0.241-2.814M2+7.721M+0.003T+0.026MT (3) western red cedar heartwood at 0.8 kV: e" = -9.932M2+18.972M+0.017T 99 Figure 26. 3-D plot of the loss factor for western hemlock sapwood at 0.8 kV. Figure 27. 3-D plot of the loss factor for western hemlock sapwood at 1.1 kV. 100 Figure 28. 3-D plot of the loss factor for western hemlock heartwood at 0.8 kV. 8 f Figure 29. 3-D plot of the loss factor for western hemlock heartwood at 1.1 kV. 101 Figure 30. 3-D plot of the loss factor for western red cedar heartwood at 0.8 kV. Figure 31. 3-D plot of the loss factor for western red cedar heartwood at 1.1 kV. 102 at 1.1 kV: e"=-4.545M2+7.07M+0.005T+0.097MT (42) These six empirical regression equations were three-dimensionally plotted in Figures (26) -(31). It can be seen from these figures that the trends of the loss factor change with the moisture content and temperature are in a good agreement with the discussion in Chapter 4.3. For all practical purposes, the empirical regression equations made possible the calculation of the loss factor of wood and in turn the power density deposited within wood during the RF heating without the requirement of direct measurements. It should be emphasized that the above empirical regression equations are applicable only within the test conditions applied to this study. 4.5 Calculated power density The power density values for the three species at two rms electrode voltages, various moisture contents and temperatures, were calculated according to Eq. (15) and the empirical regression equations (37) to (42). These calculated power densities are tabulated in Tables (13 a), (13b), (14) and (15). From Eq. (15), it can be seen that the power density was proportional to the square of the electric field strength, in turn the rms electrode voltage, the frequency and the loss factor. Although the relationship between the loss factor and the electric field strength was negative as discussed in Chapter 4.3.3, the effect of the electric field strength on the power density was positive. This means that the higher the electric field strength, the larger the power density if the other conditions being the same. On the other hand, as seen in Chapters 4.3.1 and 103 Table 13a. Calculated power density for western hemlock sapwood at 0.8 kV, various moisture contents and temperatures. Species V(kV) T(°C) M (%) PD (kW/m3) 25 10 39.60 25 20 69.88 25 30 96.81 25 40 120.41 25 50 140.67 25 60 157.59 25 70 171.16 25 80 181.40 35 10 44.54 35 20 77.37 35 30 106.86 35 40 133.01 35 50 155.82 35 60 175.29 35 70 191.42 WHS 0.8 35 80 204.21 45 10 49.49 45 20 84.87 45 30 116.91 45 40 145.61 45 50 170.98 45 60 193.00 45 70 211.68 45 80 227.02 55 10 54.43 55 20 92.37 55 30 126.96 55 40 158.22 55 50 186.13 55 60 210.70 55 70 231.94 55 80 249.83 104 Table 13b. Calculated power density for western hemlock sapwood at 1.1 kV, various moisture contents and temperatures. Species V(kV) T(°C) M (%) PD (kW/m3) 25 10 62.84 25 20 113.49 25 30 160.48 25 40 203.79 25 50 243.45 25 60 279.43 25 70 311.75 25 80 340.40 35 10 74.60 35 20 127.21 35 30 176.16 35 40 221.44 35 50 263.05 35 60 300.99 35 70 335.27 WHS 1.1 35 80 365.88 45 10 86.36 45 20 140.93 45 30 191.84 45 40 239.08 45 50 282.65 45 60 322.55 45 70 358.79 45 80 391.36 55 10 98.12 55 20 154.66 55 30 207.52 55 40 256.72 55 50 302.25 55 60 344.12 55 70 382.31 55 80 416.85 105 Table 14. Calculated power density for western hemlock heartwood at two rms electrode voltages, various moisture contents and temperatures. Species V(kV) T(°C) M (%) PD (kW/m3) 25 10 79.89 25 20 136.10 25 30 185.58 25 40 228.32 25 50 264.33 35 10 86.67 35 20 142.88 35 30 192.35 35 40 235.10 0.8 35 50 271.11 45 10 93.45 45 20 149.66 45 30 199.13 45 40 241.88 45 50 277.89 55 10 100.23 55 20 156.44 55 30 205.91 55 40 248.66 WHH 55 50 284.67 25 10 84.81 25 20 141.56 25 30 194.07 25 40 242.33 25 50 286.34 35 10 89.04 35 20 147.74 35 30 202.21 35 40 252.43 1.1 35 50 298.41 45 10 93.26 45 20 153.93 45 30 210.35 45 40 262.53 45 50 310.47 55 10 97.48 55 20 160.11 55 30 218.49 55 40 272.64 55 50 322.53 106 Table 15. Calculated power density for western red cedar heartwood at two rms electrode voltages, various moisture contents and temperatures. Species V(kV) T(°C) M (%) PD (kW/m3) 25 10 107.26 25 20 184.42 25 30 252.01 25 40 310.00 25 50 358.42 35 10 115.46 35 20 192.63 35 30 260.21 35 40 318.21 0.8 35 50 366.62 45 10 123.66 45 20 200.83 45 30 268.41 45 40 326.41 45 50 374.82 55 10 131.87 55 20 209.03 55 30 276.61 55 40 334.61 RCH 55 50 383.02 25 10 93.88 25 20 168.06 25 30 233.95 25 40 291.54 25 50 340.84 35 10 107.29 35 20 190.32 35 30 265.05 35 40 331.50 1.1 35 50 389.65 45 10 120.70 45 20 212.58 45 30 296.16 45 40 371.46 45 50 438.46 55 10 134.11 55 20 234.83 55 30 327.27 55 40 411.41 55 50 487.26 107 4.3.2, the moisture content and temperature had positive effects on the loss factor. These relationships also existed between the power density and the moisture content and temperature. 108 Chapter 5 5.0 Conclusions The objective of this study was to investigate the loss factor of two B.C. species, namely, western hemlock and western red cedar at different moisture content, temperature and rms electrode voltage levels, by using the direct calorimetric data method. In the light of this study, the following conclusions can be summarized below: (1) The direct calorimetric method was practical and feasible for the measurement of the full-sized dielectric wood specimens. The accuracy of this method and the data measured were comparable with that of the Q-meter method and the published data, respectively. (2) The moisture content of wood had a significant effect on its loss factor, and the loss factor increased proportionally to the second power of moisture content. (3) The temperature significantly affected the loss factor of wood, and the relationship between them was directly proportional and linear. (4) The electric field strength exhibited a significant effect on the loss factor of wood. The higher the electric filed strength, the lower was the loss factor. (5) Different species had different loss factors. From the practical point of view, the empirical regression equations obtained based on the experimental data made the calculation of the loss factor possible, and this was expected to provide useful information for the development of a mathematical model to describe heat and moisture transfer mechanisms in wood during RF/V drying. Since this study was carried out only at a limited range of the variables, further research 109 work is recommended to investigate the dielectric properties of wood covering a wider range of the variables, such as the other ISM frequency bands and principal structural directions. Although the effect of the electric field strength on the dielectric properties of wood was found, further study still needs to be carried out as recommended in Chapter 4.3.3. 110 Literature Cited Anderson, J. C. 1964. Dielectrics. Chapman and Hall Ltd., London. Avramidis, S. and J. F. Graham Mackay. 1988. Development of kiln schedules for 4-inch by 4-inch pacific coast hemlock. Forest Products Journal. 38(9): 45 - 48. Avramidis, S. and L. Oliveira. 1993. Influence of presteaming on kiln-drying of thick hem-fir lumber. Forest Products Journal. 43(11/12): 7 - 12. Avramidis, S. and R. L. Zwick. 1992. Exploratory radio-frequency/vacuum drying of three B.C. coastal softwood. Forest Products Journal. 42(7/8): 17-24. Avramidis, S., R. L. Zwick and J. B. Neilson. 1996a. Commercial-scale RF/V drying of softwood lumber. Part 1.Basic kiln design considerations. Forest Products Journal. 46 (5): 44- 51. Avramidis, S. and R. L. Zwick. 1996b. Commercial-scale RF/V drying of softwood lumber. Part 2. Drying characteristics and lumber quality. Forest Products Journal. 46 (6): 27 - 36. Avramidis, S. and R. L. Zwick. 1997. Commercial-scale RF/V drying of softwood lumber. Part 3. Energy consumption and economics. Forest Products Journal. 47 (1): 48 - 56 Avramidis, S., F. Liu and B. J. Neilson. 1994. Radio-frequency/vacuum drying of softwood: ' drying of thick western red cedar with constant electrode voltage. Forest Products Journal. 44(1): 41 - 47. Avramidis, S. and J. Dubois. 1993. The study of dielectric properties of spruce, hemlock, western red cedar and Douglas-fir at varying MC, temperature, grain orientation and radio frequency. Report No. 93 (SA-3) (unpublished). Science Council of British Columbia, Vancouver, B.C., Canada. Biryukov, V. A. 1961. Dielectric heating and drying of wood. GLBI Goslesbumizdat, Moskva-Leningrad. Bramhall, G. and R. W. Wellwood. 1976. Kiln drying of western Canadian lumber, Fisheries and Environment, Canada. Briggs, D. R. 1928. The determination of the C-potential on cellulose - a method. Journal of Physical Chemistry. 32: 641 - 675. Brown, G. H., C. N. Hoyler and R. A. Bierwirth. 1947. Theory and application of radio-frequency heating. D. Van Nostrand Company, Inc., New York. I l l Brown, J. H., R. W. Davidson and C. Skaar. 1963. Mechanism of electrical conduction in wood. Forest Products Journal. 13 (10): 455 - 459. Cao, L. J., Z. Q. Liu, Y. X. Liu and C. Y. Dai. 1986. Study on the dielectric constant of wood. Journal of Northeast Forestry University. 14(3): 57 - 66 (in Chinese). Dai, C. Y., Y. X. Liu, Z. Q. Liu and L. J. Cao. 1989. Study on the parameters of dielectric loss of wood. Journal of Northeast Forestry University. 17(3): 42 - 47. (in Chinese). Gebhart, B. 1993. Heat conduction and mass diffusion. McGraw-Hill, Inc., New York. Hargis, L. G. 1988. Analytical chemistry. Prentice Hall. Englewood Cliffs, New Jersey 07632. Harris, R. A. and Andy W. C. Lee. 1985. Properties of white pine lumber dried by radio-frequency/vacuum process and conventional kiln process. Wood and Fiber Science. 17(4): 549 - 552. Harris, R. A. and M. A Taras. 1984. Comparison of moisture content distribution, stress distribution, and shrinkage of red oak lumber dried by a radio-frequency/vacuum drying process and a conventional kiln. Forest Products Journal. 34(1): 44-54. Harris, R. A. 1988. Dimensional stability of red oak and eastern white pine dried by radio-frequency/vacuum and conventional drying process. Forest Products Journal. 38(2): 25 -26. Hearmon, R. F. S. and J. N. Burcham. 1954. The dielectric properties of wood. Department of Science and Industry research, London. Forestry Products Research Special Report No. 8. Holmes, S. and C. J. Kozlik. 1989. Collapse and moisture distribution in presteamed and kiln-dried incense-cedar squares. Forest Products Journal. 39(2): 14 - 16. Hosie, R. C. 1976. Native trees of Canada. Canadian Forestry Service. James, W. L. 1975. Dielectric properties of wood and hardboard: variation with temperature, frequency, moisture content and grain orientation. USD A, Forest Service Research Paper, FPL 245. James, W. L. 1977. Dielectric behaviour of Douglas-fir at various combinations of temperature, frequency, and moisture content. Forest Products Journal. 27(6): 44 - 48. James, W. L. 1981. Influence of electrode design on measurements of dielectric properties of wood. Wood Science. 13(4): 185 - 198. 112 James, W. L. 1983. Dielectric properties of lumber loads in a dry kiln. USD A, Forest Service Research Paper, FPL 436. James, W. L. and D. W. Hamill. 1965. Dielectric properties of Douglas-fir measured at microwave frequencies. Forestry Products Journal. 15(2): 51-56. Kozlik, C. J. 1970. Problems of drying western hemlock heartwood to a uniform final moisture content. Proceedings of the 21st Annual Meeting of the Western Dry Kiln Clubs, Missoula, Montana, pp. 55 - 61. Kozlik, C. 1981. Shrinkage of western hemlock heartwood after conventional and high-temperature kiln-drying. Forest Products Journal. 31(12): 45 - 50. Kraszewski, A. 1977. Journal of Microwave Power. 12(3): 215 - 222. Lee, Andy W. C and R. A. Harris. 1984. Properties of red oak lumber dried by radio-frequency/vacuum process and dehumidification process.Forest Products Journal. 34(5): 56 - 58. Lever, A. E. 1966. The plastics manual (3rd edition). The Scientific Press Ltd., London. Li, M. 1996. A study on the effects of air gaps, air velocities and fan revolutions on the drying characteristics of western hemlock baby squares. MSc Thesis. University of British Columbia. Lin, R. T. 1965. A study on the electrical conduction in wood. Forest Products Journal. 15(11): 506 - 514. Lin, R. T. 1967. Review of the dielectric properties of wood and cellulose. Forest Products Journal. 17(7): 61 - 66. Lin, R. T. 1973. Wood as an orthotropic dielectric material. Wood and Fiber. 5(3): 26 - 236. Metaxas, A. C. and R. J. Meredith. 1983. Industrial microwave heating. IEE Power Engineering Series No. 4, Peter Peregrinus Ltd., London. Meyer, R. W. and G. M. Barton. 1971. A relationship between collapse and extractives in western red cedar. Forest Products Journal. 21(4):58 - 60. Miller, D. G. 1966. Radio-frequency lumber drying: methods, equipment and costs. Canadian Forest Industries. June, 1966: 53 - 57. Miller, D. G. 1969. The state of the art in electronic drying of wood. Canadian Forest Industries. 113 December, 1969: 42-43. Miller, D. G. 1971. Combining radio-frequency heating with kiln-drying to provide fast drying without degrade. Forest Products Journal. 21(12): 17-21. Miller, D. G. 1972. Further report on combining radio-frequency heating with kiln-drying. Forest Products Journal. 23(7): 31 - 32. Nanassy, A. J. 1970. Overlapping of the dielectric relaxation spectra in oven-dry yellow birch at temperatures from 20 to 100°C. Wood Science and Technology. 4(2): 104 - 121. Nanassy, A. J. 1972. Dielectric measurement of moist wood in a sealed system. Wood Science and Technology. 6: 67-77. Nelson, S. O. and A. W. Kraszewski. 1990. Dielectric properties of materials and measurement techniques. Drying Technology. 8(5): 1123 - 1142. Norimoto, M. 1976. Dielectric properties of wood. Wood Research. No. 59/60: 106 - 152. Peterson, R. W. 1951. Dielectric heating as applied to the woodworking industries. Department of Northern Affairs and Natural Resources. Forestry Branch. Bulletin No. 110. Peyskens, E., M. de Pourcq, M. Stevens and J. Schalck. 1984. Dielectric properties of softwood species at microwave frequencies. Wood Science and Technology. 18: 267 - 280. Pound, J. 1966. Radio-frequency drying of timber. Wood. 12: 43 - 45. Pound, J. 1973. Radio frequency heating in the timber industry. E. F. N. Spon Ltd., London. Schiffmann, R. F. 1987. Microwave and dielectric drying. Handbook of Industrial Drying, pp. 327 - 356. Skaar, C. 1948. The dielectrical properties of wood at several radio frequencies. New York State College of Forestry Technical Publication No. 69 Siau, J. F. 1995. Wood: influence of moisture on physical properties. Department of Wood Science and Forest Products, Virginia Polytechnic Institute and State University. Siau, J. F. 1984. Transport process in wood. Springer-Verlag. Berlin, Heidelberg, New York, Tokyo, 1984. Simpson, W. T 1980. Radio-frequency dielectric drying of short lengths of northern red oak. USD A, Forest Service Research Paper, FPL 377. 114 Simpson, W. T. (Ed.) 1991. Dry kiln operator's manual. USD A, Madison, Wisconsin. Tinga, W. R. and S. O. Nelson. 1973. Dielectric properties of materials for microwave processing - Tabulated. Journal of Microwave Power. 8(1):23 - 65. Torgovnikov, G. I. 1993. Dielectric properties of wood and wood-based material. Springer-Verlag, Berlin Heidelberg, New York. Trofatter, G., R. A. Harris, J. Schroeder and M. A. Taras. 1986. Comparison of moisture content variation in red oak lumber dried by a radio-frequency/ vacuum process and a conventional kiln. Forest Products Journal. 36(5): 25 - 28. Venkateswaran, A. and S. G. Tiwari. 1964. Dielectric properties of moist wood. Tappi. 47(1): 25 -28. Vermaas, H. F. 1971. The dielectric properties of wood. Ph. D Thesis. University of Stellenbosch. Vermaas, H. F. 1973. Regression equations for determining the dielectric properties of wood. Holzforschung. 27 (1973): 132 - 136. Vermaas, H. F., J. Pound and K. B. Borgin. 1974. The loss tangent of wood and its importance in dielectric heating. South African Forestry Journal. 89: 5-8. Von Hippie, A. R. (Ed.) 1954. Dielectric materials and applications. The Technology Press ofM. I. T. And John Wiley & Sons, Inc., New York; Chapman & Hall, Ltd., London. Wittkopf, J. J. and M. D. Macdonald. 1949a. Dielectric properties of ponderosa pine at high frequencies. Engineering Experiment Station, Oregon State System of Higher Education, Oregon State College, Corvallis. Bulletin No. 29. Wittkopf, J. J. and M. D. Macdonald. 1949b. Dielectric properties of Douglas fir at high frequencies. Engineering Experiment Station, Oregon State System of Higher Education, Oregon State College, Corvallis. Bulletin No. 29. Yavorsky, J. M. 1951. A review of electrical properties of wood. New York State College of Forestry Technical Publication No. 73. Zaky, A. A. and R. Hawley. 1970. Dielectric solids. Routledge & Kegan Paul Ltd., London; Dover Publicationa Inc., New York. Zwick, R. L. (Principal investigator). 1995. The COFI radio frequency/vacuum kiln project. The Council of Forest Industries, Vancouver. 115 

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:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0075272/manifest

Comment

Related Items