A new method of determining moisture gradient in wood

Transcription

1 A new ethod of deterining oisture gradient in wood Zhiyong Cai Abstract Moisture gradient in wood and wood coposites is one of ost iportant factors that affects both physical stability and echanical perforance. This paper describes a ethod for easuring oisture gradient in luber and engineering wood coposites as it varies across aterial thickness. This innovative ethod eploys a colliated radiation bea (x rays or rays) eitted fro a radiation source that discretely scans the aterial in steps or continuously through its thickness direction. A radiation detector on the other side detects intensities of the transitted bea, which directly relates to the aterial density profile through scanning direction. Moisture gradient can be calculated by contrasting it to the ovendried density profile. A series of solid wood saples were tested to verify the accuracy of this technique. The estiated oisture content (MC) gradients easured using the radiation ethod were copared with the actually easured MC gradients using the ovendry ethod of icrotoed sections. The high correlation between the two MC gradient ethods proved that the radiation ethod could provide an accurate and propt estiation of internal wood MC gradient. Moisture gradients in hygroscopic coposite aterials such as wood are known to affect internal stresses that cause diensional changes and eventual developent of defects. Severe deforation and other defects of finished products reduce product quality and have the potential to daage a anufacturer s reputation and significantly increase the cost of anufacturing. One of the oldest ethods for easuring oisture gradient in wood was the bandsaw slicing technique (McMillen 955). This technique is not entirely accurate because of kerf and oisture losses caused by heat generated during high-speed cutting. Methods of slicing thin layers were iproved later by using a drill bit (Feng and Suchsland 993) and icrotone knife (Wand and Youngs 996). Because the saples were destroyed after easuring oisture gradient, these ethods failed to provide a continuous onitoring of oisture oveents inside saples. This ade it difficult to develop accurate odels of oisture gradient and oveent during wood drying. The search for a nondestructive ethod of easuring oisture in wood started in early 900. The electrical resistance of wood was found to be a good indicator of oisture content (MC) (Myer and Rees 926) and was widely used after an extensive study on the electronic properties of wood (Jaes 963). Instead of using a single paraeter (electrical resistance), any advanced devices eployed radio frequency signals to easure MC in wood (Dennis and Beall 977, Steele and Cooper 2006). These ethods were based priarily on principles of electrical resistance, capacitance, and phase. Because the relationship between MC in wood and its electrical properties were not fully understood, the radio frequency technique was dependent on experiental data to create an epirical relationship for each species. Radiation techniques were used to deterine MC and density of wood and other aterials (Loos 96, 965; Gardner et al. 972; Spolek and Plub 98). The principle of these radiation techniques is that penetration of the radiation ray (x ray, beta ray, and gaa ray) into aterials is dependent on gross density of the aterial and how deep radiation penetrated into the aterial. Estiated density is the average value of volue where radiation penetrates. For nonunifor aterials, usually the radiation shapes (spot, plane, or plue) ight have soe effects on estiated density. A narrow bea of onoenergetic photons with an incident intensity I 0, penetrating a coposite aterial (i.e., wood) with transitted distance t and density, eerges with intensity I given by the exponential attenuation law: i i I = I t [] 0 e i where µ i is ass attenuation coefficient of ith coponent with density of i in the coposite. The ass attenuation coefficient is a basic quantity used in calculations of penetration The author is a Materials Research Engineer, USDA Forest Serv., Forest Product Lab., Madison, Wisconsin (zcai@fs.fed.us). This paper was received for publication in Noveber Article No Forest Products Society Meber. Forest Products Society Forest Prod. J. 58(7/8):4 45. FOREST PRODUCTS JOURNAL VOL. 58, NO. 7/8 4

2 and energy deposition by photons. Given energy of incident radiation depends on aterial copositions and can be obtained by siple addition, such as cobining values for the eleents according to their proportions by weight. By coparing incident intensity I o and transitted intensity I, the exponential attenuation law (Eq. []) will deterine the density of a given specien with a known ass attenuation coefficient and the penetration thickness. The relationship between ass attenuation coefficient and oisture is used to deterine MC. Because radiation covers a large portion of the specien, the ethod is used to estiate average MC throughout a given volue of wood and is related to size of the radiation plue. When shrinkage initiates as the wood dries to below the fiber-saturation point, MC in wood is no longer linear with density. Therefore, to assure accuracy, all ethods require that separate radiation odels be used above and below the fiber saturation point. Hattori and Kanagawa (985) estiated oisture in wood with a edical x-ray coputer toograph (CT) scanner. In this study, the CT scanner transitted the x ray through wood saples in their longitudinal direction. The CT iages of 35 saples with different MCs (fro wet to ovendried) but the sae diensions (2 c by 2 c by c) were analyzed. The CT nubers within the 2-c by 2-c square of each saple were used to correlate to average MC of the saple. After establishing a regression odel to accurately estiate average MC of sall saples, Kanagawa and Hattori (985) used the sae CT scanner to investigate oisture distributions of two 0.5-c square tiber saples during conventional kilndrying. The CT iages were taken through the saples longitudinal direction on the 0.5-c square radial tangential surface. The iages were then eshed into 20 by 20 grids of sall squares (a series of about 0.5-c grid squares). Instead of using the basic radiation attenuation law, CT nubers in each sall square were counted and used to estiate oisture distributions aong the sall squares based on the previously discussed average MC-regression odel (Hattori and Kanagawa 985) developed for the entire piece. Moisture distributions estiated using CT iages were actually the average oisture in the 400 sall grid areas. Moreover, any such regression odel will be soewhat dependent on species and the CT scanner. Xu et al. (996) investigated a procedure to deterine water-absorption distribution in particleboard, ediu density fiberboard (MDF), and oriented strandboard (OSB). The procedure was based on direct easureent of the vertical density distribution fro a gaa-ray densitoeter before and after water soak and vertical-density distribution after the water-soaked speciens had been reconditioned to their presoak weights. The researchers then used these data sets to separate wood ass and water fro different vertical density distributions. Thus, Xu et al. estiated water absorption inside the saple. To separate water fro wood ass, the ethod assued that wood within a divided volue before soaking is the sae location and ass as wood after soaking. However, after water soaking, each layer inside the saple expanded evenly. It is possible to divide the saple into layers so that each pair of corresponding layers before and after water soaking have the sae wood weights, but it is ipossible to precisely locate those layers after soaking so that they are identical to those before soaking. Moisture gradient in coposite aterial (wood) is one of iportant physical paraeters that needs to be accurately deterined if we are to better understand internal stresses and diensional stability. The priary objective of our study was to investigate radiation techniques that could be used to deterine oisture-content gradient in wood. Methodology Wood-based coposites usually consist of cellulose, heicellulose, lignin, extractive, and other cheical additives (e.g., resin and wax). Lindgren (99) and Macedo et al. (2002) investigated the effect of wood cheical coposition on ass attenuation coefficients. After studying the ratios 50/ 25/25, 25/25/50, and 0/0/80 for cellulose/heicellulose/ lignin, Lindgren (99) found that the average value of the ass attenuation coefficient was c with a 0.4 percent coefficient of variation (COV). These researchers concluded that no significant changes were found in the ass attenuation coefficient because of coposition changes. Laufenberg (986) found that a sall aount of resin and wax in coposites had very liited influence on the ass attenuation coefficient. However, the presence of water in saples was found to affect the coefficient (Lindgren 99, Macedo et al. 2002). Usually, ass attenuation coefficient is dependent upon the energy of incident radiation. By selecting a proper incident radiation energy level, the ass attenuation coefficient of water could be about the sae as that of wood. Equation [] for wood with certain MC will then be: I = I 0 e t [2] where µ is ass attenuation coefficient. After calibrating a given aterial for its ass attenuation coefficient, overall density of the aterial could be calculated using the radiation ethod: = ln 0 I [3] t I According to the Wood Handbook (USDA Forest Serv. 999), wood density is defined as: W wood + W water = [4] V where W wood = weight of ovendried wood; W water = weight of water; and V = volue at the MC. The MC of wood on a dry basis is defined as: W water = 00% [5] W wood Wood shrinkage norally begins at about the fiber saturation point (usually 30%) and is continuous in a fairly linear anner until the wood is copletely dry. Shrinkage of any segent of wood fro the green condition (above the saturation point) to any MC is reported and described (USDA Forest Serv. 999): 30 S = S 0 [6] 30 and V = V S [7] green 42 JULY/AUGUST 2008

3 where S is voluetric shrinkage fro the green condition to MC (<30%), S 0 is total voluetric shrinkage fro the green to ovendry condition, and Vgreen is volue easured at the green condition. After taking into account the shrinkage when MC is below 30 percent, aterial density in Equation [2] can be rewritten as: W wood = = wood [8] V S S green where wood is wood density based on ovendried weight and green volue. The exponential attenuation law (Eq. [2]) can be rewritten as: + 00 ln 0 I = t wood [9] I S For a given wood saple with predeterined paraeters such as density wood, total shrinkage S 0, and ass attenuation coefficient µ, MC can be calculated using: t wood S 0 ln I 0 I = [0] S 0 t ln 0 I wood 30 I 00 Another practical way to deterine MC of the saple is to copare exponential attenuation relationships between the ovendried condition and certain oisture conditions. Specifically, fro Eq. [8] we have the density of the saple when it is ovendried: 0 = wood S 0 [] Siilarly, we have the density of the saple when it has MC : + 00 = wood S [2] If both ovendried density 0 and density with MC in Equations [] and [2] are deterined fro the radiation ethod using Equation [3], MC of the saple can be calculated fro: when 30%: when < 30% 0 = 00 [3] 0 0 = 00 [4] 0 0 S S 0 During wood drying, the saple is usually drying fro the green condition to a target MC. The ovendried exponential attenuation cannot be easured until the saple is ovendried. Figure. Typical configuration of radiation easureent of vertical density profiles for a flat-sawn specien across its thickness (radial direction). Alternatively, the ovendried density 0 can be estiated earlier by perforing the radiation test on the edge of sideatched saples, and the calibrated odel could be used to deterine the saple MC during drying. All radiation easureent techniques require a source of radiation and a radiation detector. A typical source detector configuration for easuring density profile across a flat-sawn board thickness is shown in Figure. Usually, the radiation bea is passed through a slit to provide a colliated x-ray bea that penetrates the saple. Diension of the saple is typically 50.8 by 50.8 by 20.3, which is in L by T by R directions. The colliated x-ray bea penetrates the saple through either its longitudinal or tangential directions. The colliated radiation bea then increentally scans the saple at ultiple sites through the thickness direction (usually radial). The scanning result will produce a continuous density curve which is called a vertical density profile (VDP). Figure 2 shows typical ovendried density 0 profile and profile for the sae saple at MC. Coparing the two profiles, Equations [3] and [4] can be used to deterine oisture gradient along the thickness direction. We noticed that saple thickness changed when MC of the saple was different, especially when the MC is below the fiber saturation point. Thus, it is necessary to precisely ap the two profiles to ensure that 0 and atch. Equation [6] can be used to copensate thickness shrinkage according to the oisture profile. Test procedures and results Eight solidwood saples (four oak saples, two pine saples, and two spruce saples) with noinal diension of 50.8 by 50.8 by 20.3 (L by T by R directions) were equilibrated at 22 C and 65 percent relative huidity (RH) and tested to verify the technique previously discussed. Equilibrated saples were then edge-sealed and ovendried for 24 hours under teperatures of 05 C. The saples ovendried density profiles were easured using QMS x-ray density profiler (QMS, Knoxville, Tennessee). Then saples were placed in water with one L by T face suberged and about half of each saple block iersed in water to purposely induce oisture gradients in the radial direction of the QMS is a tradearked product. Any reference in this paper is not an endorseent but provided for the benefit of readers. FOREST PRODUCTS JOURNAL VOL. 58, NO. 7/8 43

4 Figure 2. Vertical density profiles of a saple when it is wet and ovendried. saples. Saples were then taken out at different tie intervals for density-profile easureents as shown in Figure. Each saple had to be diensionally easured and weighed before it was placed in the QMS x-ray density profiler to test. This is the potential biggest hurdle to applying this technique in wood drying, because the sall saples could lose considerable oisture during the transportation fro a dry kiln to the density profiler. Figure 3 shows density profiles at different intervals for a typical oak saple with inducing oisture penetration fro one side. After 7 days, all saples were sliced in the radial direction with a icrotoe knife. Each sliced layer (about - thick) was carefully arked and ovendried to deterine individual MC of each icrotoed laina according to Eq. [5]. These data were then plotted across the radial direction to estiate actual oisture gradient. About 2 to 4 layers were sliced fro the surface that was iersed in water. Figure 4 shows the easured oisture gradient and estiated oisture gradients for the sae specien of Oak #. The easured oisture gradient was deterined using the icrotoe ovendry ethod after 7-day iersion in water. Estiated oisture gradients were deterined using VDP inforation obtained using the x-ray profiler at various iersing intervals and could be used to estiate oisture gradients over the course of iersions. Figure 4 shows that estiated gradients see uneven and overly sensitive. The reason is that saple thickness changes as wood absorbs oisture (below the saturation point) and it is ipossible to exactly atch 0 and when thickness changes. Although Equation [6] was used to copensate thickness shrinkage according to the oisture profile, the profile could slightly offset the 0 profile. In addition, diension changes (i.e., warp, twist, and cup) during oisture absorption and density differences between late- and earlywood could also ake oisture gradient uneven. Coparison between estiated gradient and actual easured gradient after 7-day iersion in water indicated that the two gradients atched well. After the estiated MC gradient was soothed using a polynoial regression ethod, estiated MC gradient was thought to ore accurately atch the easured gradient. Figure 5 shows the soothing MC gradient and Figure 6 presents high correlation between the two MC gradients. The siilar procedure was used to analyze data for other saples and species. Table shows correlation coefficients between the estiated MC gradient and actual easured MC gradient for the eight saples before and after Figure 3. Density profiles of a typical oak saple when it was soaked in the water fro one side for different intervals. Figure 4. Estiated oisture content (MC) gradients and actually easured MC gradient after 7-day soak in water. Figure 5. Coparison between estiated and easured MC gradients. soothing the MC profile. The high correlation (average R- square after soothing is 0.97) indicates that the radiation ethod discussed in this paper provides an accurate and rapid estiation of MC gradient. The new ethod could provide a very useful tool for understanding internal oisture oveent and oisture-related stress developent. The propt inforation about internal MC gradient could help to dynaically control wood drying processes. Inforation about inside oisture distribution 44 JULY/AUGUST 2008

5 Figure 6. Linear regression between estiated and easured MC gradients. Table. Regression coefficients between the estiated and easured MC gradient. Linear regression Linear regression after soothing Wood ID Slopes Intercepts R-square Slopes Intercepts R-square Oak # Oak # Oak # Oak # Pine # Pine # Spruce # Spruce # could also assist in understanding warping perforance of wood coposites and allow engineers to find solutions to iniize product-diensional instability. Suary and conclusion We have described a new ethod using a radiation ethod to deterine oisture gradient of a rectangular- or squareshaped aterial. This inventive ethod eployed a colliated radiation bea (i.e., x rays or rays) fro a radiation source. The radiation bea scans aterial through its thickness direction continuously or discretely. A detector with the sae scanning speed is placed on the other side of the aterial and detects the transitted radiation bea accordingly. Because the linear attenuation coefficient between the transitted and incident radiation bea is related to density, the signal fro the detector will provide the aterial density profile through scanning direction. After obtaining the density profile, oisture gradient can be calculated on the basis of diensional copensation and the previously deterined or estiated dried-density profile. Calculated oisture gradient is highly correlated through a polynoial regression technique to actual easured oisture gradient. The new ethod can be used to onitor internal oisture oveent in luber to provide propt inforation about internal stresses that could result in warping and defect developent. The ethod could be useful to dynaically control the kiln-drying process by accounting for internal oisture gradient. This technique could reduce drying tie and defects and thereby lower drying energy and iprove luber value. Literature cited Dennis, J.R. and F.C. Beall Evaluation of a new portable radiofrequency oisture eter on luber with drying gradients. Forest Prod. J. 27(8): Feng, Y. and O. Suchsland Iproved technique for easuring oisture content gradients in wood. Forest Prod. J. 43(3): Gardner, W.H., G.S. Capbell, and C. Calissendorf Systeatic and rando errors in dual gaa energy soil bulk density and water content easureents. Soil Sci. Soc. of A. Proc. 36: Hattori, Y. and Y. Kanagawa Nondestructive easureent of oisture distribution in wood with a edical x-ray CT scanner I. Accuracy and influencing factor. J. Japan Wood Res. Soc. 3(2): Jaes, W.L Electric oisture eters for wood. USDA Forest Serv. Res. Note FPL-08. Forest Products Lab., Madison, Wisconsin. Kanagawa, Y. and Y. Hattori Nondestructive easureent of oisture distribution in wood with a edical x-ray CT scanner II. Changes in oisture distribution with drying. J. Japan Wood Res. Soc. 3(2): Laufenberg, T.L Using gaa radiation to easure density gradients in reconstituted wood products. Forest Prod. J. 36(2): Lindgren, L.O. 99. Medical CAT-scanning: X-ray absorption coeffiecients, Connecticut-studies and their relations to wood density. Wood Sci. and Tech. 25: Loos, W.E. 96. The relationship between gaa-ray absorption and wood oisture content and density. Forest Prod. J. (3): A review of ethods for deterining oisture content and density of wood by nuclear radiation techniques. Forest Prod. J. 5(3): Macedo, A., C.M.P. Vaz, J.C.D. Pereira, J.M. Naie, P.E. Cruvinel, and S. Crestana Wood density deterination by x- and gaa-ray toography. Holzforschung 56: McMillen, J.M Drying stresses in red oak. Forest Prod. J. 5():7 76. Myer, J.T. and L.W. Rees Electrical resistance of wood with special reference to the fiber-saturation point. Tech. Bull. No. 9. New York State College of Forestry at Syracuse Univ. Spolek, G.A. and O.A. Plub. 98. Capillary pressure in softwoods. Wood Sci. and Tech. 5: Steele, P.H. and J.E. Cooper Moisture and density detector (MDD). United States Patent No. 7,068,050. Wand, H. and R.L. Youngs Drying stress and check developent in the wood of two oaks. IAWA J. 7():5 30. USDA Forest Serv., Forest Products Lab Wood Handbook: Wood as an Engineering Material. GTR-FPL-3. Forest Product Lab., Madison, Wisconsin. 463 pp. Xu, W., P.M. Winistorfer, and W.W. Moschler A procedure to deterine water absorption distribution in wood coposite panels. Wood and Fiber Sci. 28(3): FOREST PRODUCTS JOURNAL VOL. 58, NO. 7/8 45