Density (D) and moisture content (MC) are two important physical properties of wood and bamboo, which are highly correlated with many other physical and mechanical properties. In this study, the X-ray computed tomography (CT) technique was used to determine the D and MC of poplar (Populus xiangchengensis) and bamboo (Phyllostachys edulis). There was a statistically significant difference in the CT-measured numbers for D and MC between these species. The D-CT and MC-CT linear models for both species were independently established: Dpoplar = 0.00098 × H + 1.02603, Dbamboo = 0.00118 × H + 0.98684, MCpoplar = 0.00309 × H + 1.89982, and MCbamboo = 0.00131 × H + 0.31488, where H is the CT number. The determination coefficients, R2, of the models were all higher than 0.97. Additionally, the R2 values obtained for model validation were also all higher than 0.97. These results indicated that it is feasible to reliably determine D and MC of wood and bamboo using the X-ray CT technique. This study aims to provide reference data for CT detection of the D and MC of wood and bamboo.Abstract
Forest resources are important renewable ecofriendly materials and play a key role in economic development. Wood and bamboo are two important components of forest resources. With the increasing scarcity of forest resources (Dong et al. 2008, State Forestry Administration of the People's Republic of China 2014), it is important to improve the comprehensive utilization rate of wood and bamboo for the sustainable development of the wood industry in China. Accordingly, a thorough knowledge of the physical properties of wood and bamboo and the measurement of such physical properties are critical to ensure that these materials are used to maximum efficiency. Density (D) and moisture content (MC) are two physical properties of major importance in the utilization of wood and bamboo. For industrial uses (e.g., construction, furniture, decoration, pulp, and paper), D and MC are two determinants of the quality as they are highly correlated with the physical and mechanical properties of wood and bamboo, such as strength, toughness, shrinkage, impregnation, and durability. Krauss (2009) analyzed the D and mechanical properties of 100 timber species and found that the correlation between wood D and mechanical properties was approximately a power function. Branco et al. (2014) found that with the increase of the MC, the physical and mechanical properties of wood decreased.
Computed tomography (CT) techniques can be used to measure internal properties of test objects in a nondestructive and noninvasive way. Currently, there are several CT techniques that use X-ray, gamma-ray, ultrasound, microwave, etc. (Bucur 2003). Among these radiation techniques, X-ray CT is widely used due to its higher image resolution compared with other radiation techniques. Additionally, according to Planck's law, X-ray radiation has sufficient energy (over 100 eV) to penetrate test objects (Wei et al. 2011). The X-ray CT technique was first developed for medical applications (Cormack 1963, Hounsfield 1973). Since the 1980s, the use of the X-ray CT technique as a tool to measure the D and MC of wood and bamboo has received considerable attention. Lindgren (1991) found a linear relationship between the D and CT number, and that at the same D, the CT number clearly changes with different MCs. Other studies have also reached similar conclusions. Hou et al. (2009) determined the green D of balsam fir (Abies balsamea) and eastern beech (Fagus grandifolia) using the X-ray CT technique. Steffenrem et al. (2009) and Witomski et al. (2010) developed linear models for the CT numbers and D of Norway spruce (Picea abies) at 12 percent MC and Scots pine (Pinus sylvestris) at 6 percent MC. Gu et al. (2010) fitted a linear model for the air-dry D and CT numbers of 12 wood species. Shan et al. (2014) also found a significant correlation between the CT numbers and air-dry D of bamboo (Dendrocalamus farinosus). Fromm et al. (2001) measured the D distribution and water content in the xylem of spruce (Picea abies) and oak (Quercus robur) by the X-ray CT technique. He and Qi (2013) reported the linear relationship between the MC and CT number in heartwood and sapwood of Dahurian larch (Larix gmelinii).
Most previous studies on the detection of the D, MC, and CT number of wood (or bamboo) by CT mainly focus on the wood (or bamboo) at a certain MC (e.g., green, air-dry, ovendry, etc.). In the present study, the relationship between the D, MC, and CT number of native Chinese poplar (Populus xiangchengensis) and Moso bamboo (Phyllostachys edulis) was investigated at 10 stages of MC. Additionally, the differences between the models for species were analyzed. This study aims to provide reference data for determining the D and MC of wood and bamboo using the X-ray CT technique.
Materials and Methods
Sample preparation
Populus xiangchengensis trees of four ages (12, 13, 14, and 15 yr old) were collected from Ya'an City, Sichuan Province, China. The sampled Populus xiangchengensis trees averaged 21.0 to 25.4 cm in diameter at breast height. Samples of Populus xiangchengensis trees with dimensions of 20 (length) by 20 (width) by 20 mm (thickness) were taken from 500-mm-thick discs at breast height (1.3 m; Standardization Administration of China [SAC] 2009). Phyllostachys edulis plants of 7 ages (2, 4, 6, 8, 10, 12, and 14 yr old) were collected from Huangshan City, Anhui Province, China. Phyllostachys edulis plants were sampled at a breast height diameter of 9.0 to 11.3 cm. Samples of Phyllostachys edulis plants with dimensions of 10 (length) by 10 (width) by t mm (thickness; t represents the natural thickness of the bamboo culm) were taken from 1.5 m above the ground (SAC 1995). For each species, the prepared samples were randomly divided into two groups, which were used to independently establish and validate the fitted models.
Measurement of the density and moisture content
Two methods, an ovendry method and a saturated salt solution method, were used to control the MCs of wood and bamboo samples. The control process of MC was as follows: (1) the prepared samples were labeled and then saturated with water; (2) subsequently, the samples were transferred into an oven at a temperature of 103°C ± 2°C and dried for 1 and 2 hours, respectively; (3) then, the samples were, in turn, placed in six desiccators with six saturated salt solutions (i.e., KNO3 [94.62% relative humidity {RH}], KCl [85.11% RH], NaCl [75.47% RH], NaBr [59.14% RH], K2CO3 [43.16% RH], and CH3COOK [23.11% RH]) and conditioned to the corresponding equilibrium MC (laboratory temperature of 20°C to 22°C); and (4) finally, the samples were further oven-dried at a temperature of 103°C ± 2°C to 0 percent MC. The D and MC under each moisture stage were measured, referring to ISO 13061-1: 2014 (International Organization for Standardization 2014a) and ISO 13061-2: 2014 (International Organization for Standardization 2014b) standards. The mass was measured to 0.001 g using an electronic balance and the dimensions were determined to 0.01 mm using a vernier caliper.
X-ray CT scanning
According to the Lambert-Beer exponential law (Davis and Wells 1992), the attenuation of the radiation is described by
where I denotes the intensity of transmitted radiation (J/cm2·s), I0 represents the intensity of the incident radiation (J/cm2·s), e is the Euler's constant (e = 2.718), μ is the linear attenuation coefficient (cm−1), and t is path length (cm).
Then based on the theorem of Radon (1986), the attenuation coefficient at any point within the tomographic images can be obtained (Fig. 1) as follows:
where I (u,θ) denotes the transmitted intensity at different angles (J/cm2·s), ϕ is the incident angle (rad), and I0, t, and μ represent the same parameters as in Equation 1.
Citation: Forest Products Journal 70, 2; 10.13073/FPJ-D-20-00001
The calculated X-ray linear attenuation coefficient is normalized to the corresponding linear attenuation coefficient for water according to Equation 3 (Davis and Wells 1992, Lindgren et al. 1992). This normalized value is referred to as the CT number:
where HCT is the CT number of the test object (Hu), μmaterial is the linear attenuation coefficient of the test object (cm–1), and μwater is the linear attenuation coefficient of water (cm–1).
The CT scanning was performed simultaneously during the measurement of the D and MC. CT images were collected using a medical BrightSpeed Excel Select 4slices CT scanner (GE Healthcare, Chicago, Illinois, USA). The scanning parameters were set according to previous studies (Freyburger et al. 2009; Wang et al. 2016, 2019) as follows: peak voltage of 120 kV, tube current of 160 mA, scan thickness of 1.25 mm, scan distance of 1.25 mm, field of view of 9.6, reconstruction algorithm of “std,” and laboratory temperature of 20°C to 22°C. Fifteen and seven CT images were obtained for each specimen of Populus xiangchengensis and Phyllostachys edulis, resulting in a total of 3,600 and 1,960 slices at 10 stages of MC, respectively. The measurement unit in the CT scanner software was used to measure the mean CT number for each CT image. The CT number was measured along the axial direction of the wood and bamboo samples.
Statistical analysis
The independent samples t test was used to test differences in CT number, D, and MC between both species. The univariate analysis of covariance was used to analyze differences between models for both species. To evaluate the autocorrelation of the models, a standardized residual plot was drawn. The characteristic of the residual distribution was investigated by the Kolmogorov-Smirnov test.
Results and Discussion
CT numbers' characteristics
The CT number of both species decreased with decreasing MC (Fig. 2). This indicated that there was a positive correlation between the CT number, D, and MC. This apparent correlation of the CT number and MC offers a possibility for modeling. Also, as shown in Figure 2, the CT number ranges of the two species were clearly different. The CT numbers of Populus xiangchengensis trees ranged from −110.000 to −630.462 Hu. The CT number range of Phyllostachys edulis plants was 215.497 to −231.417 Hu. The data in Table 1 reveal that there is a statistically significant difference in CT numbers between the two species. Lindgren (1991) found that the wood substance has more influence on the X-ray linear attenuation coefficient than its MC. As shown in Figure 3, the CT images clearly revealed that Phyllostachys edulis contains more wood substance than Populus xiangchengensis. This finding explains the difference in CT numbers between wood and bamboo. Additionally, it also shows that the X-ray CT technique can distinguish the material differences between wood and bamboo.
Citation: Forest Products Journal 70, 2; 10.13073/FPJ-D-20-00001
Citation: Forest Products Journal 70, 2; 10.13073/FPJ-D-20-00001
Model development
The data shown in Tables 2 and 3 revealed that the differences in D and MC between Populus xiangchengensis and Phyllostachys edulis were also statistically significant. Accordingly, it is a rational presumption for independent modeling that there is significant difference between the independent variable sets and the dependent variable sets. Thus, the fitted models for both species were independently developed.
The D-CT models for Populus xiangchengensis and Phyllostachys edulis were as follows:
where D is the density (g/cm3), and H is the CT number (Hu). The determination coefficient values of the D-CT models for Populus xiangchengensis and Phyllostachys edulis were R2 = 0.9975 and R2 = 0.9852, respectively (Figs. 4a and 4c).
Citation: Forest Products Journal 70, 2; 10.13073/FPJ-D-20-00001
The MC-CT models for Populus xiangchengensis and Phyllostachys edulis were as follows:
where MC is the moisture content (%), and H is the CT number (Hu). The determination coefficient values of the MC-CT models for Populus xiangchengensis and Phyllostachys edulis were R2 = 0.9847 and R2 = 0.9764, respectively (Figs. 4b and 4d).
The R2 values of the D-CT and MC-CT models were all higher than 0.97. This suggests that it is feasible to measure the D and MC of Populus xiangchengensis and Phyllostachys edulis using the X-ray CT technique.
Model validation
As shown in Table 4, the differences in the fitted models between both species are statistically significant. This justified the reasonableness of independently modeling for both species. As shown in Figure 5, the standardized residuals of the fitted models were mainly distributed between the intervals (−2, +2), which implied that the fitted models had no obvious autocorrelation (Nobre and Singer 2007). Additionally, the residuals were normally distributed in all models (Table 5). Furthermore, the D-CT and MC-CT fitted models were validated. The R2 values obtained from the model validation were all higher than 0.97 (Fig. 6). These results proved the reliability of the measurement of the D and MC of wood and bamboo by the X-ray CT technique.
Citation: Forest Products Journal 70, 2; 10.13073/FPJ-D-20-00001
Citation: Forest Products Journal 70, 2; 10.13073/FPJ-D-20-00001
Conclusions
In this study, the X-ray CT technique was used to measure the D and MC of Populus xiangchengensis and Phyllostachys edulis. There was statistically significant difference in CT numbers between both species. The D-CT and MC-CT linear models for both species were, respectively, Dpoplar = 0.00098 × H + 1.02603, R2 = 0.9975; MCpoplar = 0.00309 × H + 1.89982, R2 = 0.9847; Dbamboo = 0.00118 × H + 0.98684, R2 = 0.9852; and MCbamboo = 0.00131 × H + 0.31488, R2 = 0.9764, where D is the density, MC is the moisture content and H is the CT number. The R2 values in the model validation were all greater than 0.97. These results suggest that it is feasible to reliably measure the D and MC of wood and bamboo using the X-ray CT technique.
Illustration of X-ray computed tomography scanning.
Variation trend of computed tomography (CT) numbers of (a) Populus xiangchengensis and (b) Phyllostachys edulis at 10 stages of moisture content (MC). The percentages in the figure are the average MC values of the samples using the corresponding control methods.
Comparison and three-dimensional simulation of typical computed tomography images of (a) Populus xiangchengensis and (b) Phyllostachys edulis under different moisture content (MC) stages.
Density–computed tomography (D-CT) and moisture content (MC)-CT fitted models of (a, b) Populus xiangchengensis and (c, d) Phyllostachys edulis. Square icon and round icon indicate D and MC, respectively.
Standardized residuals of the density–computed tomography (D-CT) and moisture content (MC)-CT models: (a) D of Populus xiangchengensis, (b) MC of Populus xiangchengensis, (c) D of Phyllostachys edulis, and (d) MC of Phyllostachys edulis. Square icon and round icon indicate D and MC, respectively.
Validation of the density–computed tomography (D-CT) and moisture content (MC)-CT models: (a, b) Populus xiangchengensis and (c, d) Phyllostachys edulis. Square icon and round icon indicate D and MC, respectively.
Contributor Notes
The authors are, respectively, PhD Candidate, Bamboo and Rattan Sci. and Technol. Lab., International Centre for Bamboo and Rattan, and College of Materials Sci. and Technol., Beijing Forestry Univ., and Research Inst. of Wood Industry, Chinese Academy of Forestry, Beijing, People's Republic of China (wongchingping@outlook.com); and Professor (liuxe@icbr.ac.cn) and Associate Professor (yangsm@icbr.ac.cn [corresponding author]), Bamboo and Rattan Sci. and Technol. Lab., International Centre for Bamboo and Rattan, Beijing, People's Republic of China. This paper was received for publication in January 2020. Article no. 20-00001.