ESTIMATION OF THE DYNAMIC ELASTIC PROPERTIES OF WOOD FROM Copaifera langsdorffi i Desf USING RESONANCE ANALYSIS

Analysis of transverse vibrations has been identifi ed as a simple, inexpensive, fast and effective method of characterizing the elastic properties of several materials, including wood. The objective of this study is to evaluate the dynamic elastic properties of wood from Copaifera langsdorffi i through resonance analysis. A linear correlation was observed, with a coeffi cient of determination of 0.88, between the dynamic modulus of elasticity (Ed), as determined by BING (resonance analysis), and the static modulus of elasticity (E), as determined by a universal testing machine, for wood samples of Copaifera langsdorffi i. A moderate correlation was observed (0.69) between the Ed and the apparent density (D) of the samples. The shear modulus (G) was negatively correlated with D, E, and Ed, yet not signifi cantly. The specifi c modulus, which is the stiffness to density ratio, was linearly correlated with the internal damping. This viscoelastic correlation is rarely reported in literature and indicates, for instance, wood quality for the manufacture of musical instruments. The BING system provided fast, reliable results for estimation of elastic properties, particularly stiffness, in wood of Copaifera langsdorffi i.


INTRODUCTION
Copaiba (Copaifera langsdorffi i Desf.) is a species belonging to the family Leguminosae-Caesalpinoideae which provides moderately dense timber (~700 kg m -³) and has good durability under natural conditions.For this reason, Copaiba wood has been recommended for use in building construction, furniture making and lathed parts.Therefore, investigating the elastic properties of this wood species is critical to defi ning its applications in a way that is suitable and safe.
The mechanical properties of a wood species indicate its capability to withstand efforts and are usually gauged by its strength and stiffness against external forces.The use of a wood species for building construction and structures is conditional on its stiffness properties, given by the modulus of elasticity (MADSEN, 1992).And for this reason, stiffness is one of the most commonly studied mechanical wood properties.
Classic testing determines the static modulus of elasticity (E) using a universal testing machine.As a rule, the static bending test method is performed in expensive Hein, P. R. G. et al.
equipment that demand high maintenance costs, requiring an average 5 to 15 minutes per sample depending on the species and test conditions being used.Therefore, searching for faster, safer alternative methods for determining elastic mechanical properties takes on critical importance within the context of wood science and technology.
Several methods exist which are categorized as nondestructive for determining the modulus of elasticity of various materials.In wood, the most commonly used nondestructive methods for assessing physical, mechanical and elastic properties are based on ultrasound (BALLARIN;NOGUEIRA, 2005;BUCUR, 1984), stress waves (ROSS, 1985), transverse vibration (MURPHY, 1997) and, more recently, near-infrared spectroscopy (SCHIMLECK et al., 2001).According to Bucur (1995), nondestructive methods are also capable of detecting defects in the manufacturing process of plastics, steel and ceramics, which are homogeneous, isotropic materials.In wood, defects occur naturally and can be detected by these methods with reasonable accuracy.
Analysis of transverse vibrations, or resonance, has been identifi ed as the simplest, most inexpensive and effective method of characterizing elastic properties of wood (BRANCHERIAU;BAILLÈRES, 2002).In it, the natural vibration frequency of a material is correlated with its stiffness against bending or compression.This analytical technique has been well described in literature (CALIL JUNIOR;MINÁ, 2003;TARGA et al., 2005) and has been used in the timber industry for the purpose of grading materials.Several studies have demonstrated a strong linear correlation between the dynamic modulus and the static modulus of elasticity as measured in various wood species (BALLARIN et al., 2002;BRANCHERIAU;BAILLÈRES, 2002;CALIL JUNIOR;MINÁ, 2003;GREEN et al., 2004;TARGA et al., 2005;WANG et al., 2001).In addition to industrial applications, vibrational properties can be also used, for instance, for assessing wood quality and potential for the manufacture of musical instruments (OBATAYA et al., 2000).
The objective of this study was thus to evaluate the dynamic elastic properties of wood from Copaifera langsdorffi i Desf.by using resonance analysis.

Sampling
Wood samples of Copaiba (Copaifera langsdorffi i Desf.) were obtained from the campus of Universidade Federal de Lavras, southern Minas Gerais state.The site (21°14' S and 45°00' W) is located at an average altitude of 900 meters, lying in a transition zone between the Cerrado and Atlantic Rainforest biomes, and the local climate falls under the Cwb category according to Köppen classifi cation (PEEL et al., 2007).Twenty-nine samples were cut from a central board one meter in length which had been taken from the trunk base.The defect-free specimens used in bending tests were 300 x 20 x 20 mm in size and had well defi ned tangential and radial surfaces.It was not possible to determine tree age.

Static bending
The modulus of elasticity (E) was determined by means of a bending test procedure following standard D 143-94 (Standard Methods of Testing Small Clear Specimens of Timber) as set out by the American Society for Testing and Materials -ASTM (1997).Tests were performed in a universal testing machine (EMIC DL -30 kN model, EMIC Equipamentos e Sistemas de Ensaio LTDA).

Transverse vibration
The transverse vibration of wood specimens was evaluated in time and frequency domain by a BING system (Beam Identifi cation by Nondestructive Grading, http://www.xylo-metry.org),as illustrated in Figure 1.The samples were placed in elastic supports in order to generate free vibrations.Vibration is produced by lightly taping a wood-tipped hammer on one end of the specimen.Input and output signals were transmitted using a low-pass fi lter and a data acquisition card coupled to a computer and then recorded as described by Brancheriau and Baillères (2003).Figura 1 -Dispositivo experimental usado para medição do tempo e da frequência de vibração transversal das amostras de madeira por meio do sistema BING.Estimation of the dynamic elastic properties of wood ... Spectral signal analysis and selection of individual vibration frequency peaks were performed using the software BING® version 9.1.3.The signal sampling frequency was 39,062 Hz with spectral acquisition of 32,768 points.

Estimation of dynamic parameters
The theoretical model used for describing the resonance of a prismatic, uniform, freely vibrating specimen was proposed by Timoshenko (1921) as follows: Eq. ( 1) where E X is modulus of elasticity (or Young's modulus); I Gz is moment of inertia, v is transverse movement; x is distance along the beam axis; D is density; K is a constant that depends on the geometry of the section (for a rectangular section, K = 5 / 6); G XY is shear modulus; t is time and A is the cross-sectional area of the beam.Using the solution proposed by Bordonné (1989) for this differential equation, the effect of shear is taken into account while the effects of support are ignored.The modulus of elasticity (E X ) and shear modulus (G XY ) can be calculated as follows: Eq. ( 2) where f i is the ith resonance frequency, P n is Bernoulli's solution (n rank) and L is prism length.The calculation of parameters m, P n , Q, F 1 (m) and F 2 (m) and θ is described in Hein et al. (2012).Internal damping (tg δ) (or friction) was calculated according to the equation provided by Brancheriau et al. (2006): where f 1 is fi rst resonance frequency and α 1 is temporal damping associated with f 1 .Only the fi rst frequency was computed because of its high energy.A detailed discussion on different theoretical models about the movement, approximate solutions and respective hypotheses concerning longitudinal and transverse vibrations, as well as the effects of the elastic support used in the BING procedure, is provided in Brancheriau and Baillères (2002).
The samples were kept in a temperature-controlled room at 21°C and 65% relative humidity.Under these conditions, the moisture of wood samples stabilized at 14%.The apparent density (D) of stabilized specimens was given as the ratio of specimen size, as measured with a caliper, to specimen mass, as measured with a scale accurate to 0.001 grams, immediately before the dynamic test.The specifi c modulus (E') was given as the ratio of dynamic modulus of elasticity to apparent density.
Correlations between properties, analyses of variance, linear regressions and prediction errors were determined using the software Statgraphics Plus for Windows v.2.1 (Statistical Graphics Corp., Maryland, USA).

Dynamic and static determination of elastic parameters
Table 1 provides results of elastic parameters against static bending as measured by a universal testing machine, and against dynamic bending as measured by transverse vibration, in wood specimens of Copaifera langsdorffi i.
With moisture content stabilized at 14%, the apparent density of specimens ranged from 709 to 914 kg m -³.Stiffness, as indicated by the modulus of elasticity, had wider variation: 20% under static bending and as measured by the testing machine, and 22.6% under dynamic bending.The mean dynamic modulus of elasticity was higher than the static value, an expected pattern also reported by other authors (BURDZIK; NKWERA, 2002;ILIC, 2001;TARGA et al., 2005).Shear modulus (G) had the greatest variation among samples.Due to the diffi culty in obtaining accurate measurements of internal damping (tg δ) (or friction), this property is rarely reported in literature.The variation (16%<CV<44%) found in the viscoelastic properties of these wood specimens is considered normal.Trees are subjected to many infl uences that include moisture, soil conditions and tree spacing, producing wood with widely varying characteristics.

Correlation between physical and elastic parameters in Copaiba wood
Table 2 illustrates the correlation between dynamic properties in specimens of Copaifera langsdorffi i. Elastic constants measured by destructive (E) and nondestructive (E d ) testing had a correlation of 0.94; these elastic constants had a less signifi cant correlation with apparent density (D) (0.72<R<0.67).Shear modulus (G) had low signifi cance correlations with other properties.The correlation between internal damping (tg δ) and specifi c modulus (E') was -0.61.BAILLÈRES, 2003;CALIL JUNIOR;MINÁ, 2003;HAINES et al., 1996;ILIC, 2003;OHSAKI et al., 2007;TARGA et al., 2005), yet using other wood species.Ilic (2001), for instance, evaluated Eucalyptus wood and found a strong correlation (r=0.99) between the dynamic modulus and the static modulus of elasticity.Again, this author did not fi nd signifi cant correlations between shear modulus (G) and elasticity.Figure 2 illustrates the correlation between the destructive and the nondestructive method for determination of stiffness in wood specimens of Copaifera langsdorffi i. Results are consistent with correlations found in other studies between dynamic and static moduli of elasticity, applying the same methodology (BRANCHERIAU; Estimation of the dynamic elastic properties of wood ...

Viscoelasticity of the wood specimens
Among other purposes, viscoelastic properties are used for evaluating wood quality prior to the manufacture of musical instruments (OBATAYA et al., 2000).According to Ono and Norimoto (1984), woods with a high specifi c modulus and low internal damping are suitable for musical instruments such as pianos and violins.For this reason, the few existing studies exploring this issue were performed using hardwood.Another limiting factor for studying viscoelastic properties is the diffi culty found in accurately measuring internal damping.Hence the correlation between internal damping and specifi c modulus (ratio of dynamic modulus of elasticity to apparent density) in tropical woods being rarely reported in literature.
In this work, an approximately linear correlation was found between the specifi c modulus and the internal damping (Figure 3), indicating that wood from Copaifera langsdorffi i (or at least part of it) has interesting acoustic properties for the manufacture of musical instruments.The E' x tg δ correlation for Copaifera langsdorffi i was found to be very similar to that reported by Ono and Norimoto (1984), who evaluated this correlation in 30 types of hardwood.Obataya et al. (2000) found a similar E' x tg δ correlation for wood from Picea sitchensis.According to Ono and Norimoto (1985), the modulus of elasticity and internal damping of a wood specimen are determined to a large degree by the angle of microfi brils in the S 2 layer of its fi bers.the genera Picea and Acer and found good correlations between the two gauging methods, with deviations of 0.1% to 0.6% between the dynamic modulus and the static modulus of elasticity.The consistency between these previous studies and results found in this work (Table 2) indicates that the BING system is suffi ciently reliable and accurate for evaluating elastic properties in wood.
Table 3 provides the analysis of variance of the linear regression model to describe the correlation of E with E d .The P-value in Table 3 being less than 0.01 is an indication of a statistically signifi cant correlation between E and E d at the 99% confi dence level.The R² statistic indicates that the model fi t explains 88% of the variability in static elasticity.The correlation coeffi cient was 0.94 (Table 2), indicating a strong correlation between the methods of analysis.The standard error of estimate shows a standard deviation of residuals of 980 MPa.This value could be used for setting prediction thresholds in future elasticity estimates based on transverse vibration.

CONCLUSIONS
Despite the limited number of samples investigated in this work, results revealed that the BING system was capable of determining the stiffness as well as the shear modulus and internal damping of wood from Copaifera langsdorffi i.A strong correlation was found (R²=0.88) between the modulus of elasticity determined by transverse vibration and the value determined by the universal testing machine.A moderate correlation was found (0.69) between the E d and the apparent density of the wood specimens.The shear modulus was negatively correlated with density and with elastic constants, yet not signifi cantly.

Figure 1 -
Figure 1 -Experimental device used for measuring time and frequency of transverse vibration in wood samples through the BING system.

Figure 2 -
Figure 2 -Correlation between the static modulus of elasticity (E) as measured by a universal testing machine and the dynamic modulus of elasticity (E d ) as measured by the BING system.

Tabela 3 -
Análise de variância da regressão que descreve a relação entre E e E d . of squares; DF: degrees of freedom; MS: mean square; Calc-F: calculated F; P-Value: comparison value with an α signifi cance level.

Table 3 -
Analysis of variance of the regression describing the E x E d correlation.