Vibration Based NDT Methods to Verify Wood Drying Efficiency

The capability of application of fl exural vibration test in nondestructive evaluation of wood drying was studied. Poplar timbers were conventionally kiln dried under three different drying schedules down to the fi nal moisture content of 12 ±2 %. Dynamic Modulus of Elasticity, damping (internal friction), acoustic coeffi cient, and acoustic converting effi ciency of each sample were evaluated using free fl exural vibration test of both-ends free bars before and after the drying, while dynamic responses in each individual step were also monitored. Results revealed that emerging of some defects in the dried timber caused signifi cant changes through the studied parameters of samples by different extents. Cracks were amongst the most prevalent defects infl uencing the studied factors. The desired schedule, which caused the least abundance of defects in the dried timber, was then identifi ed using this technique. It can be stated that the existing destructive and traditional methods could be replaced by the free vibration method to evaluate the fi nal drying quality accurately.


INTRODUCTION 1. UVOD
Nondestructive testing (NDT) techniques have recently been implemented by many researchers for quality assessment of materials or for identifying their defects (Lee, 2001;Axmon et al., 2004;Baskaran and Janawadkar, 2007;Roohnia et al., 2011a;Biechele et al., 2011).These techniques are effective means of testing and evaluating the properties of materials, which do not destroy the physical, chemical, and mechanical properties of materials and have no infl uence on their future performance.The exploitation and application of this technology have also been quickly developed in wood and wood-products fi elds for its evident advantages.Appropriate correlation between modulus of elasticity and other mechanical properties of wood (Bodig and Jayne, 1993) has encouraged researchers to engage dynamic modulus driven by nondestructive tests to estimate the mechanical properties of wood and wood products (Wang et al., 2008).
In the process of timber manufacturing, drying is the single most demanding step in terms of energy consumption and time.A reduction in drying time and/or an improvement in the quality of the dried timber offers potential economic benefi ts.Improvements in the wood drying process are accompanied by the better understanding of the drying process (Watanabe et al., 2008).Hence, many studies have been conducted to investigate the effect of different wood drying conditions on characteristics of the dried timber (Möttönen, 2006;Milić and Kolin, 2008;Mugabi et al., 2011;Tarmian et al., 2010;Oltean and Teischinger, 2011).All of these researches imply the importance of being aware of wood properties affected by the drying processes.
In general, destructive methods are used for quality controlling of the dried timber (Perré and Roger, 2006).In addition to being time-consuming, destructive nature of these methods makes them infeasible to be applied for the control and inspection of the entire timber.Consequently, common methods do not provide researchers and industrialists with suffi cient information on properties of every individual sample.Meanwhile, some of the defects induced by wood drying schedules, such as occurrence of surface and internal cracks, which are among the most prevalent defects in wood drying processes, cannot be detected by these methods or they might be insuffi cient in precision.
Given the application of non-destructive tests create no change in the properties of wood, the possibility of timely evaluation of each single sample is provided.Measurement of moisture content has been investigated during drying by means of non-destructive methods (Watanabe et al., 2008;Tanaka et al., 2009;Lazarescu et al., 2010;Watanabe et al., 2012).However, the potential of the vibration-based NDT method was extended here for evaluation of the wood properties subjected to different drying conditions.Following the previous studies, this paper aims to identify the vibrational properties of drying timber infl uenced by different drying conditions.

Materijali
Freshly-cut logs of poplar (Populus alba L.) with approximately 17-20 years of growth and 30-35 cm diameter, belonging to the Taleghan region in Iran were selected for the study.The logs were cut through the tangential direction to produce uniform timbers with a nominal thickness of 7 cm, a length of 220 cm, and a width of 14 cm.Some sound samples (i.e.without any crack, knot, etc.) were selected before the drying for fl exural vibration test for each schedule; 6, 6, and 5 samples for three different schedules, respectively, named A, B and C.
Next, their cross sections were immediately completely covered with oil paint in order to prevent moisture loss and occurrence of end cracks.A conventional semi-automatic kiln with the capacity of 3 m 3 was used.Estimation of the current moisture content of kiln stacks was done by means of samples that were provided from the kiln stack and identifi ed as control samples.From each side of the control samples, cookies for the moisture content measurement with a length of 2.5-cm were cut and immediately weighed with an electric balance.Finally, they were oven-dried at 103 ±2 °C to a constant weight 2.2 Drying Procedure 2.2.Proces sušenja Three different moisture-based drying schedules were used, including schedule A (T 5 -D 2) , schedule B (T 5 -D 4 ), and schedule C (T 5 -D 6 ).Drying process terminated when the fi nal moisture content (MC) of 12 ± 2 % was reached.To introduce the new conditions in the kiln, the control samples, depending on the drying rate, were weighed at least once a day.This change was made based on the average MC of the wettest half of the control samples (Simpson, 1991).Finally, the drying process was terminated without any conditioning treatment for all of the drying runs.

Flexural Vibration Procedure 2.3. Vibracijski test savijanja
The cut samples were maintained in the open air for 2 days prior to drying.Then, fl exural vibration in free-free bar test was conducted on wet specimens in order to have primary information for further investigations of the acoustic properties of the dried timber (Fig. 1).This test was also carried out immediately after completion of drying process.
Based on Timoshenko's improved theory of fl exural free vibration test, as described in the literature (Bordonné, 1989;Brancheriau and Baillères, 2002;Roohnia et al. 2011a, b), after obtaining the n th modal frequency through Fast Fourier Transform (FFT), based on coordinates of evaluated spots from at least three initial modes of vibration, longitudinal specifi c modulus of elasticity was determined through a linear regression (Equation 1). in computing acoustic converting effi ciency (ACE).Either of factors (K or ACE) were applied as a criterion in selection of woods used in musical instrument industry and it is highly believed that as much the rate of these two factors in wood is higher, that wood will be more appropriate for the musical instrument industry (Roohnia et al., 2011b;Rujinirun et al., 2005).This phenomenon would surely benefi t from a better quality of wood with lower amounts of defects.
Drying-induced cracks were of the most important and major defects during the wood drying processes.They increase the internal friction, which consequently leads to the increase of the damping factor in the wood.In fact, as much the damping factor is lower and the acoustic coeffi cient is higher, the wood would have better resonance properties.Therefore, by considering the acoustic coeffi cient combined with damping in the shape of acoustical converting effi ciency (ACE), the impact of defects for investigating the wood drying effi ciency seems to become more and more facilitated.
Fig. 2 denotes that all the three drying schedules led to an increase in dynamic modulus of samples after the drying process, and schedule A resulted in a higher increase than the others, however, a lower correlation coeffi cient was observed.
The effect of each schedule on damping factor is shown in Fig. 3. Drying schedules of B and C led to an increase in damping factor, which was higher in the latter one.Conversely, schedule A, unlike B and C, decreased the damping factor.As the fi nal moisture content was the same for all the schedules, it might inspire the preferences of schedule A.
Fig. 4 shows that the acoustic coeffi cient of samples dried under all the three schedules has been higher than the corresponding fi gure before the drying process.Schedule A experienced a higher increase than schedules C and B, respectively.All the three drying schedules led to an increase in acoustic converting effi ciency in samples and the rate of increase in schedule A was higher than in schedule B and C, respectively (Fig. 5).It can be observed that not only the samples applied in this test showed unequal dynamic response, but the rate of these changes was also different depending on the schedule applied.
All the three drying schedules led to an increase in Dynamic Modulus of Elasticity of dried timber as against the non-dried ones, which can be attributed to mass reduction of samples due to moisture loss during the drying process (equation 1).In addition, dynamic modulus of samples dried by schedule A showed a higher increase than those dried by the schedule C and B, respectively.Although, changes in dynamic modulus are considered as one of the most important factors in qualitative evaluation of wood products, it should be born in mind that the place of defects, especially cracks, on wooden bars is highly effective on dynamic modulus of elasticity.Hence, we could not just rely on changes of this factor for judging and distinguishing the schedules applied.
In addition to the role of defects on damping factor, wood species (density, extractive contents, wood , where the inte rcept (E/ρ) is fl exural specifi c modulus of elasticity, G ij is shear modulus in LR or LT plane (not followed in present approach), K is shape coefficient, equivalent to 0.833 (Brancheriau and Baillères, 2002), a n and b n are the coordinates of the evaluated spots from at least three initial modes of fl exural vibration, n corresponds to mode number and R is the correlation coeffi cient of the Timoshenko's linear trend.
The origin and background to the derivation of the Equation (1) were presented in details in the above introduced group of literature.NDT-lab ® Portable System Setup for fl exural free vibration test developed using MATLAB ® 7.1 (Roohnia et al., 2006;Roohnia, 2007) was used here to obtain Fast Fourier Transform and related vibrational parameters (Fig. 1).Damping due to internal friction (tan) was calculated from logarithmic decrement (Bodig and Jayne, 1993;Brémaud, 2008).
Acoustic coeffi cient (K) and acoustical converting effi ciency (ACE) were calculated based on modulus of elasticity, density, and damping factor using Equations 2 and 3: (2) Where, K is the acoustic coeffi cient (m 4 s -1 kg -1 ), E is the modulus of elasticity (Pa), ρ is the density of wooden specimens (kgm -3 ), ACE is the acoustical converting effi ciency (m 4 s -1 kg -1 ).
The recording rate for the audio fi les was adjusted to 44100 Hz with 16 bits encoded (bit depth).
The scatter plots were used to compare the acoustical properties and their related change patterns.

REZULTATI I RASPRAVA
It seems critical to have an idea about the acoustic coeffi cient and acoustical converting effi ciency.Acoustic coeffi cient (K) is an indicative of the proportion of elasticity of wood to its density, and in combination with damping due to internal friction is crucial ) and moisture content are also infl uential (Tsoumis 1991).By reducing the moisture content of samples, the damping factor is expected to be reduced in the dried samples.However, this only happened in schedule A; and in the other two schedules (B and C) this factor increased and its rate was higher in schedule C than in schedule B. In this regard, existence of cracks in wood could be considered as a culprit in increasing of damping factor (Ouis, 2004;Roohnia et al., 2010;Hossein et al., 2009;Roohnia et al., 2011a).Therefore, the reduction in damping factor in schedule A, unlike the other two schedules, can be attributed to less formation of defects.In the preliminary stage of the pre-sent research, Shahverdi et al. (2012) reported a table, which has been extended (Table 1), to demonstrate the amount of visually recognized cracks attributed to the drying schedules A, B and C.
The rate of internal defect propagation due to drying processes severity has been higher in schedules B and C than in schedule A. In addition, considering total crack lengths and their abundance, schedule A has resulted in smaller amounts of drying defects i.e. crack (Table 1).
As mentioned before, acoustic coeffi cient is one of the most important parameters in estimating the acoustic specifi cations of a material, which is only in-  fl uenced by modulus of elasticity and density.Therefore, observing a trend like that observed in modulus of elasticity has not been a suffi cient criterion.By considering the explanations expressed about modulus of elasticity, the changes that occurred in this factor cannot be solely a basis for judgment between the drying schedules used.Especially not in case when the trend R squares were not suffi ciently high.Though, this correlation was strengthened in ACE comparisons.
As shown in equation 2, there is a direct relationship between acoustic coeffi cient and dynamic modulus of elasticity and, on the other hand, the same relationship can be found between dynamic modulus of elasticity and mechanical properties of wood.Since acoustic converting effi ciency is a proportion of acoustic coeffi cient to damping factor (equation 5) and damping factor itself might be under the infl uence of cracks, it can be predicted that acoustic converting effi ciency would be a proportion of wood resistance to the extent of defects induced by the drying process.The amount of this factor in the wood dried by schedule A was over twice as much as the non-dried one; whereas, in the wood dried by schedules B and C, this amount showed a 32 and 9 % increase, respectively.
The increasing trend of acoustic converting efficiency in the dried timber as opposed to non-dried one in all the three drying schedules was predictable due to a reduction of MC during drying followed by the reduction of damping factor.However, the increasing rate of this factor indicates that each schedule was ap-plied to the extent to which secondary defects could be caused in timber.Hence, schedule A caused the least abundance of cracks in the dried samples as against schedule B and C, respectively.
It could be concluded that the free vibration method has the potential to substitute the existing destructive and traditional methods in qualitative nondestructive grading of dried timber, especially in determining the effect of the drying schedule on the development of cracks as the most common defect in the drying processes.Meanwhile, the acoustical converting effi ciency (ACE), which is used as a criterion for wood quality in musical instrument industry, can be introduced as one of the key factors in grading of the dried timber under different drying conditions.A visual or destructive traditional inspection of wood drying quality may neglect hidden defects, unlike the proposed non-destructive evaluation.

ZAKLJUČAK
This research evaluated the potential of the vibration-based non-destructive testing method of timber infl uenced by the wood drying process.Different dynamic responses of samples before and after each drying schedule provided the ability to determine the importance of either internal or external defects in wood.In other words, non-destructive testing methods are capable of being used in sorting strategies of timber as they are more and more sensitive in detecting the defects.2) There was a signifi cant difference between samples concerning the dynamic modulus of elasticity, damping factor, acoustic coeffi cient, and acoustic converting effi ciency before and after the drying process.3) Dynamic modulus of elasticity and acoustic coefficient may not be exclusively suffi cient for choosing between the drying schedules applied.4) Considering the negative impact of cracks on damping factor and acoustic coeffi cient, it was shown that changes in acoustic converting effi ciency could be considered as an effective way for evaluating the drying schedules.