Air-coupled ultrasound and electrical impedance analyses of normally dried and thermally modified Scots pine (Pinus sylvestris)

Transcription

1 Wood Material Science & Engineering ISSN: (Print) (Online) Journal homepage: Air-coupled ultrasound and electrical impedance analyses of normally dried and thermally modified Scots pine (Pinus sylvestris) Laura Tomppo, Markku Tiitta & Reijo Lappalainen To cite this article: Laura Tomppo, Markku Tiitta & Reijo Lappalainen (2016) Air-coupled ultrasound and electrical impedance analyses of normally dried and thermally modified Scots pine (Pinus sylvestris), Wood Material Science & Engineering, 11:5, , DOI: / To link to this article: Published online: 03 Dec Submit your article to this journal Article views: 113 View related articles View Crossmark data Citing articles: 1 View citing articles Full Terms & Conditions of access and use can be found at

2 Wood Material Science & Engineering, 2016 Vol. 11, No. 5, , ORIGINAL ARTICLE Air-coupled ultrasound and electrical impedance analyses of normally dried and thermally modified Scots pine (Pinus sylvestris) LAURA TOMPPO, MARKKU TIITTA, & REIJO LAPPALAINEN Department of Applied Physics, University of Eastern Finland, Kuopio, Finland Abstract Wood is graded according to strength in various applications. The ultimate strength can only be determined by breaking the specimen, and thus other characteristics like density and modulus of elasticity (MOE) are used for estimation of the strength. In this study, the properties of normally dried and thermally modified Scots pine were studied using electrical impedance and air-coupled ultrasound (ACU) methods. Density, hardness, MOE and strength were analysed and compared with the electrical and ultrasonic responses. The measurements were conducted in stable laboratory conditions with well equilibrated samples and the moisture content was not determined. Both the ultrasonic and electrical parameters correlated with the density and mechanical properties. Using multivariate analysis, density could be estimated with accuracy of 21 kg/m 3 (SD) for normally dried timber and 13 kg/m 3 (SD) for thermally modified timber (TMT; N = 15). MOE could be estimated with accuracy of 0.7 kn/mm 2 (SD) for normally dried timber and 1 kn/mm 2 (SD) for TMT (N = 14). According to the study, electrical impedance spectroscopy combined with ACU measured across the grain is a potential non-destructive technique for the strength estimation of wood. Keywords: Air-coupled ultrasound, electrical impedance spectroscopy, strength Introduction Density affects the mechanical properties of wood. Other factors affecting the mechanical properties are moisture content (MC) and temperature. In addition, natural and processing-based defects of wood reduce its strength. Natural defects include knots and sloping grain, whereas splits and checks are usually processing-based. Material heterogeneity, which is increased by defects, causes stress concentrations in subject under load and thus the strength is reduced (Zink-Sharp 2003). In the European structural timber system the timber is assigned to a grade according to its bending strength, modulus of elasticity (MOE) and density. The timber is graded visually or with a machine grader, and bending machines are commonly used machine graders. The stiffness of timber can be determined by bending the piece into fixed deflection or by bending it with fixed force. Acoustic techniques are also used in grading; acoustic vibration is induced into the timber and the sound wave velocity or resonance frequency can be used for calculation of the MOE. The strength grading techniques are reviewed by Pellerin and Ross (2002), Brännström (2009) and Ranta-Maunus et al. (2011). Thermally modified timber (TMT) exhibits reduced hygroscopicity, lower equilibrium MC and lower density compared with normally dried wood. In thermal modification, wood is kept at C for several hours in an atmosphere that prevents it from burning. Thermal modification affects both the physical and chemical properties of wood. For example, it increases the dimensional stability, decay resistance and thermal insulating capacity of wood. On the other hand, the strength properties of wood, e.g. modulus of rupture (MOR) and impact bending strength, tend to decrease. The extent of the changes is affected by the treatment temperature and the wood species (Esteves and Pereira 2009). In electrical impedance spectroscopy (EIS), an alternating electric field at varying frequencies is induced into a specimen, and the responses are measured. Complex impedance spectrum can be determined Correspondence: Laura Tomppo, Department of Applied Physics, University of Eastern Finland (Kuopio campus), PO Box 1627, FI-70211, Kuopio, Finland. Tel: Fax: laura.tomppo@uef.fi (Received 1 November 2013; revised 16 September 2014; accepted 28 October 2014) 2014 Informa UK Limited, trading as Taylor & Francis Group

3 Ultrasound and electrical impedance analyses of pine 275 when impedance modulus and phase responses are measured and thus both capacitive and conductive properties can be evaluated. MC is the main factor affecting electric and dielectric properties of wood, but density, temperature and grain angle have an effect as well (Skaar 1988, Torgovnikov 1993). EIS has been used for the characterization of wood and TMT (Tiitta et al. 1999, Tiitta and Olkkonen 2002, Zelinka et al. 2007, 2008, Tomppo et al. 2011, van der Beek et al. 2011). Ultrasound and other acoustic techniques have been used as non-destructive testing (NDT) techniques for strength determination and detection of internal defects. Density, wood structure and different types of defects exert various effects on the ultrasound propagation. The development of aircoupled ultrasound (ACU) techniques (Bhardwaj 2004a, 2004b) has made the ultrasound method more feasible for online industrial applications. Online wood MC measurement based on ACU has been proposed (Vun et al. 2008), and density and defects in wood samples have been studied by ACU (Marchetti et al. 2004). In addition, throughtransmission ACU has been utilised in the detection of checks in timber and TMT (Gan et al. 2005, Tomppo 2013, Tomppo et al. 2014), delamination in wood (Bucur 2010, 2011) and in glulam beams of arbitrary number of lamellas up to 500 mm thickness (Sanabria et al. 2011, Sanabria 2012, Sanabria et al. 2013a). In wood-based panels, both density and particle type affect the ultrasound transmission (Vun and Bhardwaj 2004, Hilbers et al. 2012, Sanabria et al. 2013b). In the current study, normally dried and TMT pine samples were studied under laboratory conditions. The main goal was to determine whether EIS and ACU measurements across the grain could be used for estimation of the mechanical properties. Materials and methods Fifteen Scots pine (Pinus sylvestris) boards were cut in two pieces. One part was normally dried at 70 C and another was heat-treated at temperature of 228 C using Thermowood procedure. The dried and thermally modified samples were mm 3 in size, and they were further cut into two pieces; 300 mm specimens were cut for strength and density determinations, and 400 mm specimen was used for the NDT measurements (Figure 1). MOR and MOE were determined in three-point bending. Dimensions of the test pieces were 300 mm 20 mm 20 mm. Distance between the centres of the device supports was 240 mm. The loading was carried out uniformly at constant speed (constant loading rate). The test pieces were broken in 1.5 ± 0.5 min from the beginning of loading. MOR was determined as: MOR ¼ 3P maxl 2bh 2 ð1þ where P max is the breaking load (N), L is the distance between the centres of the supports (mm), b is the width of the test piece (mm) and h is the thickness of the test piece (mm). MOE was determined as: MOE ¼ L3 k 4bh 3 ð2þ where k is the slope of the straight-line portion of the load-deflection curve (N/mm). The MOR and MOE were determined separately for heartwood and sapwood parts of the boards, and then average of the two was calculated and used in the further analyses. There were nine samples that broke from a knot, and thus they were excluded from the analyses. The parallel heartwood/sapwood sample, if successfully tested, was then used in further analyses. For both normally dried and TMT there was one sample with both heartwood and sapwood sample unsuccessfully tested, and thus without MOR or MOE value. Brinell hardness was tested with a method adapted from EN 1534 (2000); a steel ball of 10 mm diameter and normal load of 490 N, which deviates from the norm. Maximum load was reached in 15 s with a constant rate of loading. Force was kept constant for 30 s and then reduced with a constant speed to zero within 15 s. Brinell hardness was calculated as follows: Figure 1. Samples were cut in pieces to measure mechanical, electrical and acoustical properties of wood. For the EIS measurement, the electrodes were set on the side of the flat face on the board.

4 276 L. Tomppo et al. 2P BH ¼ h i ð3þ pd D D 2 d where P is the maximum load (N), D is the diameter of the steel ball and d is the diameter of the impression. Samples were stored in laboratory conditions before the tests, and measurement conditions were T=22 ± 2 C and RH 37 ± 4%. Density was determined from the samples dried at 103 ± 2 C. Longitudinal contact ultrasound, EIS and ACU measurements were made from the samples 40 mm in length. The transit time of longitudinal ultrasound was measured with Pundit (C.N.S Electronics Ltd, London, England) and it was used to calculate the longitudinal ultrasound velocity (v US ). Ultrasound gel was used between the transducer and the wood to enhance the contact. The impedance measurements were made using Hioki 3531 Z HiTester (Hioki E. E. Corporation, Ueda, Japan) at frequency range of 5 khz 2 MHz, with 26 frequencies in total. The analyser was calibrated using open and short circuit compensations. Parallel plate electrodes were set on the wood for the measurements. The stainless steel electrodes were 40 mm 50 mm in size and the distance between the electrodes was 3 mm. The measurements were conducted in longitudinal direction (Figure 1) from heartwood and sapwood side of the board. From the analyser output, the impedance modulus Z and phase angle Φ were used in further analyses. ACU set-up included three through transmission measurements across the grain (Figure 2). Three measurement frequencies were used: 90 khz through 100 mm of wood, 190 khz and khz through 32 mm. The goal was to measure the whole cross section of the board. The frequencies were chosen according to the best penetration and transducer frequency range. The samples were scanned from the centre section: eight measurements with 10 mm intervals were made; the measurements at 190 khz and khz were made from the both sides. The ACU measurement system consisted of function generator (Agilent 33220A, Santa Clara, CA, USA) and a custom made linear amplifier as a signal source. Olympus 5058PR (Olympus Corporation, Tokyo, Japan) and a PC with Compuscope 8380 digitizer (Gage Applied Technologies Inc. Lachine, Quebec, Canada) were used for receiving and saving signals. Gas Matrix Piezoelectric transducers (The Ultran Group, State College, PA, USA) were used as follows; 90 khz measurement: NCG100-D50- P150 (transmitter, focused) and NCG100-S25 (receiver); 190 khz measurement: NCG200-D25; and khz measurement: NCG500-D13-P38. The distance between the sensor and the board was 150 mm for the 90 khz transmitter and 50 mm for the receiver. The distances between sensors and board were 40 mm for the both higher frequencies. Several parameters describing the shape and the amplitude of the signal were determined (Table I), and for each board the median of the measured parameters was used in statistical analyses. In frequency domain, the signal contained typically two peaks near the resonance frequency; the peaks were analysed separately and referred to as lower frequency (fmax1 and fmaxamp1) and higher frequency Figure 2. The transducer set-up in ACU measurements.

5 Ultrasound and electrical impedance analyses of pine 277 Table I. The determined ultrasound parameters for ACU signals. Abbreviation Parameter Time domain maxamp Maximum amplitude tmax Location of maximum peak tt Transit time (threshold crossing) Frequency af Area of the signal domain cf Centroid frequency fmax1 Location of lower frequency maximum peak fmax2 Location of higher frequency maximum peak fmaxamp1 Maximum amplitude of lower frequency fmaxamp2 Maximum amplitude of higher frequency (fmax2 and fmaxamp2). These frequencies were different for each transducer pair. The analyses of two frequency peaks may provide more information on attenuation and dispersion of the signal than analyses of one peak. The ultrasound signal analyses and statistical analyses were carried out with Matlab R2012a (Mathworks, Natick, MA, USA). Results The average values and standard deviations of density, mechanical properties and longitudinal ultrasound speed are presented in Table II and their correlations are presented in Table III. The mechanical properties, MOR and MOE, correlated with the density for normally dried and thermally modified Scots pine timber as well as with each other. For both materials, there was also very significant correlations between the Brinell hardness and density. For the normally dried timber, hardness was the only parameter that correlated significantly with the longitudinally measured ultrasound velocity. For the TMT, the ultrasound speed correlated significantly only with the MOE. Table II. Average values and standard deviations of the determined speed of longitudinal contact ultrasound (v US ), density (ρ), Brinell hardness, MOE and MOR for the 70 C dried and 228 C thermally modified samples. N is the number of the samples. 70 C dried TMT 228 C Mean SD N Mean SD N v US (m/s) ρ (kg/m 3 ) Hardness (N/mm 2 ) MOE (kn/mm 2 ) MOR(MPa) Table III. Correlations for the determined speed of longitudinal contact ultrasound (v US ), density (ρ), Brinell hardness, MOE and MOR. Number of the samples N was 15, except for correlations with MOR and MOE, for which N = 14 and for MOR/MOE correlation for TMT N = 13. v US ρ Hardness MOR MOE 70 C dried v US 1.00 ρ Hardness 0.54* 0.80*** 1.00 MOR *** 0.72** 1.00 MOE *** 0.77** 0.96*** 1.00 TMT 228 C v US 1.00 ρ Hardness *** 1.00 MOR ** 0.86*** 1.00 MOE 0.62* 0.76** 0.83*** 0.81*** 1.00 *p < 0.05; **p < 0.01; ***p < Examples of the measured ACU signals are presented in Figure 3. There were significant correlations between ACU parameters and density or mechanical properties for both normally dried timber and TMT for 90 khz measurement (Tables IV and V, Figure 4), 190 khz and khz measurements. For impedance modulus, the correlations were in general stronger, and there were consistent correlations between Z and density, hardness, MOR and MOE (Table VI, Figure 4). For the phase angle Φ, there were only few nearly significant (p < 0.05) correlations for TMT (e.g. ρ and Φ at 300 khz, r = 0.61, p < 0.05). The multiple regression models were calculated (Table VII, Figure 5). For density, the EIS model and combined model (EIS and ACU) had higher explanatory power than the ACU model. The explanatory power of the models for MOR and MOE was in range r 2 = for ACU alone, whereas for EIS alone the range was r 2 = For the combined models, the range was r 2 = Discussion The current study provided encouraging results for developing a measurement method based on the EIS and transverse ACU for evaluation of strength of timber. The method showed potential for both normally dried and TMT. The limitations of the sample set, especially the number of the samples, are acknowledged and thus further studies with more extensive sample set are pursued. However, the sample set covered the typical density range of Scots pine, and is thereby representative. Due to the thermal modification, the density decreased 6%, which is in the range of values reported

6 278 L. Tomppo et al. (a) U(V) U(V) (c) t(μs) t(μs) (b) U(au) (d) U(au) f (khz) f (khz) Figure 3. Examples of ACU responses of normally dried (a/b) and heat-treated wood (c/d). Time and frequency responses for the ACU measurements at 190 khz. The specimens are from the same board. earlier (Boonstra et al. 2007, Metsä-Kortelainen and Viitanen 2010). The defects of wood could affect the changes of mechanical properties during thermal modification; Boonstra et al. (2007) reported stronger decrease in MOR due to thermal modification for full size Norway spruce boards compared with defect free specimens. Thus, the rather strong changes of MOR ( 25%) and MOE ( 8%) in the current study could be affected by defects. Several changes, both advantageous and disadvantageous for the ultrasound propagation, take place in wood during the thermal modification. For example the crystallinity of cellulose increases during thermal modification (Boonstra and Tjeerdsma 2006). Furthermore, the increased crosslinking of lignin strengthens the middle lamella (Boonstra et al. 2007) and the increased crystallinity and the degradation of hemicelluloses decreases equilibrium MC of TMT (Esteves and Pereira 2009). These changes can enhance the propagation of ultrasound. On the other hand, the thermal modification especially in high temperatures can cause micro-cracking and cracking in wood. Cracking hinders the propagation of ultrasound, particularly when the cracks are aligned perpendicularly to the ultrasound beam. For the TMT, the correlation coefficient between MOR and MOE in this study was in the range of earlier reported for flatwise bending (reviewed by Pellerin and Ross 2002), and for dried wood it was even higher. The longitudinally measured stress wave velocity can be used for estimation of MOE, with typical correlations close to 1 for clear wood. For knotty wood, the stress wave approach tends to overestimate the MOE, but a correlation of 0.87 has been reported (Gerhards 1982). In the Table IV. Correlations between wood properties and ACU parameters at 90 khz. Through-transmission measurement for normally dried wood (70 C). N = 15 for longitudinal ultrasound velocity (v US ), density (ρ) and hardness and N = 14 for MOR and MOE. The ultrasound parameters are explained in Table I. Table V. Correlations between wood properties and ACU parameters at 90 khz. Through-transmission measurement, TMT (228 C). N = 15 for longitudinal ultrasound velocity (v US ), density (ρ) and hardness and N = 14 for MOR and MOE. The ultrasound parameters are explained in Table I. v US ρ Hardness MOR MOE tt cf * 0.76*** tmax * 0.58* maxamp * 0.53* fmax fmaxamp fmax fmaxamp af ** 0.66** *p < 0.05; **p < 0.01; ***p < v US ρ Hardness MOR MOE tt 0.65** cf tmaxamp maxamp * fmax fmaxamp * 0.57* fmax fmaxamp * af * 0.67** *p < 0.05; **p < 0.01.

7 Ultrasound and electrical impedance analyses of pine 279 (a) Z at 80 khz (Ω) 5 x C heart 228 C sap 70 C heart 70 C sap (b) ACU 90 khz, fmaxamp U (au) 228 C 70 C Density (kg/m 3 ) Density (kg/m 3 ) Figure 4. Impedance modulus at 80 khz as a function of the density (a) and fmaxamp of 90 khz ACU measurement as a function of the density (b). The lines represent the linear regressions for each subgroup of the graph. current study, the longitudinal ultrasound velocity was measured from the samples parallel to ones used in the mechanical testing. Since the specimens were not the same, although from same board, the rather low correlations between the v US and MOE are affected by the wood defects. However, the main goal for the measurement of the longitudinal ultrasound velocity was to compare it with the other NDT measurements. In the current study the MOE and MOR were calculated as an average of ones measured from heartwood and sapwood. In the ACU and EIS measurements, there was varying amount of both in the measurement region, which can cause variation in the results. In addition, for some samples the analysed value is only from heartwood or sapwood, since one of the two was broken from a knot. The NDT measurements were not made from the smaller cross-sections corresponding to the bending test samples, because the sample size affects the signal reflections for ultrasound, and the goal was to measure authentic timber material. Due to the measurement set-up, the edgewise ACU measurement covered the board cross section in whole, whereas the other measurements covered only part of the board. Therefore, the stronger Table VI. Correlations between normally dried/tmt characteristics and impedance magnitude (log Z) at different frequencies (f). N = 15 for longitudinal ultrasound velocity (v US ), density (ρ) and hardness and N = 14 for MOR and MOE. f (khz) v US ρ Hardness MOR MOE Dried 70 C *** 0.68** 0.66* 0.61* * 0.76** 0.77*** 0.76** 0.69** * 0.85*** 0.77*** 0.78*** 0.74** * 0.75** 0.76** 0.73** 0.68** * 0.78*** 0.76** 0.74** 0.69** * 0.81*** 0.78*** 0.76** 0.71** * 0.66** 0.63* 0.69** 0.75** TMT 228 C *** 0.58* * *** 0.67** 0.63* 0.58* * 0.90*** 0.70** 0.57* 0.59* ** 0.86*** 0.70** 0.60* 0.74** * 0.94*** 0.74** 0.65* 0.68** * 0.93*** 0.70** 0.58* 0.68** *p < 0.05; **p < 0.01; ***p <

8 280 L. Tomppo et al. Table VII. Summary of the linear regression models. The applied measurement method, explanatory variables in the model, the coefficient of determination (r 2 ) and the standard error of the estimate. The significance of the coefficients for different parameters in the model are presented as asterisk. For density N = 15 and for MOR and MOE N = C 228 C ρ f (khz) Parameters r 2 of the estimates f (khz) Parameters r 2 of the estimates Standard error Standard error ACU 90 EIS EIS and ACU 20 MOR cf** fmaxamp2** log Z, sap*** Φ, sap* log Z, sap*** fmaxamp2** kg/m kg/m kg/m fmaxamp2* tmax* log Z, sap*** Φ, sap* log Z, sap*** af* kg/m kg/m kg/m 3 ACU 190 EIS EIS and ACU MOE tmax** tmax** MPa log Z, heart*** MPa 80 Φ, sap* 80 log Z, heart** MPa 80 tmax ns 190 tt* fmax* log Z, sap** Φ, heart** log Z, sap** tt* MPa MPa MPa ACU 90 EIS EIS and ACU 500 cf* tmax* log Z, heart** log Z, heart* log Z, heart*** tt** kn/mm kn/mm kn/mm 2 40 tt** tmax ns kn/mm 2 log Z, sap*** kn/mm 2 Φ, heart** log Z, heart** kn/mm 2 fmax2** *p < 0.05; **p <0.01; ***p < correlations between the reference parameters and the edgewise ACU compared with those between the reference parameters and flatwise were expected. For example, the edgewise measured area of the signal in frequency domain (af) correlated strongly with MOR and MOE for both dried and TMT. In general, the amplitude-related parameters correlated negatively with density, MOR and MOE. In addition, the longitudinal ultrasound velocity correlated positively with the transit time (tt). There are two possible reasons for this. Firstly, the specimens with higher density, MOR and MOE were also harder. For harder wood, a larger portion of ultrasound can be reflected from the airwood interfaces than for softer and less dense wood. Secondly, the more dense wood has typically more frequent year rings. Thus the ultrasound in denser wood has to cross more earlywood latewood interfaces when travelling in the radial direction, and each interface-crossing causes also reflections which reduce the energy of transmitted signal. (a) Predicted r (kg/m 3 ) Dried 70 C TMT 228 C (b) Predicted MOE (N/mm 2 ) Measured r (kg/m 3 ) Measured MOE (N/mm 2 ) Figure 5. Density (a) and MOE (b) predicted by linear regression models (models in Table VII, ACU + EIS) for normally dried wood and TMT.

9 Ultrasound and electrical impedance analyses of pine 281 In wood with knots and other defects, the ultrasound tends to travel through the fastest pathway. The effect of knot or other defect on transverse ultrasound depends on the dimensions and orientation of the defect. For example, checks perpendicular to the ultrasound propagation tend to attenuate the signal (Bucur 2011, Tomppo 2013). On the other hand, for example in veneer a transverse ultrasound was not affected by the knots (Wang et al. 2001). Compared with the longitudinal ultrasound measurement, the transverse measurement could provide more accurate spatial information of the strength of the wood. The spatial information would be valuable tool for detecting the weakest proportion of the timber board. In the current study, the modulus of electrical impedance was strongly correlated with the mechanical characteristics and density of the timber, which indicates that EIS alone could be used for measurement of MOR or MOE of conditioned samples. However, the MC of timber in industrial scale is rarely constant between batches, and thus the effect of MC on EIS measurements would become a problem and should be compensated. Since the phase angle was not correlated with strength properties, it could be used for estimation of the MC. In the Table VII, the regression models were limited to two explanatory parameters. Thus, the model containing both ACU and EIS parameters was in certain cases weaker than one containing only measurements with a single technique. It is assumed that the model combining both ACU and EIS data would perform better for material with MC variation, since ultrasound attenuation is not very sensitive to MC changes below the fibre saturation point (Sakai et al. 1990). For TMT, the intensity of the treatment also affects the electrical responses. For example colour measurement as an independent tool for evaluation of the treatment intensity has been used to enhance the prediction of MC by resistivity of TMT (Brischke et al. 2014), and presumably it could as well be used to enhance the prediction of other properties like density and strength. In an extensive comparison of different NDT techniques for strength estimation, the best correlations for longitudinal ultrasound measurements were achieved when combined with electrical resistance (Hanhijärvi and Ranta-Maunus 2008); the r 2 for MOE was 0.76 and for MOR was The current study indicates that the ACU measurement across the grain combined with EIS measurement could be used for estimation of strength too, but in order to estimate the accuracy of the method, measurements with more extensive sample set with larger natural variance in MC needs to be conducted. Conclusions Linear regression models for the density, MOR and MOE of normally dried and thermally modified Scots pine timber were constructed based on ACU and EIS data. Both series consisted of 15 equilibrated samples which were measured in laboratory conditions. The MC of the samples was not determined during the NDT measurements. However, the current study indicates that the ACU measurement across the grain, combined with EIS measurement could be used for estimation of strength of dried and TMT. Acknowledgements The authors thank Mr Pekka Miettinen and Mr Ville-Veikko Wettenhovi for their contribution in the experimental part of the study. Technical assistance from Mr Juhani Hakala and Mr Aimo Tiihonen is acknowledged. References Bhardwaj, M. C. (2004a) Evolution of piezoelectric transducers to full scale non-contact ultrasonic analysis mode. 16th World Conference on Nondestructive Testing, Montreal, Canada, 30 August 3 September. Bhardwaj, M. C. (2004b) High efficiency non-contact transducers and a very high coupling piezoelectric composite. 16th World Conference on Nondestructive Testing, Montreal, Canada, 30 August 3 September. Boonstra M. J. and Tjeerdsma B. (2006) Chemical analysis of heat treated softwoods. Holz Als Roh-Und Werkstoff, 64, Boonstra, M. J., Van Acker, J., Tjeerdsma, B. F. and Kegel E. V. (2007) Strength properties of thermally modified softwoods and its relation to polymeric structural wood constituents. Annals of Forest Science, 64, Brännström, M. (2009) Integrated strength grading. PhD thesis, Luleå University of Technology, Skellefteå, Sweden. Brischke, C., Sachse, K. A. and Welzbacher, C. R. (2014) Modeling the influence of thermal modification on the electrical conductivity of wood. Holzforschung, 68, Bucur, V. (2010) Delamination detection in wood based composites, a methodological review. Proceedings of 20th International Congress on Acoustics, ICA, Sydney, Australia, August, pp Bucur, V. (2011) Delamination in Wood, Wood Products and Woodbased Composites (Dordrecht: Springer). EN 1534 (2000) Wood and Parquet Flooring Determination of Resistance to Indentation (Brinell) Test Method (Brussels: CEN European Committee for Standardization). Esteves, B. M. and Pereira, H. M. (2009) Wood modification by heat treatment: A review. Bioresources, 4, Gan, T. H., Hutchins, D. A., Green, R. J., Andrews, M. K. and Harris, P. D. (2005). Noncontact, high-resolution ultrasonic imaging of wood samples using coded chirp waveforms. IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control, 52, Gerhards, C. (1982). Effect of knots on stress waves in lumber. Effect of Knots on Stress Waves in Lumber. Research paper FPL-384 (Madison, WI: Forest Products Laboratory, U.S. Department of Agriculture, Forest Service).

10 282 L. Tomppo et al. Hanhijärvi, A. and Ranta-Maunus, A. (2008) Development of strength grading of timber using combined measurement techniques. Report of the Combigrade Project-Phase 2 (Espoo: VTT Publications 686). Hilbers, U., Thoemen, H., Hasener, J. and Fruehwald, A. (2012) Effects of panel density and particle type on the ultrasonic transmission through wood-based panels. Wood Science and Technology, 46, Marchetti, B., Munaretto, R., Revel, G., Tomasini, E. P. and Bianche, V. B. (2004). Non-contact ultrasonic sensor for density measurement and defect detection on wood16th World Conference on Nondestructive Testing, Montreal, Canada, 30 August 3 September. Metsä-Kortelainen, S. and Viitanen, H. (2010) Effect of fungal exposure on the strength of thermally modified Norway spruce and Scots pine. Wood Material Science & Engineering, 5, Pellerin, R. F. and Ross, R. J. (2002). Nondestructive Evaluation of Wood (Madison, WI: Forest Products Society). Ranta-Maunus, A., Denzler, J. K. and Stapel, P. (2011) Strength of European timber Part 2. Properties of spruce and pine tested in Gradewood project: VTT Working papers 179 (Finland: Espoo). Sakai H., Minamisawa, A. and Takagi, K. (1990) Effect of moisture content on ultrasonic velocity and attenuation in woods. Ultrasonics, 28, Sanabria, S. (2012) Air-coupled ultrasound propagation and novel non-destructive bonding quality assessment of timber composites. PhD thesis, ETH No 20404, Swiss Federal Institute of Technology, Zürich, Switzerland. Sanabria, S. J., Furrer, R., Neuenschwander, J., Niemz, P. and Sennhauser, U. (2011) Air-coupled ultrasound inspection of glued laminated timber. Holzforschung, 65, Sanabria, S. J., Furrer, R., Neuenschwander, J., Niemz, P. and Sennhauser, U. (2013a) Novel slanted incidence air-coupled ultrasound method for delamination assessment in individual bonding planes of structural multi-layered glued timber laminates. Ultrasonics, 53, Sanabria, S. J., Hilbers U., Neuenschwander J., Niemz P., Sennhauser U., Thömen H. and Wenker J. L. (2013b) Modeling and prediction of density distribution and microstructure in particleboards from acoustic properties by correlation of noncontact high-resolution pulsed air-coupled ultrasound and X-ray images. Ultrasonics, 53, Skaar, C. (1988) Wood-water Relations (Berlin: Springer Verlag). Tiitta, M. and Olkkonen, H. (2002) Electrical impedance spectroscopy device for measurement of moisture gradients in wood. Review of Scientific Instruments, 73, Tiitta, M., Savolainen, T., Olkkonen, H. and Kanko, T. (1999). Wood moisture gradient analysis by electrical impedance spectroscopy. Holzforschung, 53, Tomppo, L. (2013) Novel applications of electrical impedance and ultrasound methods for wood quality assessment. PhD thesis, University of Eastern Finland, Kuopio, Finland. Tomppo, L., Tiitta, M., Laakso, T., Harju, A., Venäläinen, M. and Lappalainen, R. (2011) Study of stilbene and resin acid content of Scots pine heartwood by electrical impedance spectroscopy (EIS). Holzforschung, 65, Tomppo, L., Tiitta, M. and Lappalainen, R. (2014) Nondestructive evaluation of checking in thermally modified timber. Wood Science and Technology, 48, Torgovnikov, G. I. (1993) Dielectric Properties of Wood and Woodbased Materials (Berlin: Springer-Verlag). Van der Beek, J., Tiitta, M., Tomppo, L. and Lappalainen, R. (2011) Moisture content determination of thermally modified timber by electrical and ultrasound methods. International Wood Products Journal, Vun, R. Y. and Bhardwaj, M. C. (2004) Non-contact ultrasonic characterization of in-plane density variation in oriented strandboard. 16th World Conference on Nondestructive Testing, Montreal, Canada, 30 August 3 September. Vun, R. Y., Hoover, K., Janowiak, J. and Bhardwaj, M. (2008). Calibration of non-contact ultrasound as an online sensor for wood characterization: Effects of temperature, moisture, and scanning direction. Applied Physics A, 90, Wang, J., Biernacki, J. M. and Lam, F. (2001). Nondestructive evaluation of veneer quality using acoustic wave measurements. Wood Science and Technology, 34, Zelinka, S. L., Rammer, D. R. and Stone, D. S. (2008). Impedance spectroscopy and circuit modeling of southern pine above 20% moisture content. Holzforschung, 62, Zelinka, S. L., Stone, D. R. and Rammer, D. S. (2007) Equivalent circuit modeling of wood at 12% moisture content. Wood and Fiber Science, 39, Zink-Sharp, A. (2003) The mechanical properties of wood. In J. H. Barnett and G. Jeronimidis (eds.) Wood Quality and Its Biological Basis (Boca Raton, FL: CRC Press), pp