influence of some structural wood defects on
Transkrypt
influence of some structural wood defects on
INFLUENCE OF SOME STRUCTURAL WOOD DEFECTS ON ULTRASONIC WAVE SPECTRUM WITOLD DZBEŃSKI, TOMASZ WIKTORSKI Department of Wood Science and Wood Protection, Faculty of Wood Technology, Warsaw Agricultural University Abstract: Influence of some structural wood defects on ultrasonic wave spectrum. Methods for evaluation of wood's quality and technical fitness by using ultrasonic and other acoustic techniques are commonly used in practice all over the world. To this end, first and foremost, the following parameters are analysed: signal transition time, maximum amplitude of received signal, signal attenuation factor. Based on these parameters overall quality of wood can be forecast but types of wood defects cannot. As results from the recorded spectrum of signals propagated in wood followed by the spectrum analysis, it is also possible to predict anatomical structure of wood at the section tested. Based on preliminary model-scale tests using the Sylvatest Duo apparatus, a correlation has been found between wood's anatomical defects, encountered by the signal passing through the material, and the signal's time and frequency performance being analysed. Key words: wood, non-destructive testing, ultrasounds, identification of structural defects INTRODUCTION The application of non-destructive methods for testing wood properties and evaluating its quality by using sonic and ultrasonic techniques has practically been imperceptible in the Polish wood technology sector although the related fundamental research studies have been completed [Dzbeński 1978, Dzbeński 1984, Mańkowski 2004, Moliński at al., 2000, Moliński 2002]. Ultrasounds commonly used in various fields of science and technology [Lewińska-Romnicka 2001, Śliwiński 2001] and huge development potential for using the above mentioned techniques for wood [Bucur 1996, Zombori 2001] encourage further research work in this field. In recent decades, the techniques based on measurements of parameters of elastic waves propagated in various materials, also including wood [Sandoz, Benoit, Demay 2001], have considerably been developed. Experiments based on measurements of natural vibration frequencies were also conducted [Bozhang, Pellerin 1996]. The team led by prof. Sandoz has developed a conventional method of ultrasonic wave propagation while designing transceivers operating in both sonic and ultrasonic bands [Sandoz 1996, Sandoz, Benoit, Demay 2001]. The team's work results were used in the Sylvatest Duo device thus helping it to improve the range of waves penetrating the material tested. The device was used to generate signals in the experiment as described below. The device's basic indications include: time necessary for the signal to pass through the specimen tested, wave velocity and maximum signal vibration amplitude referred to as maximum peak energy. The wave velocity is the most clearly correlated to the modulus of elasticity of wood and its mechanical properties, whereas, it is suggested to evaluate the material's biological degradation by using the vibration amplitude. The measurement results allow the tested wood specimen's quality and technical fitness to be evaluated in general terms. However, the results cannot be used to predict what types of structural defects are in the signal's way. According to Bozhang and Pellerin's study as mentioned above, the natural frequencies of wood with various defects vary, which may influence on selective attenuation of elastic waves. TESTING SET Defected and non-defected pinewood specimens were tested. The measuring signal was propagated in wood at the following sections: (1) without visible structural defects, (2) with slope of grain, (3) with knots - a comparative measurement was taken (4) across the trunk under an angle of about 45o (Figure 1). The test signal was generated by the Sylvatest Duo apparatus and recorded by digital oscilloscope HM407. Then, the signal was analysed by using MATLAB 5.1, a calculating software. The signal for analysis was sampled at 400 kHz for 5 ms. Such test parameters allowed all the essential signal's harmonics to be measured with a high margin of safety. Having made the signal analysis it was found that it would also be possible to use a lower sampling frequency - 120 kHz minimum - thus decreasing time resolution in favour of frequency accuracy and the extension of interval of signal tested. No evident harmonics above the frequency of approx. 60 kHz were found in the signal tested. Lower sampling frequencies will result in systematic loss of information on the signal's harmonics having a frequency lower than 60 kHz. Figures 1 Specimens used during tests. Fig. 1A During test. Fig. 1B B-D section with knot. Fig. 1C D-E section with slope of grain, D-G section with no visible structural defects. RESULTS The reference run of the signal in a homogeneous material - the Plexiglas specimen supplied together with the device - was recorded. Figure 2a shows characteristics versus time for phenomena accompanying a full measurement. Theoretical model for piezoelectric phenomena identifies deformations of piezoelectric element as being proportional to the element's electric field intensity. The motion of piezoelectric element is described by a second order differential equation (Equation 1) [Śliwiński 2001, on the basis of other studies]. Equation 1 Equation of piezoelectric element's motion, where: x – deformation, q – electrical charge, i – current intensity, U – voltage, t – time, ω – angular frequency; m,r,k – parameters of piezoelectric element. mx 2 d 2 i rx 2 di kx 2 2 i U 0 cos t 2 q dt q t q So, it can be stated that the emission of ultrasounds is triggered by a step increase in current intensity in the system upon starting measurement, and voltage is expressed as a current integral. The current intensity characteristic is similar to the theoretical characteristic of condenser charging [Bolkowski 1986] and is identical (with some proportionality factor) to the voltage derivative curve. The initial rapid increase in amplitude during the test is caused by a transient state of the current in the transmitting head's piezoelectric element. Once the transmitting head's current and voltage have stabilised in the second half of the graph, the signal vibration amplitudes also stabilise. It should be noted that the cumulative signal energy is correlated with the transmitting head's voltage, whereas, the absolute values of vibration amplitude with the transmitting head voltage's derivative. The correlation factors are r1 = 0.79 and r2 = 0.42, respectively. Therefore, a conclusion has been drawn that the envelope of signal recorded by the receiving head is chiefly dependent on the transmitting head's voltage and current characteristics. Fig. 2a Measurement with Sylvatest Duo 0.03 0.02 Dashed line - transmitting head`s voltage in [V]*1e+4 Amplitude [V] 0.01 0 Dot line - voltage derivative in [V]*1e+2 -0.01 Full line - time spectrum of signal recorded by receiving head; graph moved by - 0.01V -0.02 -0.03 0 0.02 0.04 0.06 Time [s] 0.08 0.1 0.12 Fig. 2b Time spectrum for the first 5 ms of test in homogeneous material 0.4 Amplitude mV 0.2 0 -0.2 -0.4 0 0.5 1 1.5 2 2.5 Time [s]*1E-3 3 3.5 4 4.5 5 -3 x 10 Fig. 2c Frequency spectrum 50 Amplitude 40 30 20 10 0 0 1 2 3 4 Frequency [kHz]*1E+4 5 6 7 4 x 10 Figure 2. Phenomena accompanying tests with Sylvatest Duo device. 2a. Full measurement, 2b. Graph for the first 5 ms of test, 2c. Frequency spectrum Figure 2 b shows the first interval of 5 ms of the test conducted on the homogeneous material, and Figure 2 c shows the corresponding frequency spectrum for a range of 0 - 60 kHz. The device indications are consistent with the graphs, with the measured values of signal transition time and maximum signal amplitude to be 50 ns and 588 mV, respectively. Seven main harmonics of the signal with approximate frequencies of 1.3; 3.4; 14.7; 26.7; 31.6; 36.4 and 50.7 kHz can be identified in the graphs. 0.02 0.02 0.02 0 0 0 0 -0.02 -0.02 -0.02 -0.02 0 5 -3 0.02 x 10 0 5 -3 0.02 x 10 0.02 0 5 -3 0.02 x 10 0 0 0 0 0 -0.02 -0.02 -0.02 -0.02 0 5 -3 0.02 x 10 0 5 -3 0.02 x 10 0 5 -3 0.02 x 10 0 0 0 0 0 -0.02 -0.02 -0.02 0 5 -3 x 10 0 5 -3 0.02 x 10 0 5 -3 0.02 x 10 0 0 0 0 0 -0.02 -0.02 -0.02 0 5 0 -3 5 0 -3 x 10 5 5 -3 x 10 0 -3 x 10 5 -3 x 10 0.02 -0.02 0.2 x 10 0.02 -0.02 0.02 5 -3 0.02 5 -3 x 10 x 10 0 -0.2 0 5 -3 x 10 Figure 3 Graphs 3.1-3.17 Time spectrums for ultrasound signal propagated in wood. Axis X - time [ms], axis Y- amplitude [mV]. Column 1, graphs 3.1-3.4 – defect free wood; column 2, graphs 3.6-3.9 – slope of grain; column 3, graphs 3.10-3.13– knots; column 4, graphs 3.14-3.17 – trunk crosswise measurement under an angle of 45°. 5 5 0 5 5 0 0 5 4 5 0 0 4 5 5 4 4 x 10 0 40 0 4 x 10 0 4 x 10 5 4 x 10 0 0 5 4 5 0 0 x 10 0 5 4 x 10 5 4 x 10 0 5 4 x 10 5 5 x 10 4 x 10 0 5 5 0 0 0 5 5 0 0 5 x 10 x 10 5 5 4 0 0 5 5 0 0 0 0 4 x 10 0 5 x 10 0 0 5 5 5 4 x 10 0 0 5 5 0 5 4 x 10 20 0 0 5 4 x 10 Figure 4 Graphs 4.1-4.17 Frequency spectrums for signals propagated in wood; axis X - frequency [kHz], axis Y- amplitude. The layout is analogous to that above Graphs 3.1 through 3.17 and 4.1 through 4.17 show the time and frequency spectrums arranged in columns with each column corresponding to the signal propagated at the successive sections as mentioned earlier, i.e. specimens without structural defects, with slope of grain, with knots and finally in column 4 for the trunk crosswise measurement under an angle of 45°. The last graph in column 1 shows the results obtained for the homogeneous material - the Plexiglas specimen. When evaluating the time spectrums on a visual basis, some differences can be found in the maximum signal amplitude (parameter determined as standard one by Sylvatest Duo) and in the amplitudes of main harmonics. As results from graphs 3.1 through 3.4 and 3.9 through 3.12, the amplitudes of harmonics with higher frequencies dominate over those with lower frequencies, in contrast to graphs 3.13 through 3.16 and 3.5 through 3.8 where the amplitudes with lower frequencies are prevailing. Additionally, the variance was calculated for the results obtained. The initial findings are confirmed by the variance. For the signals with prevailing amplitudes with higher frequencies, the variance is higher and for the specimen without defects and that with knots amounts to 0.351*e-3 and 0.399*e-3, respectively. The variance for the specimen with slope of grain and for the crosswise measurement amounts to 0.156*e-3 and 0.091*e-3, respectively. In the case of homogenous material the variance is greater by two orders of magnitude. Based on the modelscale experiment conducted, it can be concluded that wood's structural defects that are present in the way of ultrasonic signal, affect the signal's variance. So, there are reasonable grounds to conduct further investigations allowing for this parameter as an indicator specific to the type of defect present. It would also be advisable to take measurements of frequency characteristics of the signal tested. To this end, Fourier transformations were calculated for all the signals received. The parts of frequency characteristics concerned are shown in graphs 4.1 through 4.17. The Fourier transformation results were pre-filtered to eliminate the harmonics of lesser importance, i.e. those with amplitudes lower than 0.05. Then, percentage energy distributions for signals with frequency bands of 0-5 kHz, 5-22 kHz, 22-40 kHz and 40-60 kHz were calculated. The signal energies were calculated as sums of signal amplitudes in the individual bands. Energy contributions to the low frequency band strongly increase for wood specimens with structural defects, slope of grain, knots and for the crosswise measurement, which can be regarded here as a measurement at a section with exceptionally high slope of grain. The energy contributions for wood specimens with structural defects, slope of grain, knots and for the crosswise measurement amount to 14, 22, 37 and 64%, respectively. By contrast, the corresponding energy contributions for the band of 40-60 kHz are as follows: 28, 28, 20 and 4%. The basic measurement band, i.e. 22-40 kHz, shows a similar distribution of energy with the highest energy contribution for wood specimens free from defects and lower energy contributions for wood specimens with slope of grain, knots and for the crosswise measurement. The respective energy contributions are as follows: 55, 48, 48 and 31%. The signal frequency band of 5-22 kHz shows the most stable energy distribution with the respective percentages of 28, 26, 20 and 25%. It should be noted for the band of 40-60 kHz that its percentage energy contribution is higher than that of the reference signal in the homogeneous material. The results of experiment are consistent with those reported by Bozhang and Pellerin and confirm that the signal frequencies, excluding the material's natural frequency, are more strongly attenuated. The evaluation of energy accumulated in the individual frequency bands is a successive candidate for the parameter, which allows us to predict structural defects in wood by using ultrasonic signal propagated in wood. CONCLUSIONS In order to use the time spectrums in practice, it is necessary to select parameters that will be calculated by using digital signal processing methods and be closely correlated with actual wood properties. Due to a variety of structural defects in materials of natural origin and possibility of coexistence of various defects in the material, it is extremely difficult to match a suitable set of parameters. Currently, it is of vital importance to propose the suitable set of parameters and collect a big enough base of wood specimens and the related signals propagated in such specimens. The signal covariance and percentage signal energy distribution by frequency band have been proposed as parameters to be used for predicting structural defects present in wood. However, the analysis of signal envelope has been rejected as it depends on electric parameters that trigger ultrasounds. As tasks for the future it is proposed to considerably extend the base of specimens with defects and the defect analysis through the use of band-pass filters. REFERENCES: 1. BOLKOWSKI St. et al., 1986: Elektrotechnika teoretyczna. Teoria obwodów elektrycznych. (Theoretical electrical engineering. Theory of electric circuits) WNT, Warszawa. 2. BOZHANG S., PELLERIN R.F., 1996: Non-destructive evaluation of the degree of deterioration in wood: stress wave frequency spectrum analysis. 10th International Symposium on Non-destructive Testing of Wood Proceedings, p. 99, Lusanne. 3. BUCUR V., 1996: Acoustics of wood as a tool for non-destructive testing. 10th International Symposium on Non-destructive Testing of Wood Proceedings. p. 53, Lausanne. 4. DZBEŃSKI W., 1978: Metody kontroli jakości drzewnych materiałów konstrukcyjnych (Quality control methods for arboreal structural materials) Skrypt. Wyd. SGGW-AR, Warszawa 5. DZBEŃSKI W., 1984: Nieniszczące badania mechanicznych właściwości iglastej tarcicy konstrukcyjnej wybranymi metodami statystycznymi i dynamicznymi (Non-destructive testing of mechanical properties of coniferous structural timber using selected statistical and dynamical methods) Wyd. SGGW-AR. Warszawa. 6. LEWIŃSKA-ROMICKA A., 2001: Badania nieniszczące. Podstawy defektoskopii (Nondestructive testing. Bases of Defectoscopy) WNT, Warszawa. 7. MAŃKOWSKI P. 2004: Badanie modułu sprężystości drewna przy użyciu technik radiacyjnych i ultradźwiękowych (Modulus of elasticity testing using isotope and ultrasonic techniques). Thesis for the degree of Ph.D. WTD, SGGW. 8. MOLIŃSKI W. et al., 2000: Badanie prędkości propagacji ultradźwięku w drewnie napięciowym buka Fagus Silvatica L. (Testing of propagation velocity of ultrasound in beech tensionwood, Fagus Silvatica L.) Materiały 14 konferencji naukowej WTD SGGW. Drewno - Materiał Wszechczasów (Proceedings of 14th Scientific Conference of WTD SGGW on Wood as Material forever) Wyd. Fundacja Rozwój SGGW. 9. MOLIŃSKI W., FABISIAK E., 2002: Selected aspects of ultrasonic wave propagation in wood of Larix decidua Mill. Ann. of Warsaw Agric. Univ., Forestry and Wood Tech., 2002 Special No. I. p. 225-229. 10.SANDOZ J.L., 1996: Ultrasonic solid wood evaluation in industrial application. 10th International Symposium on Nondestructive Testing of Wood Proceedings. p.135, Lausanne. 11.SANDOZ J.L., BENOIT Y., DEMAY L., 2000: Wood testing using acusto-ultrasonic. WCTE 2000 12.ŚLIWIŃSKI A., 2001: Ultradźwięki i ich zastosowania (Ultrasounds and their applications). WNT, Warszawa. 13.ZOMBORI B., 2001: "In situ" non-destructive testing of built in wooden members. NDT.net 2001 vol. 6, no. 3. Streszczenie: Wpływ niektórych wad strukturalnych drewna na charakter widma fali ultradźwiękowej. Metody oceny jakości i przydatności technicznej drewna za pomocą technik ultradźwiękowych i innych akustycznych są na świecie szeroko stosowane w praktyce. W tym celu analizowane są przede wszystkim takie parametry jak: czas przejścia fali przez materiał, maksymalna amplituda sygnału odebranego, współczynnik tłumienia sygnału. Na podstawie tych parametrów można przewidywać ogólną jakość drewna, ale nie można przewidywać jakiego rodzaju wady występują w materiale. Rejestracja widma sygnału propagowanego w drewnie i jego dalsza analiza wskazują, że istnieje również możliwość prognozowania typu budowy anatomicznej drewna na badanym odcinku. Na podstawie wstępnych badań modelowych za pomocą aparatu Sylvatest Duo zauważono korelację pomiędzy wadami anatomicznymi drewna występującymi na drodze przepuszczanego sygnału, a parametrami czasowymi i częstotliwościowymi analizowanego sygnału. Author’s adress: Witold Dzbeński, Tomasz Wiktorski Department of Wood Science and Wood Protection, Warsaw Agricultural University – SGGW, Nowoursynowska 159 str., 02-776, Warsaw, e-mail: [email protected]