NDT.net • Apr 2006 • Vol. 11 No.4

Inspection of Glue-Lines of Glued-Laminated Timber by Means Of Ultrasonic Testing

G. Dill-Langer, W.Bernauer, S. Aicher
Material Testing and Research Institute (MPA),
Otto-Graf-Institute University of Stuttgart, Germany
E-mail: dill-langer@po.uni-stuttgart.de

14th International Symposium on Nondestructive Testing of Wood
May 2005, University of Applied Sciences, Germany, Eberswalde.
Published by Shaker Verlag (ISBN 3-8322-3949-9).


Abstract

In the frame of a recently started research project the feasibility of glue-line inspection by means of ultrasonic testing has been studied. Both, glue-lines within glulam and so called "block gluings", i.e. flatwise bonding of two or more glulam cross-sections have been evaluated. Unlike earlier research attempts on similar topics (e.g. [1-3]), the tests were not restricted to artificial defects like sawcuts or holes, but were performed with realistic glue-line defects such as partially missing adhesive. Specially adopted ultrasound devices with low frequency characteristic and high excitation energy have been used. One focus of the first project phase, reported here, consisted in the identification of appropriate characteristic signal features from the fully recorded ultrasound data and the study of coupling conditions.

Introduction

The introduction and further development of glued wood composites such as, e. g., glued laminated timber ('glulam') built-up of e.g. combined lamination strength classes, laminated veneer lumber ('LVL') and multi-crosslayered wooden plates built-up of lower quality boards ('board plywood') contributed considerably to a more efficient and thereby increasingly sustainable use of wood. However, glued timber products in particular need appropriate quality control, as the decisive advantages (high stiffness and strength) are accompanied by some weaknesses such as sensitivity of strength to production conditions.

Non-destructive evaluation (NDE) methods, being in principle applicable for quality control of every produced piece of glued timber, could be used complementary to todays usually applied destructive testing of small populations of random samples. Furthermore, the trend in timber engineering towards large glued joints and so-called 'block-gluings' increases the need for NDE (post-) production quality control, also at the building site. Compared to NDE application in quality monitoring of the competing non-renewable construction materials steel and concrete, the level of NDE methods applicable to large glued timber structures is still relatively low.

Recently an applied research project on non-destructive evaluation of glued timber products for structural use has been started at Otto-Graf-Institute. Thereby the integrity of glue lines shall be characterised by means of ultrasonic methods. The main goals of the project are

  • development of reliable and easy to use methods for non-destructive detection of glue-line defects in glued laminated timber and so called "block" glue-lines
  • application and calibration of the developed NDE methods for use in production control and in-situ inspection of existing glued timber structures
  • determination of sensitivity of the developed methods against interfering factors such as natural variability of wood and coupling conditions
In the first period of the project the studies focussed on the following topics:
  • testing of appropriate ultrasound equipment,
  • proof of feasibility for detection of glue-line defects in model specimens,
  • evaluation of ultrasound signals by identification of characteristic parameters and
  • optimisation of coupling methods.
In the following some problems of experimental techniques, evaluation procedures and first preliminary results from the on-going research work are discussed.

Experimental Configuration

a) Equipment The ultrasound transmission measurements have been performed by means of a high energy ultrasound generator, (multiple) resonant transducers, broad band amplifiers and a PC based transient recorder system. The ultrasound generator (GEOTRON INC.) being designed for application to US measurement of concrete produces burst-like pulses with central frequencies adjustable in coarse steps between 20 kHz and 350 kHz. Two different types of transducers have been used:

  • Low frequency transducers (main resonance at 20 kHz, GEOTRON INC.) with cone-type tapered ends and a resulting coupling diameter of about 3 mm
  • Medium frequency transducers (main resonance at 350 kHz, GEOTRON INC.) with flat ends and a coupling diameter of 64 mm
The broad band amplifiers (3 dB points at 1 kHz and 1 MHz) provided a built-in band pass filter and adjustable amplification factors up to 100 dB. The entire signals have been recorded by means of fast transient recorder cards with 20 MHz time resolution and 12 bit amplitude resolution.

b) Specimens and test procedure
Two different experimental configurations have been studied: First, the glue-line of two flatwise bonded glulam lamellas has been tested. The model specimen (named specimen A in the following) for this configuration consists of two lamellas with width thickness length of 140mm 32mm 1000mm bonded at the laboratory by means of PRF adhesive with a resulting bond-line thickness of about 0.1 mm. In the middle section of the specimen length no adhesive has been applied over full width and a length of 300 mm, thus representing e.g. a temporary dysfunction of the adhesive spraying device. The transmission measurement has been performed as a line-(B-) scan parallel to grain (fiber) direction with coupling of the sensors to positions at mid-with of the boards. For this test series the medium frequency transducers with flat ends (diameter 64 mm) have been applied. The transducers have been pressed to the timber surface by means of a frame with a screw for compression application. No coupling medium has been used (termed dry coupling). The specimen built-up and the scheme of the test configuration are shown in Fig. 1.


Fig. 1 Specimen and scheme of test configuration of specimen A: bondline of two glulam lamellas

Second, the "block"- glue-lines between two glulam blocks have been evaluated. The "block-glued" specimens consisted of two glulam blocks cut from commercially produced glulam beams then bonded together by means of a two component expoxy adhesive with a resulting glue-line thickness of about 1 mm. The depth of the individual glulam blocks was D = 900 mm consisting of 22 lamellas of 40.1 mm thickness. The width of the lamellas (i.e. thickness of the individual blocks) was W = 120 mm, the length of the lamellas (and thereby of the complete specimen) was L = 900 mm. Three different model specimens have been studied, whereby for all specimens artificial glue-line defects on an inlying area (i.e. area of artificial defect completely surrounded by intact glue-line) of circular shape have been realised (compare Fig. 2):

Specimen B: circular area of 300 mm on one glulam block without adhesive, manufactured with a millcut of 1.5 mm depth and a compressible sealing to prevent adhesive penetration
Specimen C: circular area of 200 mm on one glulam block without adhesive, manufactured with a millcut of 1.5 mm depth and a compressible sealing to prevent adhesive penetration
Specimen D: circular area of 200 mm diameter with an adhesive layer of the same thickness (1 mm) as the intact glue-line areas, however cured before bonding of the two blocks

The model specimens B and C represent the case of missing adhesive or big air bubbles in "block" glue-lines. Model specimen D represents the case of "kissing bond" due to e. g. overrun of open / closed assembly time. The measurements have been performed as line scans perpendicular to the bond-line layer and perpendicular to the grain direction of the glulam lamellas. For this test series the low frequency transducers with tapered ends (diameter 3 mm) have been applied. The transducers were pressed to the timber surface by means of a frame containing springs for application of a rather constant compression force. No coupling medium has been used (dry coupling). The specimen built-up and the scheme of test configuration of specimens B, C and D are shown in Fig. 2.


Fig. 2 Specimen and test configuration of specimen B,C and D: "block" glue-line of two glulam blocks

Evaluation of Ultrasound Signals

Several factors complicate the evaluation of ultrasound signals obtained from the outlined through transmission method applied to glued timber products in structural dimensions:

  • In order to maximise excitation energy, burst-like pulses are generated and applied to resonant transmitters. Thereby high energy US pulses are excited without exact control of shape and frequency content.
  • The US signals exhibit long tails resulting from within-sensor resonances and multiple reflections from the specimen boundaries.
  • Usually not only the (desired) p-waves but also transversal waves, surface waves, plate waves and beam waves are excited interfering among each other and with the respective reflected components.
However, in a simplified consideration (for limitations see e.g. [4]) the begin / on-set of the signals contains only the directly transmitted p-wave, as the wave components detectable in the first part of the signal obviously exhibit the highest phase velocity and travelled the shortest distance from transmitter to transducer. Therefore the evaluation of the signal on-set has been used for quantitative evaluation of low frequency ultrasound signals for characterisation of materials like concrete [5]. The methodology is also well approved for analogous problems of signal processing in the field of geophysics, i.e. location of earthquake hypo-centres [6] and has been adopted for damage location in timber at tension loading by means of acoustic emission analysis [7, 8].

Recently, in the context of ultrasonic detection and characterisation of cracks in glulam beams [9] three characteristic signal parameters have been identified, whereby two out of the three parameters are connected to the on-set of the ultrasound signal and one is related to the global characteristic of the entire signal. The parameters are

  • the time of flight (TOF), defined as the time difference between the excitation of the ultrasound signal and the on-set of the recorded signal
  • the initial amplitude of the first recorded oscillation of the ultrasound signal
  • the global peak-to-peak (pp) amplitude of the entire recorded signal.
The definition of the three characteristic parameters is illustrated in Fig. 3 for one exemplary given ultrasound signal.


Fig. 3 Definition of characteristic parameters derived from recorded ultrasound transmissions

The three identified parameters, evaluated positively in the case of crack detection are here applied to the problem of detection of glue-line defects.

First Results of Transmission Measurements at Model Specimens

The recorded signals of the transmission line scans are evaluated with respect to the above defined three characteristic parameters: time-of-flight, initial amplitude and global amplitude. In all following figures the parameters are depicted in a normalised representation: The respective normalised parameter PN(x) at the position x is calculated from the actual parameter P(x) and the minimum / maximum values of the line-scan Pmax and Pmin according to . (1) Thus the variation of all parameters is jointly normalised to the span between 0 and 1. Detection of defects in a glulam bond-line For an exemplary model specimen with one glulam bond-line (type A, configuration see Fig. 1) the normalised amplitude results, given as function of position x along grain direction are shown in Figure 4a. In Figure 4b the normalised time-of-flight values are presented accordingly. The section of the "artificial" glue-line defects (missing adhesive) is marked in both Figures 4a and b by a grey shaded box. The transmission test results show a considerable increase of time-of-flight and a considerable attenuation of both, initial and global amplitude. For the tested configuration with two lamellas and one bond-line, the global transmission response (represented here by global peak-to-peak amplitude) yields a quite high contrast between well glued sections and the area of missing adhesive. Future tests with specimens of structural dimensions (depths larger than 500 mm) containing more glue-lines will reveal how the contrast of the three different characteristic signal parameters depends on the increased thickness and higher number of glue-lines.


Figure 4 Results of ultrasound through-transmission B-scan measurements of specimen A (inspection of glulam bond-line, for test configuration see figure 1)
a) Results of normalised initial amplitude and global pp-amplitude parameters
b) Results of normalised time of flight values

Detection of defects in a "block-" bond-line The result of transmission measurements at three model specimens (type B, C and D, configuration see Fig. 2) are given in Figs. 5a,b to 7a,b. In all figures the results of line (B-) scans in direction perpendicular to grain direction are depicted. The respective area of the glue-line defect is given as a grey shaded box in all figures. Figure 5a shows the amplitude results for specimen B (missing adhesive within a circular area of diameter = 300 mm). The resulting normalised initial amplitude and the normalised global pp-amplitude are depicted as function of position x. Figure 5b gives the normalised time-of-flight results of specimen B.

Figure 6a shows the amplitude results for specimen C (missing adhesive within a circular area of diameter = 200 mm). The resulting normalised initial amplitude and the normalised global amplitude are depicted as function of position x. Figure 6b gives the normalised time-of-flight results of specimen C.

Comparing the results of specimens B and C, which deviate only with respect to the diameter of the "artificial" glue-line defect, it becomes evident, that the time-of-flight results are less sensitive to the glue-line defects hereby showing large fluctuations and only minor contrasts with respect to glue-line defect detection, particularly in case of smaller defect diameter (specimen C). The global pp-amplitude results show a better performance with limited variations and higher contrast between the sound glue-line and the glue line defect. The parameter initial amplitude turned out to be most appropriate for detection of defect type "missing glue" in "block" glue-lines.


Figure 5: Results of ultrasound through-transmission measurements of specimen B (representing a "block" bond-line with missing adhesive within a circular area with diameter d = 300 mm, for test configuration, see Figure 2)
a) Results of normalised initial amplitude and global pp-amplitude
b) Results of normalised time of flight values


Figure 6: Results of ultrasound through-transmission measurements of specimen C representing a "block" bond-line with missing adhesive within a circular area with diameter d = 200 mm, for test configuration, see figure 2)
a) Results of normalised initial amplitude and global pp-amplitude
b) Results of normalised time of flight values

Figure 7a shows the amplitude results for the "kissing bond" specimen D (cured adhesive within a circular area of diameter of 200 mm). The resulting normalised initial amplitude and the normalised global pp-amplitude are depicted as function of position x. Figure 7b gives the normalised time-of-flight results of specimen D.

As in the case of missing glue with a defined air filled gap between the two adherents, also in the case of cured glue (i.e. "kissing bond" without a defined gap) the parameter initial amplitude is most appropriate for detection of an irregularity in the adhesive layer. Both, global amplitude and time-of-flight show large fluctuations and limited contrast even in the given case of a model specimen.


Figure 7: Results of ultrasound through-transmission measurements of specimen D representing a "block" bond-line with cured adhesive within a circular area with diameter d = 200 mm, for test configuration, see figure 2)
a) Results of normalised initial amplitude and global pp-amplitude
b) Results of normalised time of flight values

Coupling Conditions

The preliminary results presented above have been achieved by "dry coupling", i.e. by direct contact of transducer and timber surface. In order to minimise measurement fluctuations and to increase the signal-to-noise ratio, both important for application to structural sized specimens, some comparative tests on different coupling conditions have been performed. Three coupling conditions have been realised and repeated measurements at well controlled conditions have been performed. The studied coupling conditions with defined force application (see below) were
  • dry coupling (no coupling medium, direct contact between transducer and timber surface)
  • coupling by means of an elastomer film (transducer coupled to the elastomer film of 0.9 mm thickness by means of silicon paste, elastomer film coupled directly to the timber surface)
  • coupling by means of silicon paste

For all three conditions medium frequency transducers with flat ends and a diameter of 64 mm were used. The sensors were pressed to the timber by means of a frame with a compression screw; the compression forces for the transmitter and the receiver sensors were measured by means of two load cells and the compression forces were adjusted equally for all measurements to 1030 N 1.5 %. The coupling procedure was repeated 10 times and the mean value and the standard deviation of global pp-amplitude and time-of-flight have been evaluated. From the measured global pp-amplitude and the peak-to-peak value of noise (evaluated in a zero voltage reference section of the signal) the signal-to-noise ratio (SNR) has been calculated as normalised amplitude value. The results of signal-to-noise ratio are given in Fig. 8a, whereby the SNR-axis is scaled logarithmic. The 2 times magnified values of standard deviation are given as error bars. The results of time-of-flight evaluations are depicted in Fig. 8b, whereby the time-of-flight axis is scaled linear. The 10 times magnified values of standard deviation are depicted in an error bar representation.

The test results reveal, that the use of coupling media lead to a very high increase of the signal-to-noise ratio as compared to the dry coupling by a factor of 36 in case of elastomer film and 82 in case of silicon paste. Somehow surprising, it has to be stated, that the relative coefficients of variation with values of 14.1% (dry coupling), 12.6% (film) and 10.4% (paste) remain in the same order of magnitude for the SNR values.

As anticipated, time-of-flight shows only a minor dependency on coupling conditions. However, it is interesting to note that the mean time-of-flight values for dry coupling are slightly higher (8 to 15%) as compared to those obtained with the more efficient coupling by film or paste. This fact may be explained by the relatively higher noise resulting in some problems of proper identification of signal on-set interfering with noise. Accordingly, the variability of time-of-flight evaluation decreases considerable from dry coupling ( 1.5%) to much lower values in case of film ( 0.75%) and paste ( 0.8%) coupling conditions.

From the comparative study it was concluded that the use of an elastomer film as coupling medium is the best compromise between optimal coupling performance and best practical handling conditions. As a spin-off from the performed study on coupling conditions it has become evident, that the control of the compression force of the transducers pressed to the timber surface is a prerequisite for reproducible measurements.


Figure 8: Results of repetitive measurements with different coupling conditions
a) Results for signal-to-noise ratio (SNR)
b) Results for time-of-flight

Conclusions and Outlook

The reported first results of an ongoing research project show the feasibility of detection of glue-line defects by means of ultrasound through transmission methods. Three characteristic parameters derived from the completely recorded ultrasound signals have been identified, whereby the parameter time-of-flight exhibited best performance in case of a single glue-line between two glulam lamellas. The parameter initial amplitude related to the signal on-set exhibited the best contrast to distinguish between proper glue-lines and artificial glue-line defects in case of so-called "block" bond-lines. With respect to coupling conditions, the use of clamped transducers with compression load control combined with the use of an elastomer film as coupling medium turned out to be the best solution for both objectives, being handling conditions and energy transfer.

The presented results have been achieved with model specimens of relatively small dimensions. Therefore the evaluation of the regarded characteristic signal parameters was only partly affected by the problem of noise. In order to transfer the methods to glued members with structural dimensions, the noise of highly attenuated and highly amplified signals has to be decreased by means of averaging techniques and by digital filtering. One promising attempt in order to develop an appropriate filter algorithm by means of a digital filter based on fuzzy logic principles is presented in another paper of this proceedings [10].

The used ultrasound and signal processing methods will be further developed with regard to sensitivity on a larger number of practical glue-line defects caused by e.g. missing / wrong gluing pressure or wrong resin to hardener ratio. Moreover, the resulting non-destructive test procedure has to be evaluated for interfering influences from natural variability of timber and vs. influences from the industrial environment. The latter goals are part of the actual work plan of the ongoing research project.

Literature

  1. Kessel, M. H., Plinke, B., Augustin, R., Huse, M.: Strength grading of construction timber with large cross sections. Proc. 5th World Conf. on Timber Eng., Vol. 1, Montreux, 1998, pp. 557-562
  2. Klingsch, W., Weber, W.: Erarbeitung anwendungstechnischer Grundlagen zur zerstörungsfreien Qualitätsüberwachung von Holzleimbauteilen mittels Ultraschall. Research Report University of Wuppertal, 1991 (in German)
  3. Klingsch, W.: Zerstörungsfreie Lokalisierung äußerlich nicht sichtbarer Holzschädigungen mittels Ultraschall. Bauen mit Holz 6, 1989, pp. 421 f. (in German)
  4. Berndt, H., Schniewind, A.P., Johnson, G. C.: Ultrasonic energy propagation through wood: where, when, how much. Proc. 12th Int. Symposium on Nondestructive Testing of Wood, Sopron, Hungary, 2000, p. 57-65
  5. Grosse, C. U.; Reinhardt, H.-W., Finck, F.: Signal-based acoustic emission techniques in civil engineering. J. Mat. Civil Eng. 2003, Vol. 15, No. 3, pp. 274-279.
  6. Aki, K., Richards, P.G.: Quantitative seismology; Volume 1. Freemann and Company, New York, 1980
  7. Aicher, S., Höfflin, L., Dill-Langer, G.: Damage evolution and acoustic emission of wood at tension perpendicular to fiber. Holz Roh- Werkstoff 59, 2001, pp. 104-116
  8. Dill-Langer, G., Ringger, T., Höfflin, L., Aicher, S.: Location of acoustic emission sources in timber loaded parallel to grain. Proc. 13th International Symposium on NDT of wood, Berkley, 2002
  9. Aicher S., Dill-Langer G., Ringger T. Non-destructive detection of longitudinal cracks in glulam beams, Otto-Graf-J., 13, 2002, pp. 165-182
  10. He, Y., Manful, D., Dill-Langer, G., Bárdossy, A. *, Aicher, S.: Application of fuzzy logic to signal processing of ultrasound measurements. Proc. 14th International Symposium on NDT of Wood, Hannover, 2005

© NDT.net |Top|