|NDT.net Apr 2006 Vol. 11 No.4|
Reflection Measurements at Timber Glue-Lines by Means of Ultrasound Shear WavesG. Dill-Langer, S. Aicher, W. Bernauer
Material Testing and Research Institute (MPA),
Otto-Graf-Institute University of Stuttgart, Germany
SUMMARYIn the frame of an on-going research project the feasibility of ultrasound shear wave reflection measurements at glued laminated timber (glulam) has been studied. In the applied experimental set-up the propagation direction of the ultrasound waves was parallel to the smallest dimension of the glulam beam, being the width direction perpendicular to fiber and parallel to the board edges.
In particular the presented work should answer two questions:
KEYWORDS: Non-destructive testing, ultrasound, glued laminated timber, wood, shear waves, reflection measurements
1. INTRODUCTIONOne of the important factors for the serviceability and safety of structures made of glued timber products consists in the integrity and load bearing capacity of the incorporated glue-lines. In order to develop non-destructive test methods for characterisation of the glue-line performance an applied research project on "Non-destructive detection of glue-line failures in glued timber products" has been started at MPA Otto-Graf-Institute.
In the preparation and within the frame of the on-going project several test and evaluation methods based on transmission measurements by means of longitudinal waves have been applied successfully, see for example [AICHER ET AL. 2002, AICHER ET AL. 2004, DILL-LANGER ET AL. 2005]. The methods using longitudinal waves are especially successful in the cases of a some finite gap (> 0,1 mm) in the area of the glue-line defect, representing for example larger air bubbles within a glue-line performed by pressure free gluing with expoxy adhesives. In those cases the advantages of longitudinal waves (for example simple coupling conditions, high input energy) overbalance the shortcomings (for example the lower sensitivity to boundary layers).
However, the transmission method exhibits some drawbacks, some of them being of practical nature, for example the need of both-sided access to the structural element. Moreover, the contrast between sound glue-lines and defects achievable from the data of transmission measurements always consists in a (small) difference of (large) numbers like differences in amplitude or time-of-flight, thereby being highly susceptible for unfavorable scatter.
In the case of very thin glue-lines (<< 0,1mm) performed by means of usual gluing pressures of about 0,6 to 1,0 N/mm2, there is virtually no gap between the two assembly components, even within defect areas with missing glue. Thus, the compression part of longitudinal waves is transfered equally for sound glue-lines and for glue line defects and the US contrast for longitudinal waves with low frequency and thus large wave length is nearly vanishing.
Therefore, in the presented study shear waves in combination with the reflection method have been used for detection of defects in a thin glue-line. As an example, the secondary glue-line of a "block glued" member consisting of two glulam beam segments has been inspected. The shear wave reflection method has already be used successfully for detection of wood defects such as decay, see [HASENSTAB AND KRAUSE 2005]. However, to the knowledge of the authors, the application to glue-line defect detection has not yet been studied before.
2. METHODS AND TEST CONFIGURATIONBefore utilizing the shear wave reflection method for the inspection of glue-line defects, it has to be proven, that clear back wall reflection signals can be measured in the frame of the given boundary conditions (typical structural dimensions and usual quality of the glued laminated timber, available equipment, etc.). For this purpose a pair (transmitter / receiver) of US shear wave transducers has been applied to the planed surface of a glulam beam with the dimensions height width length = 600mm 114mm 1180mm.
The broad band transducers (PANAMETRICS V150) with a central frequency of 250 kHz and a nominal element size of 25 mm have been driven by a high voltage pulse generator (PANAMETRICS 5058 PR).
The propagation direction of the shear waves was mainly perpendicular to the grain in width direction of the beam. The test configuration is sketched in Fig. 1. The receiver was shifted from the transmitter by 100 mm in length direction (fiber direction). It turned out, that the best results were achieved with polarisation direction of the shear waves parallel to length direction (fiber direction). No coupling agent has been used (dry coupling) and the tranducers have been pressed to the timber surfaces with a given and controlled pressure force of 1000N. The signals have been amplified by a broad band amplifier and recorded by a transient recorder (12 bit amplitude resolution, 10 MHz sampling rate).
Fig. 2 shows a typical signal, whereby two different features are clearly seperated: a first peak (within about 10 to 20 s) resulting from a US wave propagating directly from the transmitter to the transducer near the surface of the timber beam. For sake of simplicity this part of the signal is named "surface wave", although from the measured US velocity it is assumed, that this wave component is a longitudinal wave propagating next to the surface but without the characteristics of the much slower surface wave. The occurrence of longitudinal waves in spite of nominally "pure" shear wave input results from the limited efficiency of the shear wave transducer on the one hand and from the mode conversion effects in the highly anisotropic material timber on the other hand.
Clearly separated from the first peak a much smaller second peak is observed throughout occurring within 160 to 200 s travel time. From literature shear wave velocity values in the range of 1320 to 1372 m/s [Bucur, 1995] and the thickness of 114 mm this peak is clearly identified as the back wall echo of the glulam beam. In the range of 360 s also a very small second back wall echo resulting from double reflection can be found. Although Fig. 2 shows one of the signals with relatively large signal to noise ratio the feasibility of back wall identification could be shown with all performed measurements.
The test specimen for detection of glue-line defects has been manufactured by two stages of gluing:
The secondary gluing has been performed by means of a PRF adhesive at a gluing pressure of 0,8 N/mm2 in a hydraulic veneer pressing device. The faces of the two glulam components have been planed directly before gluing. The resulting glue-line thickness was smaller than 0,1 mm. In the center of the interface no adhesive has been applied at an area of 400 mm length and 300 mm width. At the edge of the artificial glue-line defect a sealing inside a millcut of 1mm depth and 14 mm width prevented penetration of adhesive to the defined defect area.
The lay-up of the block-glued specimen resulted in a sound glue-line with a defined glue-line defect. Due to the surface quality and the applied gluing pressure there is virtually no gap between the two glulam components in the area of the defect.
The test configuration for the reflection measurements are analogous to the echo measurements of the single glulam segment described above: A pair of shear wave transducers is pressed to one surface of the block-glued member. Ultrasound shear waves are applied by the transmitter and the resulting signal is measured by the receiver and recorded by a amplifier / transient recorder system. The coupling conditions, the orientation and distance of the two transducers, the polarisation direction and the data of the US equipment conform to those mentioned above. The whole specimen surface (apart from some centimeters at the edges) has been scanned by shifting the transmitter / receiver pair of transducers within a grid of 50 mm in fiber direction and 32 mm in direction perpendicular thereto. In total 21*17 = 357 signals have been recorded. The test configuration is sketched schematically in Fig. 4.
3. PRELIMINARY RESULTSFigure 5 shows a typical signal recorded at a location within the section of sound glue-line. The comparison with the signal recorded at the single glulam cross-section (Fig. 2) yields some similarities and some deviations:
Figure 6 shows a typical signal recorded at a location within the section of the glue-line defect (missing adhesive). The following features can be stated:
In order to reduce scatter due to the coupling influence or due to different US attenuation of different boards the MMDglue-line is normalised with respect to the "surface" wave amplitude MMDsurface, in all cases being identic to the global MMD of the signal. In Fig. 7b the results of the reflection measurements are given as a 3dimensional C-scan representation, i.e. the parameter normalised reflection amplitude = MMDglue-line / MMDsurface is plotted vs. the location (co-ordinates parallel and perpendicular to fiber direction) of the measurements. For better comparison the location and the dimensions of the glue-line defects are depicted in Fig. 7a in a graphical similar representation, whereby the z-axis has only the meaning to differentiate between sound and defect glue-line sections. The normalised reflection amplitudes show the maximum values within the area of the glue-line defect. The mimimum values are found in the section of sound glue-line. Although there is a high amount of scatter within the defect area, a clear contrast between the sections of sound and defect glue-lines can be observed.
4. CONCLUSIONSThe reported study on reflection measurements at structural sized glue-laminated timber beams shows
ACKNOWLEDGEMENTSThe authors express their gratitude to the "Arbeitsgemeinschaft industrieller Forschungsvereinigungen (AiF)" and the "Deutsche Gesellschaft für Holzforschung (DGfH)" for the grant to the project "Non-destructive detection of glue-line defects in glued timber products" (AiF-No. 13865 N).