·Table of Contents
Dry Coupling Ultrasonic Characterisation of Flooring Tiles and Pavements
J. B. Santos
Instituto de Ciência e Engenharia de Materiais e Superfícies (ICEMS) - DEE
F.C.T.- Universidade de Coimbra - Pólo II - Pinhal de Marrocos - 3030 Coimbra - Portugal
Dry coupling inspection of flooring tiles and pavements by signature analysis have been the final objective of this work. Some difficulties have appeared in the inspection of those components due the absence of signal reproductiveness. At first, such a problem was verified to come from the air cushion formed between the sample and the silicone rubber tape used as coupling medium, due to the corrugated face of the referred ceramics although that corrugation was small. In order to overcome that problem one has proceeded with the development of a mould with a spherical contour to provide the fabrication of an adequate and functional coupling medium. The use of such a dry coupling medium has given rise to less noisy spectral responses although, whole reproducibility has not been reached at all. However, as will be demonstrated, the signature analysis can be applied with success to detect cracks in the refereed components.
The increasing requirements of high quality patterns demanded by customers have led the manufacturers of ceramic flooring tiles and pavements to implement NDE systems in their production lines. The principal claim of customers refers to surface and subsurface cracks that appear in the components. Most of these cracks are a hard problem even for skill operators as it is verified in practice with the great number of components failed. Concerning to their two faces configuration, only two NDE techniques by ultrasounds can be applied in the inspection of such structures: acousto-ultrasonics[1,2] and/or resonant inspection[3,4].
Since immersion coupling is not allowed by manufacturers, the alternatives are dry coupling and non-contact coupling. In respect of non-contact characterisation the use of low frequencies results in low spatial resolution which is inadequate to the detection of small cracks. Thus, dry coupling has appeared as the only way to provide a good transmission and at the same time to satisfy the imposed requirements.
In the experimental testing using dry contact transducers some problems were observed with the reproduction of the signals when the coupling was implemented with a flat sample of silicone rubber. More precisely, it was verified that such a configuration could not provide reproducible signals even if the transducers were positioned in the same locations and on the same sample maintaining the pressure values. Essentially, this problem has come as a result of the air cushion formed between the component and the silicone rubber tape due to the corrugated face of the referred ceramics. To overcome this difficulty a mould was projected in order to obtain a silicone rubber sample with a spherical contour. This shape has revealed effective because it has prevented the air cushion formation. Thus, as the transducers were pressed the air between the silicone-component system could escape freely providing good transmission and reproducible signals. Of course, there are some pavements which surfaces do not allow the use of these techniques because of their roughness.
A) Material samples
Acousto-ultrasonic signals were collected on several flooring tiles panels of dimensions 20 cm by 20 cm by 0.8 cm. Experimental signals were also obtained on two 12 cm by 8 cm by 0.4 cm specimens of plexiglas.
B) Acousto-ultrasonic system
Figure 1 shows the experimental set-up used for the generation, acquisition, visualisation and saving of the ultrasonic signals.
Fig 1: Schematic diagram of the acousto-ultrasonic system|
The pulser (Panametrics 5800PR) generates a broadband pulse with high amplitude. Two 0.5 MHz, 1" element size, broadband transducer manufactured by Imasonic were used firstly with 2 mm thick silicone rubber dry coupling pads and 0.6 Kg/cm2 pressure, and then with 10 mm silicone rubber spherical calottes and the same pressure. After being acquired the received signals were displayed on the screen of a digital oscilloscope then sampled and digitised and finally transferred via the RS232 serial port to the microcomputer for processing. Spectral analysis was performed with the aid of the fast Fourier transform (FFT) algorithm.
A) Plexiglas specimen spectra
Two plexiglas specimens with dimensions already mentioned, were subjected to inspection with only one objective: verify the behaviour of its signatures when all conditions of testing are maintained. Also it was studied the evolution of signal spectra collected from a same specimen. For each acquisition, the probes were put back in the same positions and submitted to the same pressure approximately. Figures 2a and 2b are the time domain signals for the two specimens.
|Fig 2: Plexiglas samples: (a) and (b) time domain signals; (c) and (d) spectral responses
Figures 2c and 2d show the frequency spectra of signals outlined in figures 2a and 2b. The acousto-ultrasonic time domain signals are, clearly, the superposition of many pulses arriving at the receiving transducer. The different time associated to each pulse leads to the spreading of the signals. It is apparent from figures 2a-b the existence of different time segments with the relatively high frequency ones near the beginning (0 - 0,25ms). The remaining signal put into evidence the dominance of lower frequencies. This behaviour is a consequence of attenuation. It is well known that higher frequency waves are more attenuated than lower ones and it also waited that waves propagating long paths be more influenced than those travelling shorter paths.
The acousto-ultrasonic signal spectra shown in figures 2c-d demonstrate the previous explanations about the time domain signals. A high magnitude component is verified for a frequency at about 50 KHz as was expected. It was also observed that higher frequencies are strongly attenuated and the receiving transducer could not receive most of them. Remember that the central frequency of transducers is 500 KHz and they have a 60% bandwidth, however, the maximum collected frequency of significance is about 220 KHz.
Meantime, the important goal to reach with this work is to provide reproductive signals from different specimens since they present the same characteristics. From figures c and d it is clearly observed that almost all resonant frequencies agree. There is a little difference in the spectra between 200 KHz and 250 KHz, which is due to a low sampling frequency.
B) Ceramic spectra
i) Dry coupling pads of silicone rubber.
Flooring tile panels (20cm x 20cm x 0.8cm) were examined to test the practicability of using signatures as a way to check their integrity. Firstly, in order to satisfy the requirements of manufacturers who impose a dry process, it was used silicone rubber pads as coupling media between transducers and the panels. For each sample inspection the transducers were submitted to the same pressure. The comparative study of the calculated spectra has permitted to conclude the following:
Figures 3a-b show the spectra of two samples, which were identified as in good condition. However, from these representations, it is not easy to confirm that presupposition.
- Pronounced differences can be observed in the spectra at resonance frequencies level;
- The fundamental resonance experiments some shifting from sample to sample.
|Fig 3: (a and b) Square amplitude responses versus frequency for two samples considered as in good condition (non-defected) using dry coupling pads of silicone rubber. |
ii) Dry coupling spherical callote of silicone rubber.
Spherical callotes of silicone rubber were produced, using a mould, in order to improve the coupling between the transducers and samples. The same samples and pressure values were used in this new coupling configuration. Again, the acquired signals were subjected to the FFT in order to have their spectra. The responses seem to be better defined and less noisy than the ones shown in figure 3. All spectral responses are between 360-500 KHz range with the maximum amplitude around 400 KHz. It is apparent a shifting of spectra towards lower frequencies. Such is a consequence of the attenuation along the callotes that present a 10 mm thickness. Also not all resonances agree. Figures 4a-b outline the spectra of two samples considered in good condition.
|Fig 4: (a and b) Square amplitude responses versus frequency for two samples considered as in good condition (non-defected) using Dry coupling spherical callote of silicone rubber|
The differences verified between the two spectra of figure 4, considering that the two samples are in good condition, are due to the non-homogeneous structure of that kind of ceramics. As a result of sinterisation micro-cracks, voids and porosity can appear leading to distinct behaviours. This conclusion seems obvious whether one compare figure 2 (c and d) with figure 4.
Remember that the goal is to detect surface and subsurface cracks that can appear in the components. So the following question can be put: - Is it possible to detect the mentioned cracks when the signals are not reproducible? The answer is yes and can be confirmed by the spectrum shown in figure 5 which represents a defected pavement. The existence of a crack on the component surface has lead to the appearance of an additional resonance around to the 220 kHz frequency. Since this resonance is well defined and separated from the envelope of the principal spectrum the characterisation of flooring tiles and pavements can be made by signature analysis with success.
Fig 5: Spectral response for a specimen with a surface crack|
Surface and sub-surface crack detection appearing in ceramic pavements and flooring tiles during sinterisation by signature analysis has been the goal of this work. Requirements of manufacturers have imposed the dry coupling as the only way of ultrasonic propagation. Silicone rubber was verified to be adequate as propagation medium between transducers and components. Two silicone rubber configurations have been used:
It was verified that the use of spherical callotes promotes a more effective transmission of ultrasounds because they prevent air cushion formations. Smaller pressure was necessary to provide good transmission.
- (i) 2-mm thickness pads and
- (ii) 10-mm thickness spherical callotes.
Signals with high degree of reproducibility were not possible at all. Such is due to the inhomogeneous character of the studied ceramics and also to the development of micro-cracks, voids and porosity after sinterisation which alter the ultrasonic responses. Yet, surface and sub-surface crack detection in ceramic pavements and flooring tiles by signature analysis is possible as was demonstrated.
Since relevant cracks appear at the component surface and sub-surface the use of angle beam transducers can provide better results because some unwanted propagation modes can eliminated leading to less noisy spectral responses. This new transducer configuration and corresponding analysis of acquired signals will be object of a future work.
- John C. Duke, Jr., Acousto-Ultrasonics - Theory and Application, Plenum Press, New York, (1988).
- K. Choudhury and K.K. Phani, Thermal Shock Damage and Thermal Fatigue of Glass-An Acoustoultrasonic Study, Materials Evaluation, (1994).
- G. Hands, A new approach with the Resonant Fingerprint Method, INSIGHT, vol.41,nº8, (1999).
- G. Hands, Resonant inspection for mass production industries, INSIGHT, vol.39, nº12, (1997).
- MATLAB for Windows, Signal Processing Toolbox, 1994.