Work programme
Firstly, modelling algorithms were studied and developed to identify the attenuation rate on this environment for buried pipe. One facet of the project was therefore to obtain realistic values for the relevant acoustic properties of the coating to be studied. Denso tape (Winn & Coales International Ltd) was used for this as it is easily applied as a spiral wrap for experimentation.
An axisymmetric wave mode, T(0,1), was used as the incident modes and its attenuation was measured. To extract the material properties, a trial and error procedure was used by gradually modifying the material properties in the SAFE model, until a good fit was found between the theoretical model and the measurements for all the frequencies. The attenuation rate of 3dB/m, is obtained for Denso coated pipe [1].
The challenge is for the guided wave instrumentation to be sensitive to the small signals, while not overloading the amplification stages with the comparatively very large signals from the low attenuation region, and causing the signals to clip. Figure 1 shows test specimen: 8” schedule 40 steel pipe (219mm diameter, 8.18mm wall), partly coated with Denso tape to simulate the transition from an unburied to a buried section, which was used to illustrate this effect.
Figure 2 shows the effect of adding a delay to the start of the data collection to reduce the total dynamic range of the responses from features of the pipe. Figure 2a shows the result when no delay was applied. The responses from weld 3, which lies under the wrap, and pipe end B, are barely visible. By introducing a 16m delay, the high amplitude welds in the uncoated part of the pipe can be removed and a much higher sensitivity may be used without the signals becoming clipped.
Further tests showed that by moving the transducer to 10.5m from end A, i.e., just before the start of the coating, the background noise level could be reduced. The results are shown in Figure 3 giving at least a 6dB improvement in SNR around weld 3. These signals are visible because a high gain is used to be able to observe the low amplitude response from the coated region. When the transducer is moved closer to the start of the coating, there is less chance for long path length signals from the uncoated region to be present in the data, so the background signals observed in the coated part are reduced.
Post Processing
The signal quality on coated pipes was still poor, therefore it was necessary to examine the possibility for further improvement by the use of post-processing technique. An advanced signal processing method: split-spectrum processing (SSP) was employed to enhance the signal-to-noise ratio (SNR) and spatial resolution of signal response by minimising the effect of dispersive wave modes (DWM) [2].
SSP divides the spectrum of a received signal using a bank of bandpass filters in order to generate a set of sub-band signals at incremental centre frequencies. These are then subjected to a number of non-linear processing algorithms in the time domain to generate an output signal. In details, A Matlab program is developed that takes an unprocessed signal in time domain and converts it to the frequency domain. Then it filters the signal in the frequency domain to generate a set of sub-bands signal and applies a number of different SSP recombination algorithms into these sub-bands to create the output signal. Furthermore, to achieve a good SNR enhancement, the selection of the SSP filtering scheme is studied [2].
Since the velocity of DWM is a function of frequency, dispersive components of such signals vary across the SSP sub-bands whereas the non-dispersive ones stay constant. Hence, use of the SSP technique would suppress regions of the signal that vary across the bandwidth, reducing the effect of DWM, which is one of the main sources of the coherent noise. Therefore, the SSP method developed in this work enables virtually all the coherent noise on the baseline of the traces to be removed. Figure 4 shows the results before and after a 9% cross-sectional area slot defect had been cut in the pipe, midway between weld 3 and pipe end B.
The green trace in Figure 4a is the unprocessed data with a continuum of background signals after the introduction of the defect. The red trace shows the processed result. The real reflectors are preserved and the background signals are removed. The response from the defect is clearly visible. Figure 4b shows a zoomed plot of the defect region; several traces are plotted. The blue line is the unprocessed result without the defect. The green trace is the unprocessed result with the defect present. The post-processed results in Figure 4b are for the cases without the defect (red) and with the defect (black). The presence of the defect response at 17.2m may be clearly seen. The amplitude of the black response is 12dB down on the response from weld 3, which is in line with the expected response amplitudes from a weld and a defect of this size.
Project outcomes
The attenuation rates for guided waves in buried pipes are sufficiently high to cause a major reduction in guided wave test capability. This is an intractable problem, but it has been shown that by developing an understanding of the acoustic properties of the coating, procedural parameters may be selected to reduce the attenuation effects.
Furthermore, careful attention to the effects of the practical data gathering procedure can produce an additional increase in signal quality. It is, nevertheless, necessary to perform further processing of the received signals to provide comparable defect detection to that achieved on bare pipes, and the post processing approach has been shown to be effective in limited trials.
Split-spectrum processing (SSP) was utilised as the post processing technique to minimize the effect of dispersive wave mode (DWM) in the UGW response. The technique significantly reduced the presence of DWM, hence increasing the spatial resolution and the SNR of the signal response in experiments. Enhancement in UGW sensitivity and spatial resolution can lead to detecting smaller defects and increase the inspection range. In this study, optimum filter bank parameters for this application were identified using a brute force search algorithm.
The proposed parameters have been tested for T(0,1) wave mode with the centre frequency of 35kHz. The results indicate that SSP improves the SNR and enhances the spatial resolution, thus, it is suggested that these parameters are suitable for the guided wave testing using T(0,1) wave mode. The Teletest unit was utilised in the experiments to identify the modes of interest using the pulse-echo method. The experimental result confirmed that the proposed technique considerably reduces the level of DWM in UGW response, and the SNR has been improved by approximately 30dB without distorting the relative amplitudes of the signal of interest. These levels of SNR enhancement can potentially make the signal interpretations easier and pave the way for highly reliable guided wave inspections.
For more information, please email contactus@twi.co.uk