Developments in Ultrasonic Inspection II Application of Real-Time Signal Processing to Ultrasonic Testing

Ultrasonic testing (UT) is widely applied for the inspection to various plants and structures. However, in the application to the inspection of high attenuation materials such as austenitic weld joints and the inspection near the surface, the SNR (Signal-to-Noise Ratio) decreases and it makes it difficult to assess the testing results. To solve this issue, systems of a signal processing by the wavelet transform have been developed. Some verification tests for the specimen with austenitic weld joints were conducted. It is confirmed that the extraction of specific frequency components and processing them could help SNR improve and the assessment of the test results. As for the thickness measurement of thermal sprayed coatings, another system, which is able to apply the discrete wavelet transform to ultrasonic signals, was developed and the field application has been started. The result of the verification tests for austenitic weld joints and the field application of the thickness measurement of thermal sprayed coatings will be presented. INTRODUCTION Ultrasonic testing (UT) has been applied to various structures and the in-service in-spection of the plants. However, the testing for the coarse grained material such as austenitic weld joints or the inspection near the surface, the signal-to-noise ratio (SNR) becomes lower than the testing for conventional testing. Also there is a growing need for the detection of the minute flaws those were difficult to detect by conven-tional UT. To solve these issues, signal processing methods including FFT and filtering processing have been studied. The signal processing used to be done after the data acquisition. Recent computers’ speedup helps the simultaneous signal processing. UT apparatuses with signal processing by the wavelet transform have been devel-oped. Examinations for the specimen with austenitic weld joints were conducted. It is confirmed that the extraction of specific frequency components could help SNR im-prove and the understanding of the test results. As for the thickness measurement of thermal sprayed coatings, another apparatus, which is able to apply the discrete wave-let transform to ultrasonic signals, was developed and the field application has been started. WAVELET TRANSFORM The Fourier transform, which provides transformed signals into the frequency domain, has been commonly applied to the signal processing for various purposes. However, the time axis information in signals is lost by this transform, as is well known. To solve this issue, the wavelet analysis has been developed [1-3] and applied to the sig-nal processing for the nondestructive testing [4-6]. The wavelet transform is like a filter, which extracts frequency components from the signals without losing time axis information. Fig. 1 shows an example of the mother wavelet. By scaling and translating the mother wavelet, similar components to the mother wavelet can be obtained from the signals. By this signal processing, it should be possible to separate and reduce noise found in raw ultrasonic signals. In addition, due to the development of the algorithms, the calculation can be speeded up. Fig. 2 shows the comparison result of calculation. Calculation after transforming the ultrasonic signals into frequency domain becomes faster than the conventional cal-culation (in time domain) and it comes true the real-time processing. APPLICATION TO ULTRASONIC TOFD METHODS Cracks in thickness direction and lack of fusion on the groove face are harmful for the weld joints, and it is necessary to accurately detect such flaws nondestructively to se-cure high structural integrity. Especially, to evaluate flaws in fracture mechanics, it is important to correctly measure the size in the thickness direction. The ultrasonic TOFD method was developed in the U.K. in 1980 and has been a useful technique for di-mensional measurement of flaws. This method is partially used for the inservice inspection of the pressurized boiler components and pressure vessels. In this examination, TOFD method was employed for data acquisition. Test results of 9%Ni steel weld specimen 9%Ni steel is used for inner tanks of the LNG storage facilities and the filler metal used is nickel based materials. Therefore, the weld metal becomes austenitic and grass can be observed in the ultrasonic signals due to development of dendrites. The grass makes it difficult to recognize the flaw indications. The specimen, 50mm-thick 9%Ni steel weld joints with the hot cracks, were pre-pared and testing by ultrasonic TOFD method was conducted. Ultrasonic transducers used were broadband of 10 11 12 13 14 15 16 17 0 2 4 6 8 10 Convolution in Freq. Domain Convolution in Time Domain C al cu la ti o n T im e (s )