Picking of first arrival times on noisy ultrasonic S-wave signals for concrete and rock materials

Abstract

Ultrasonic testing techniques are non-invasive and generally used in geosciences to obtain the longitudinal and shear wave velocities for either concrete or rock materials, which are essential to investigate the physical and mechanical properties of the materials. However, the accuracy and reliability of the material velocities depend on the precise reading of the first arrival time directly on the recorder. In ultrasonic testing, due to the heterogeneity of the samples and noise, it is often problematic especially to determine the S-wave first arrival time both directly on the recorder and from the recorded signal, and this process mainly depends on user experience. This study focuses on the semi-automatic picking of the first arrival time (FAT) of ultrasonic shear (S)-wave signals. For this, after an application of a band-pass filter (BP) to suppress the noise components, the cross-correlation (CC) technique using an operator signal estimated by Kolmogorov spectral factorization from the filtered signal is applied to determine the boundaries of the possible time interval of the FAT. An automatic search then reads the FAT which encounters the maximum amplitude value within the interval. The technique has been tested for synthetic and real data sets. The results show that the FAT can be picked within safe and acceptable limits within errors ± 2.0 µs and ensure that the velocities of materials such as rock and concrete will be obtained accurately. Therefore, this also provides the ability to calculate other related physical and mechanical parameters of the materials.