Wave-based fault detection in concrete by the Full Waveform Inversion considering noise

Abstract

In the field of non-destructive testing of concrete, the use of ultrasonic waves became state of the art, among other techniques. Nevertheless, there is still room for improvement, in particular concerning the resolution and robustness of the monitoring results.

One approach is the application of the Full Waveform Inversion (FWI) originating from geotechnical engineering, known here as seismic tomography. The forward and backpropagation of waves within a numerical model are deployed to identify material parameters, like the primary and secondary wave velocity as well as density. This approach forms a non-linear, and ill-posed mathematical problem. Its solution is computationally demanding and time-consuming.

Within the presented approach, the Full Waveform Inversion is used on a numerically simulated artificial concrete specimen of relatively large dimensions. Here, the synthetic ultrasound data is compromised by the consideration of different levels of Gaussian white noise. Additionally, several sizes and positions of faults are computed to simulate a variety of possible defects within the concrete material as in reality. Subsequently, the identifiability of the fault together with its recognition quality are investigated. Additionally, recommendations on sensor positioning and layout are given.