Signal noise based transfer function approach for reliability estimation of ultrasonic inspection

In recent years several methods for reducing the burden of experimentation for obtaining Probability of Detection (PoD) curves have been proposed, especially involving the use of numerical simulation. In particular, there is much interest in being able to estimate the PoD capabilities of a given NDE method, target embodiment and material, provided this is known for some canonical material for the same combination of method/embodiment. Ultrasonic experiments on materials with low signal to noise ratio (SNR) are often difficult and time consuming since the higher signal noise causes serious complications. Set in this context, this paper proposes an approach for transferring PoD curves among materials with different SNR values. The classical transfer function PoD approach is based on the hypothesis that the ratio of signals in related quadrants is equal, which requires large datasets for multiple quadrants. The approach proposed here directly deals with SNR instead of the ratio of signals, and thus requires only the experimental data of the parent application. The new approach is illustrated through example cases involving the prediction of PoD curves for ultrasonic inspection of an aluminium plate using the empirical PoD data for the same in austenitic stainless steel and mild steel. The approach is also demonstrated in each of the possible combinations among these three materials.