Performance evaluation of crack identification using density-based topology optimization for experimentally visualized ultrasonic wave propagation

Although techniques to visualize ultrasonic wave propagation help to detect defects, quantitative evaluation using these techniques remains a significant challenge. In this study, a topology-optimization-based damage identification method combined with an ultrasonic wave visualization technique is applied to the experimental results, and its performance is evaluated. This method estimates the target crack as the distribution of 'damage parameters' that reproduces the experimentally visualized ultrasonic wave propagation in an inverse analysis model. The objective function of the optimization was the sum of the squared error between the maximum amplitude distribution obtained in the experiment and finite element analysis. Our results suggest that the crack identification results depend on the incidence angle of the ultrasonic wave to the target crack. This is because the low amplitude in the downstream region of the ultrasonic wave to the crack causes low sensitivity of the objective function. Crack identification was performed using the dataset of ultrasonic wave propagation incident from two directions. The crack could be identified with high accuracy provided the sensitivity of the objective function was high in the entire design domain. Based on this mechanism of crack identification, this study proposes two indices to evaluate the validity of crack identification results in actual inspections.