Estimation of Flaw Parameters in Periodic Rough Surfaces via Ultrasonic Simulation

Currently, the ultrasonic pulse-echo method is widely used for thickness measurement. To enhance the performance of this technique, this study aims to develop a method to predict the internal geometry of pipes with periodic flaws typically caused by flow-accelerated corrosion. Ultrasonic wave spectra reflected from periodic surfaces show pattern specific characteristics, which we utilized for estimating flaw parameters. We conducted simulations that calculated the propagation of ultrasonic waves reflected on different periodic rough surfaces. These simulations were performed under the same conditions as experiments, which replicated pipe thickness measurement using the pulse-echo method. The simulated pulsed wave amplitude spectra showed the same characteristics as experimental waves when reflected from periodic rough surfaces. For parameter estimation, we trained and tuned a neural network model using only simulation data. To evaluate performance under real-world conditions, we tested the model using experimental data. The results showed relatively high accuracy in estimating flaws depth. While improvements are required, further flaw pitch estimation exhibited potential.