A Statistical Study on Wave Scattering on Rough Surfaces Using Ultrasonic Numerical Simulations

Ultrasonic inspection is a non-destructive evaluation method widely used in various industrial fields for its advantages including deep penetration and safety. The existence or size of a defect is estimated by using the ultrasonic waves reflected from the object to be inspected; defects initiated from natural failure have rougher surfaces. The characteristics of the reflected waves are related to the roughness of the defects. If the surface is significantly rough, ultrasonic scattering occurs and the reflected signal has a low amplitude, but when the surface is less rough, a high response signal can be obtained. In ultrasonic scattering, roughness relative to the input wavelength is more important than its absolute value. Predicting the scattering behavior can be a useful basis for improving the defect detection rate by developing an inspection method based on the scattering information. The roughness of the surface is generally expressed in a statistical manner so that reflections on rough surfaces can be described the same way. A large number of samples are required to obtain statistically meaningful results, but because there are limitations in obtaining such a number of experimental results, it is often the case that statistical results are derived based on virtual experiments. In this paper, we introduce a statistical analysis method of ultrasonic scattering on a rough surface for a virtual experiment of rough surface scattering, and compare the results based on the Kirchhoff Approximation and finite element analysis for normal incidence with vertical reflection of the plane wave.