Improved detection of rough defects for ultrasonic nondestructive evaluation inspections based on finite element modeling of elastic wave scattering

Defects which possess rough surfaces greatly affect ultrasonic wave scattering behavior, usually reducing the magnitude of reflected signals. Understanding and accurately predicting the influence of roughness on signal amplitudes is crucial, especially in nondestructive evaluation (NDE) for the inspection of safety-critical components. An extension of Kirchhoff theory has formed the basis for many practical applications; however, it is widely recognized that these predictions are pessimistic because of analytical approximations. A numerical full-field modeling approach does not fall victim to such limitations. Here, a finite element (FE) modeling approach is used to develop a realistic methodology for the prediction of expected backscattering from rough defects. The ultrasonic backscatter from multiple rough surfaces defined by the same statistical class is calculated for normal and oblique incidence. Results from FE models are compared with Kirchhoff theory predictions and experimental measurements to establish confidence in the new approach. At lower levels of roughness, excellent agreement is observed between Kirchhoff theory, FE, and experimental data, whereas at higher values, the pessimism of Kirchhoff theory is confirmed. An important distinction is made between the total, coherent, and diffuse signals and it is observed, significantly, that the total signal amplitude is representative of the information obtained during an inspection. This analysis provides a robust basis for a less sensitive, yet safe, threshold for inspection of rough defects.