Effects of randomly rough surfaces on ultrasonic inspection

Bilgen, Mehmet
Journal Title
Journal ISSN
Volume Title
Source URI
Research Projects
Organizational Units
Journal Issue

Surface finish of an industrial part affects its ultrasonic inspection and consequently the surfaces are often machined smooth before the evaluation. Ultrasonic inspection through smooth surfaces has been well studied and understood. A theoretical basis has been established for the characterization of interior cracks, voids and inclusions, and vast amount of literature exists. However, much less is known about quantitative ultrasonic inspection of such flaws in parts with rough surfaces, e.g. machine marks or "as-cast" surfaces. A question arises "When can an industrial part with randomly rough surfaces be inspected robustly using ultrasound?" This dissertation is aimed at answering this question. It (1) develops a rigorous theory used for immersion ultrasonic inspection through randomly rough surfaces, (2) gives simple formulas suitable for engineering use, (3) consequently provides a concrete understanding of the physics of the received signal with the changes in the investigating probe's parameters as well as the statistics of the randomly rough surface, and (4) helps the development of experiments and the interpretations of measurements. Some major findings presented in this dissertation are (1) the observation of a near surface dead-zone for the flaw signal (due to substantially increased attenuation for near-surface flaws), (2) a substantial reduction in roughness-induced noise for focused probes, (3) and a consequent improvement in the signal-to-noise ratio, (4) characterization of randomly rough surfaces (the determination of surface statistics from ultrasonic reflections) and (5) comparison of the theoretical results with the available experimental measurements.

Biomedical engineering, Electrical engineering and computer engineering, Electical engineering (Communication and signal processing), Communications and signal processing