Detecting small flaws in two-phase Ti-6Al-4V with rough surfaces.

Surface roughness degrades the performance of ultrasonic methods when detecting sub-wavelength flaws. In this work, a roughness-modified doubly-scattered response (DSR) model is developed to enhance the flaw detection method for two-phase Ti-6Al-4V with rough surfaces. Extreme value statistics are used to calculate the confidence bounds of grain noise, then the bounds are treated as a time-dependent threshold for segmenting flaws in C-scan images of two-phase Ti-6Al-4V with rough surfaces. Three Ti-6Al-4V samples with different surface roughness are designed and manufactured for validating the present method; ultrasonic C-scan results show that it can distinguish sub-wavelength flaws (about 1/3 wavelength) in two-phase Ti-6Al-4V with rough surfaces, which can be extended to industry applications.