Sizing Uncertainty in Time-of-Flight Diffraction by Monte Carlo Modeling

Abstract

This study provides a method to estimate of the sizing error of Time-of-Flight Diffraction (TOFD) using a calibration/inspection procedure and a TOFD instrument in accordance with ISO 10863 and ISO 16828. This estimate is based on Monte Carlo simulations using distributions of the underlying error sources. The objective is to estimate the distribution of the flaw height sizing error and to rank the factors affecting the sizing error in a TOFD inspection. The proposed method is demonstrated on a T-shaped carbon steel welded connection with wall thickness 40 mm and flaws with flaw height 5 mm from the back wall. This is a common detail in welded bridges and TOFD is a frequently used inspection technique for fatigue cracks in such a detail. This demonstration case poses two practical challenges. First, for some old bridges, there are no reference blocks of the same material for calibration. This could possibly lead to errors propagated from the calibration. Second, the practice of inspecting the weld by scanning parallel to the weld bead may result in uncertainty of the flaw location relative to the instrument midplane. We found that the errors propagated from calibration are irrelevant for the case of study, and the major contributors to the flaw height sizing error are the uncertainty in picking time-of-flight on the ultrasonic signal and the position of the flaw with respect to the midplane of the TOFD instrument. This last factor is also the main contributor to the systematic underestimation of the flaw height (95% sizing error interval of −1.1 mm in the considered inspection scenario).