Evaluating Human Factors in TFM Sizing Accuracy: Statistical Findings from a Round-Robin Programme (Part 2)

This paper reports on the results of the second part of a round-robin programme designed to assess the accuracy of TFM (Total Focussing Method) sizing as it relates to probability of detection truth values. Part 1 of the programme involved analysis of flaw responses simulated using the FMC (Full Matrix Capture) method of acquisition in CIVA simulation software. The report [1] noted that the mean standard deviation of height-sizing error for all participants 0.56mm and the mean standard deviation of error for ligament sizing was 0.88mm. CIVA simulation software is well known for its ability to reproduce ultrasonic results; however, software simulation cannot fully represent the human factors of uncertainty. To address the human factors, a second part to the programme was organised whereby welds with seeded flaws were provided to volunteers from several labs. Their task was to scan the welds using an acquisition format compatible with a TFM algorithm and analyse the data using the TFM algorithm of their choice. Height, length and smallest ligament was to be reported. This report summarises the statistical findings of Part 2 of the programme.