The NDT of austenitic stainless steel using ultrasound is difficult because the grains divert the direction of the ultrasound and scatter it.  The effect is particularly severe in welds because of the weld grain pattern.  Thus reliable ultrasonic measurements are difficult to obtain, and there is a need to improve confidence in sizing defects.  Significant research progress was made in the early 1990s in the context of the inspection of civil nuclear power plant, using computer simulation models, but there has been little funding since. This project aims to revisit this topic, exploiting new techniques in multi-path ultrasound signal processing, increased computing power, and growth of array transducer technology. This project is being done in collaboration with Andrew Temple (Visiting Professor at Imperial) who was a key contributer in the earlier proprammes of research in this topic.