Detecting microstructural variability in P91 steel using coda waves

Microstructural variability from manufacturing and post-fabrication activities have been attributed to a significant portion of failure in grade 91 steels. For example, heat treatment from welding can change the microstructure locally, which can change the hardness, phase content and grain size. This article explores the use of ultrasonic coda waves to detect microstructural changes in martensitic grade 91 steel. Tube samples with eight different microstructural conditions were simulated using different heat treatments and were tested using coda waves. Six different differential features were extracted from the coda wave signals. Only three features: Loss of correlation, differential curve length and coda wave interferometry showed accuracy > 50%. Further, clustered analysis to test the ability of the features to classify good, mid-level and bad microstructures showed that only loss of correlation had good accuracy. Overall, the study suggests that coda waves can be used to detect subtle changes in microstructure.