A multiscale model integrating artificial neural networks for failure prediction in turbine blade coatings

Abstract

The failure prediction of turbine blade coatings remains a significant challenge due to complex microstructure and multiphysics failure mechanisms. A multiscale life prediction model integrating artificial neural networks was developed, which considers the failure mechanism of oxidation, creep and thermal mismatch in microscale, and the combined effect of gas and coolant conditions, film cooling and TBCs in macroscale. The model exhibits better prediction accuracy on interface oxidation, damage evolution and failure region of TBCs on turbine vane. Based on the model, the coupled effect of thermal, oxide growth and thermal mismatch on TBCs failure of turbine vane is discovered.