ABSTRACT Hot gas components such as gas turbine blades are loaded with constant centrifugal stress at high temperatures and thus the formation of creep cavitation on grain boundaries is expedited. Conventionally cast polycrystalline Nickel-base superalloys used as blade materials are prone to creep cavitation as unfavourably orientated grain boundaries exist. This research presents a diffusion-based probabilistic creep model which describes the creep cavitation process on grain boundaries. It includes the three mechanisms: pore nucleation, growth, and coalescence. The calibration of the model has been carried out by analysing Alloy 247 as-received specimens and specimens with pre-strain. For the evaluation of the pore numbers and sizes, a deep learning model for pore detection was trained on light microscopic and scanning electron microscopy images. Electron backscatter diffraction images are analysed for further investigations regarding grain orientations and grain boundary angles to the loading direction. The calibrated model allows predictions of pore size distributions over time.
Read full abstract