Validating the predictive capabilities: A key issue in modelling thermomechanical fatigue life

Virtually all models developed to predict thermomechanical fatigue (TMF) behaviour are limited by the fact that life prediction is based on experimental data generated under conditions significantly different from the actual service conditions of most high-temperature components. The current research strongly supports the idea that life models that are closely related to the relevant microstructural processes provide a more reliable basis for life prediction. It will be discussed how microstructural arguments can be used to extend a given life model to loading situations that cannot be simulated satisfactorily in the laboratory and/or estimate the limits within which reliable life predictions can be made. Data are presented that demonstrate that non-conservative life prediction may result despite a seemingly excellent correlation of model predictions with experimentally obtained TMF life. It is emphasized that TMF tests designed to validate a life model should focus on revealing the presence of potential couplings between the various damage mechanisms such as creep, environmental degradation and cyclic plasticity.