Assessment of modelling methodologies for prediction of high-temperature creep-fatigue behaviour of Alloy 617

This study aims to assess the accuracy of various modelling methodologies in predicting creep-fatigue damage (d_cf) and cycles-to-crack-initiation (N_i) of Alloy 617 at 850 °C and 950 °C. In the decoupled methodologies, four creep damage models were employed: (i) the time fraction (TF), (ii) ductility exhaustion (DE), (iii) stress-modified ductility exhaustion (SMDE), and (iv) strain-energy density (SED), to capture the creep damage (d_c) contribution to the overall creep-fatigue damage (d_cf), while fatigue damage (d_f) contribution was always calculated using the simple “average” approach. Furthermore, we employed Holmström's integrated ϕ model, which combines the d_cf into a single parameter without a need of separate assessment of creep damage (d_c) and fatigue damage (d_f). The results show that there is no significant difference between the integrated ϕ model and decoupled creep-fatigue methodologies when the creep damage models are calibrated to creep-fatigue (CF) data. However, it is shown that the accuracy of the creep-fatigue damage (d_cf) predictions using the decoupled approach gets significantly worse when the creep damage models are calibrated against independent creep-rupture (CR) data. It is further shown that when using DE, SMDE, SED creep damage models, the d_cf predictions are arranged by strain range, suggesting that these models are better at capturing strain range variations rather than temperature changes. Based on the obtained results it is recommended that any of the employed creep damage models are all suitable for coupling with the “average” fatigue damage model when capturing the creep-fatigue behaviour of Alloy 617, as long as the employed creep damage model is calibrated directly to the experimental creep-fatigue (CF) data.