The hot tensile behavior of a typical Ni-based superalloy is elucidated using a microcomputer-controlled electronic universal testing machine at various deformation temperatures (920–1010 °C) and strain rates (0.001–0.01 s−1). Further, the influence of deformation parameters on the fracture characteristics is clarified. The results reveal that the tensile stress speedily rises to a peak level, beyond which it continuously declines until the final rupture. Moreover, numerous dimples exist on the fracture surface, and tenacity nests appear between adjacent dimples, suggesting that micro-void coalescence is the primary failure mechanism. Meanwhile, almost completed dynamic recrystallization (DRX) at elevated deformation temperature and slow strain rate promotes the intergranular fracture of the tested superalloy. Besides, an Arrhenius-type (AT) constitutive model and an artificial neural network (ANN) combined with particle swarm optimization (PSO) algorithms are established and validated. The maximum value of the mean absolute relative error (MARE) for the AT model is 9.74%, while it decreases to 4.20% for the PSO-ANN model. The correlation coefficient (R) increases from 0.9050 for the AT model to 0.9737 for the PSO-ANN model. Overall, it is demonstrated that the established PSO-ANN model has a high prediction performance and is an effective model for reproducing the hot tensile behavior of the tested superalloy.
Read full abstract