Hot deformation characteristics of NiTiV shape memory alloy and modeling using constitutive equations and artificial neural networks

In this study, an alloy conforming to the composition Ni50Ti48V2 (at. %)was cast in a vacuum induction melting furnace, and hot deformation tests were performed on the samples using a GLEEBLE 3800 thermomechanical simulator over a range of strain rates (0.01 s(-1), 0.1 s(-1), 1 s(-)(1), 10 s(-1)) and temperatures (1073 K, 1173 K, 1273 K, 1373 K). Deformation processing maps, which are a combination of instability map and efficiency map, were derived based on dynamic materials model. The mechanisms of deformation were interpreted based on stress-strain plots, kinetic analysis, process maps and microstructure. Material constants along with activation energy, which was found out to be 205.58 kJ/mol., were calculated. Since strain also plays a major role in this analysis, a Strain-Compensated Arrhenius-Type (SCAT) model was developed. A comparison between strain-compensated Arrhenius type and artificial neural network models (ANNM) was made through the use of relative error, adjusted R-2 values and root mean square errors between the predicted and experimental stress values. All three parameters unanimously indicated the use of ANNM for predicting the stresses developed in the material during hot deformation process. Finally, the flow stress required for the onset of dynamic recrystallization (DRX) was calculated using work hardening theory and correlated with the power dissipation efficiency generated from processing maps. Results show that Ni50Ti48V2 exhibits the highest power dissipation efficiency of 43.50% when deformed at 1173 K and 0.01 s(-1). The least critical stress of 58.014 MPa was also captured under the same working conditions. Results are elaborately discussed in the paper. (C) 2021 Elsevier B.V. All rights reserved.

Automated summary

This study investigates the hot deformation behavior and mechanisms of Ni50Ti48V2 alloy through hot deformation tests and dynamic materials modeling. By comparing the prediction performance of two models, the use of the artificial neural network model for predicting the stresses during hot deformation process is determined. Furthermore, the flow stress required for dynamic recrystallization and its correlation with power dissipation efficiency are analyzed.