Application of four different models for predicting the high-temperature flow behavior of TG6 titanium alloy

Abstract

The high-temperature flow behavior of TG6 titanium alloy at strain rates from 0.001 to 10 s−1 and temperatures from 850 to 1050 ℃ was systematically studied by Gleeble-3500 thermal simulation compressor, and four different constitutive equations (classical Arrhenius (AR) model, modified Johnson-Cook (MJC) model, modified Zerilli–Armstrong (MZA) model, artificial neural network (ANN) model) were derived. EBSD results show that the deformation behavior of TG6 titanium alloy has flow softening characteristics. Furthermore, the calculated correlation coefficient (R) and average absolute relative error (AARE) are used to make an error analysis of the fit accuracy of the above models to evaluate the predictive ability of the TG6 titanium alloy flow behavior. The results of the comprehensive analysis are as follows: on the one hand, it is revealed that different predictive capabilities of the high-temperature flow behavior of TG6 titanium alloy by all the four models based on the meticulous experimental comparison and validation (AR model: R2 is 97.376 % and AARE is 14.16 %; MJC model: R2 is 97.42 % and AARE is 18.5 %; MZA model: R2 is 90.598 % and AARE is 19.32 %; ANN model: R2 is 99.94 % and AARE is 2.14 %), but MZA model is a special case. On the other hand, the MJC and AR model can be considered as the best choices for relatively accurate description and prediction of the flow stress behavior of TG6 titanium alloy when the predictability of the model and the material parameters to be considered.