Optimization of flow behavior models by genetic algorithm: A case study of aluminum alloy

Abstract

Prediction of the flow stress of materials using a flow constitutive model provides strong support for engineering practice and promotes the continuous development of aluminum alloys and relevant application fields. Optimizing the parameters of flow constitutive models is a key concern to explain and predict the flow behavior. In this study, a genetic algorithm (GA) is used to optimize the parameters of flow constitutive models widely used for the flow behavior of Al alloy including modified Johnson-Cook model, hyperbolic sinusoidal Arrhenius-type model, SK-Paul model, modified Zerilli-Armstrong model, Kobayashi-Dodd model, and modified Fields-Backofen model. AA6061−T6 alloy is used in this study since it has been used as a representative Al alloy. The performance of the models optimized by GA was evaluated through comparative analysis with mechanical test. The Gleeble-3800 thermal simulation testing apparatus was employed to conduct unidirectional thermal compression tests under multi conditions, including different temperatures (573 ∼ 783 K), diverse strain rates vary from 0.001 to 1 s−1, and a range of strains (0 ∼ 0.8). The performance of all the models optimized by GA is enhanced, and the optimization effect of GA on SK-Paul model is most pronounced, which exhibits a maximum correlation coefficient (R) of 0.99731 and a minimum average absolute relative error (AARE) of 6.53%. The findings highlight the validity of GA optimization in flow constitutive models in the prediction of the flow behavior of Al alloy.