TY - JOUR
T1 - Material constitutive modeling over a wide strain rate range by integration of cylinder and cap sample SHPB tests
AU - Liu, Tongyu
AU - Zhao, Wenxiang
AU - Xie, Lijing
AU - Peng, Engao
AU - Gao, Feinong
N1 - Publisher Copyright:
© 2025
PY - 2025/10
Y1 - 2025/10
N2 - In this paper, the constitutive modelling method to cover a wide range of strain rate is studied by integrating the cylinder and cap sample SHPB (Split Hopkinson pressure bar) tests. In order to compensate the errors in the stain calculation for cap sample SHPB tests, a correct factor function of the temperature and strain rate is first established. Afterwards, a unification method is proposed to transform both the shear stress and tensile/compressive stresses in hot compression, quasi-static tension, cap and cylinder sample SHPB tests to VON MISES stress. In this way, all the data from different mechanical tests are used for data fitting the material constitutive models over a wide strain rate range. In the consideration of the coupling of thermal softening and strain rate hardening effects, both the classical and modified Johnson-Cook (J-C) constitutive models are developed for AISI9310 steel and ZL702A aluminum alloy. In addition, constitute modelling by means of Artificial Neural Network (ANN) models is explored. Two ANN (Artificial Neural Network) models with 6 and 11 neurons are trained for AISI9310 steel. According to the verification and evaluation with experiments by means of direct calculation and finite element method (FEM) simulation, the modified J-C model behaves best over the entire strain rate range, and the ANN model with 6 neurons wins over 11 neurons due to its success in avoiding the overfitting risk.
AB - In this paper, the constitutive modelling method to cover a wide range of strain rate is studied by integrating the cylinder and cap sample SHPB (Split Hopkinson pressure bar) tests. In order to compensate the errors in the stain calculation for cap sample SHPB tests, a correct factor function of the temperature and strain rate is first established. Afterwards, a unification method is proposed to transform both the shear stress and tensile/compressive stresses in hot compression, quasi-static tension, cap and cylinder sample SHPB tests to VON MISES stress. In this way, all the data from different mechanical tests are used for data fitting the material constitutive models over a wide strain rate range. In the consideration of the coupling of thermal softening and strain rate hardening effects, both the classical and modified Johnson-Cook (J-C) constitutive models are developed for AISI9310 steel and ZL702A aluminum alloy. In addition, constitute modelling by means of Artificial Neural Network (ANN) models is explored. Two ANN (Artificial Neural Network) models with 6 and 11 neurons are trained for AISI9310 steel. According to the verification and evaluation with experiments by means of direct calculation and finite element method (FEM) simulation, the modified J-C model behaves best over the entire strain rate range, and the ANN model with 6 neurons wins over 11 neurons due to its success in avoiding the overfitting risk.
KW - Artificial neural network
KW - Constitutive modelling
KW - FEM
KW - SHPB
UR - http://www.scopus.com/inward/record.url?scp=105004554126&partnerID=8YFLogxK
U2 - 10.1016/j.ijimpeng.2025.105368
DO - 10.1016/j.ijimpeng.2025.105368
M3 - Article
AN - SCOPUS:105004554126
SN - 0734-743X
VL - 204
JO - International Journal of Impact Engineering
JF - International Journal of Impact Engineering
M1 - 105368
ER -