Modeling torsional split Hopkinson bar tests at strain rates above 10,000 s−1

Abstract

Finite element analysis is used to study experiments with the torsional split Hopkinson bar technique in which strain rates on the order of 10 4 s −1 are reached by using specimens with a very short gage length. Time-dependent analysis with a viscoplastic constitutive model for the specimen material is used to analyze the whole apparatus (elastic bars and specimen). Results from modeling tests with 1100-O aluminum show that the waves in the elastic bars can be modeled accurately when a significant increase is the flow stress at strain rates of 10 4 s −1 is assumed in the constitutive model of the material. The calculated stresses, strain rates, and strains in the specimen's gage section show that the state is not exactly of pure and homogeneous shear as assumed when the experimental stress strain curve is determined from the measured waves on the elastic bars, and that the plastic zone extends beyond the gage length. Even with these discrepancies the results show that the experimental curve provides a good estimation to the true material response. A time-independent analysis with rate-dependent material model is also used to analyze only the specimen and a short section of the adjacent flanges. The results show that this analysis can also be used to determine the validity of the assumed constitutive model of the material.