Experimental/ computational evaluation of flow stress at high strain rates with application to adiabatic shear banding

Abstract

Experimental techniques are described and illustrated for direct measurement of temperature, strain-rate, and strain effects on the flow stress of metals over a broad range of strains and strain rates. The approach utilizes: (1) the dynamic recovery Hopkinson bar technique recently developed at UCSD (Nemat-Nasser et al., Proc. R. Soc. London A20, 371–391), in which samples are subjected to a single predefined stress pulse and then recovered without having been subjected to any additional loads; (2) direct measurement of sample temperature by high-speed infra-red detectors; and (3) ability to change the strain rate during the course of experiment at high strain rates. In this manner, constitutive parameters of elasto-viscoplastic flow of metals and metallic alloys are established and used in constitutive models for large-scale computational simulation of high strain-rate phenomena such as adiabatic shear banding. For application to this kind of very high strain-rate events, it becomes necessary to further tune the constitutive parameters through additional coordinated experimental and computational efforts. To this end, Taylor anvil tests are performed, accompanied by high-speed photographic recording of the deformation, and the results are compared with those obtained by finite-element simulations, leading to fine tuning of parameters in the material's flow stress. This procedure is illustrated for tantalum and tantalum/tungsten alloys, as well as for VAR 4340 steel. The simulations are performed using the PRONTO-2D finite-element code, in which an explicit constitutive algorithm, recently proposed by the first two authors and co-workers, has been implemented; Nemat-Nasser (1991, Mech. Mater. 11, 235–249), Nemat-Nasser and Chung (1992, Comput. Meth. Appl. Mech. Eng. 95, 205–219), and Nemat-Nasser and Li (1992, Comput. Struct. 44 (5), 937–963). This algorithm is always stable and incredibly accurate, independently of the time or strain increments. The constitutive model and the constitutive algorithm are used to simulate adiabatic shear bands produced in a hat-shaped specimen under controlled conditions (Beatty et al., 1991, Proc. 12th Army Symp. on Solid Mechanics, pp. 331–345), arriving at excellent correlation with experiments.