Validation of a perturbed-continuum model for shear localization

Material failure through microscale shear localization is a significant parameter in design of applications involving high deformation rates. Experiments and continuum analysis (Wright, 2002) have shown sudden stress collapse via shear localization may be related to velocity or strain rate perturbations in the vicinity of shear band initiation. This paper examines the validity of a recently developed perturbed-continuum model for predicting the timing of stress collapse. Perturbed-continuum models for rationalizing and predicting localization are highly desirable because they do not require direct modeling of microstructural defects or full numerical resolution of localized flow - for realistic problems, both would require computational resources that are well beyond current capabilities. The present work consists of split Hopkinson bar experiments and finite element analyses of stress collapse in a novel specimen geometry with controlled geometric perturbations, i. e., tilted cuboidal specimens. Experimental observations of the onset of stress collapse are compared with predictions that rely on the local strain rate field obtained with a normal finite element mesh for quantifying the theoretical perturbation.