Objective modelling of quasi-brittle failure using an anisotropic gradient-enhanced damage model

Abstract

It is widely recognized that a constant length scale parameter in integral and differential non-local damage models leads to incorrect representation of failure mechanisms either by spurious damage growth [1] or by incorrect damage initiation and propagation [2]. In this contribution we therefore propose a gradient-enhanced damage model based on an anisotropic interaction kernel with evolving length scale, inspired by the stress-based nonlocal damage model by Giry et al. [3]. Special attention will be paid to its objectivity with respect to spatial and temporal discretisation. More specifically, the relation between mesh size and length scale as well as linearisation aspects will be investigated.