A multiscale modelling procedure for predicting failure in composite textiles using an enhancement approach

Abstract

Composite materials have largely been analysed for failure and structural performance from the perspective of an anisotropic homogeneous material despite their evident hierarchical nature. Although such an assumption does substantially facilitate the analysis procedure in a computationally efficient manner, predictions on the occurrence of failure differ from reality as the failure mechanisms within composite materials are different from a constituent perspective. This paper demonstrates the ability of a multiscale dehomogenisation procedure that provides failure information at different length scales, namely: the macro- (continuum), meso- (textile) and micro-scale (constituent). This is achieved by adopting an enhancement approach performing simple matrix operations to study the strain distribution. The hierarchical dehomogenisation procedure using Strain Invariant Failure Theory is implemented in Abaqus UMAT subroutine and verified using simple sanity checks and a ‘manual’ sub-modelling technique utilising an open hole tension coupon as an example. A comparison of the two modelling techniques indicate the similar failure predictions at the meso and micro levels with the hierarchical approach presented in this paper being far more computationally efficient. The failure analysis procedure presented in this paper is subsequently demonstrated on a composite I-beam component allowing failure in composite structures to be observed from a constituent perspective where fibre and matrix modes of failure can be identified and examined from an engineering point of view.