The current thesis deals with the failure modelling in continuously reinforced composite structures taking into account the materials inherent microstructure. For that, first the implementation of a constitutive model for transversely isotropic damage in fibre reinforced composites as a User Material Subroutine for Abaqus Standard (UMAT) is presented. Damage initiation in the transversely isotropic linear elastically modelled composite is governed by the Strain Invariant Failure Theory (SIFT), utilising the first strain invariant and the second deviatoric strain invariant for damage initiation. To phenomenologically account for the materials microstructure, the homogenised macro strain is related to the micro strain by means of strain amplification factors determined from simulations of Representative Volume Elements (RVEs) with different fibre arrangements utilising Periodic Boundary Conditions (PBCs). For damage evolution, linear softening behaviour is assumed. ^The implemented constitutive model is subsequently expanded to account for nonlinearity within the matrix phase of the composite, by means of an additional damage variable evolving as a function of the distortional strain in the matrix phase. This is implemented as a User Material Subroutine for Abaqus Explicit (VUMAT).
Additionally, the Micromechanics Analysis Code based on the Generalized Method of Cells (MAC/GMC), developed by NASA at Glenn Research Centre in Cleveland, Ohio, is examined and composite Repeating Unit Cells (RUCs) are developed, having as input parameters purely the constituents material behaviour. Within the implemented RUCs, plastic deformation and failure of the composite materials matrix phase is taken into account and modelling of fibre failure is based on statistical distribution of the fibre strength by means of the Curtin fibre failure model. ^For validation of the numerically implemented models, experimental tests of composite laminates with varying layups are conducted. These experiments are simulated utilising the implemented models and the obtained results compared to the experimental data.