Residual stresses and shape distortion in composite structures

Abstract: High performance composites usually consist of continuous fibres and a thermoset matrix. One well-known example from the aerospace industry is carbon fibre epoxy composites. For that reason the manufacturing processes of thermoset composites are subject to extensive research. During the last decades process models have been developed for several different manufacturing methods. These process models are either based on linear elasticity or visco-elasticity. Linear elasticity is however too simple to accurately describe the material behaviour during processing. Visco-elasticity on the other hand requires extensive material characterisation and large computer resources for simulation. Residual stresses are formed during manufacturing of thermoset composites due to a number of factors. The crosslinking of the polymer leads to chemical shrinkage. Further shrinkage will add to this during cooling from the cure temperature to room temperature. During cure when the part usually is constrained by a mould, thermal and chemical shrinkage result in residual stresses. At demoulding the in-mould cure stresses are fully or partially released and shape distortions are formed. If the geometry of the component prohibits these shape distortions, residual stresses may lead to an apparent strength reduction or premature failure of the final product. To avoid problems related to residual stresses and shape distortions it is essential with a good understanding of the manufacturing process. This thesis, demonstrate that models for residual stress development and shape distortions must account for following mechanisms; thermal expansion (different in glassy and rubbery state), chemical shrinkage due to the crosslinking reaction and frozen-in deformations. Measured shape distortion of single shaped angle brackets manufactured at different in-mould temperatures have been used to identify and illustrate these mechanisms. The present thesis also presents a simple mechanical constitutive relation, suitable for implementation in a general purpose FE-package. The new mechanical constitutive relation accounts for all mechanisms identified in the first part of the thesis. The constitutive relation is based on linear visco-elasticity where the explicit time dependence is replaced by a path dependence on degree of cure and temperature. This means significant savings in computational time, memory requirements and costs for material characterisation compared to the visco-elastic models that has been commonly used for process modelling.