In order to provide reliable numerical simulations for the design of composite structures, both accurate, physically based material models and high performance numerical codes are necessary. The aim of this thesis is to validate one of these models: the LMT damage mesomodel for laminated composites. A new validation process, based on the evolution of the degradation mechanisms in the material up to failure, is defined. This approach is then applied on two chosen test cases: open-hole tensile tests and static indentation tests, focusing on the scaling effects. The first test case highlights the model capabilities to mirror the failure mode change with ply thickness: from a fiber breaking dominated failure to a delamination dominated one. Nevertheless, it also underlines one of the model weaknesses: the bad representation of localised damage such as splits. A study of the transverse cracking, the delamination and their interaction allow to improve the capabilities of the model and to understand in depth the role of these mechanisms in the structure failure. Concerning the static indentation, a complete experimental campaign was built and performed in collaboration with the ACCIS laboratory in Bristol. It brings out different damage evolution depending on plate thickness that can be used to validate the model. The first simulations performed show that the model does not manage to mirror all the experimental observations, and underline numerical limitations of the finite elements code used.
