Robust Multi-objective Evolutionary Optimization-Based Inverse Identification of Three-Dimensional Elastic Behaviour of Multilayer Unidirectional Fibre Composites

Abstract

This chapter focuses on mixed finite element (FE)-experimental frequency based inverse identification of three-dimensional (3D) elastic behaviour of multilayer unidirectional (UD) carbon fibre reinforced plastic (CFRP) composites using a robust multi-objective evolutionary optimization procedure. This combines numerical sensitivity analysis through FE design of experiments, response surfaces methodology-based meta-modelling and a non-sorting genetic algorithm of second generation. All identifiable 3D elastic behaviours are considered as well as uncertainties of material properties. The sensitivity analyses show that the four engineering constants that describe the two-dimensional (2D) elasticity are dominant for the considered freely vibrating UD CFRP thin plate. Differently from its manufacturer’s assumption as quasi-isotropic and its 2D inverse identification as orthotropic, this sample’s 3D elastic behaviour is identified as transversely isotropic before sensitivity analyses, but orthotropic after the latter, i.e. using the four dominant in-plane engineering constants during the multi-objective optimization.