Reliability-based design optimization of composite battery box based on modified particle swarm optimization algorithm

The application of carbon fiber reinforced polymer (CFRP) material introduces great challenges to the optimization design process, such as complex non-linear material behavior, the inherent uncertainty of design variables and multilevel characteristics of the structure. This paper aims at developing a reliability-based design optimization (RBDO) method to solve the CFRP battery box lightweight design problem considering both meso- and macro-scopic parameters. The method has three kernel parts: the uncertainty quantification and propagation part, the finite element analysis part and the optimization part. In the first part, the internal geometry variability of plain woven CFRP was obtained by X-ray micro-CT images. Representative Volume Element (RVE) models are established to predict the elastic and strength properties of the studied composites, and the constitutive model of material was adapted in stiffness and strength analysis of the battery box structure in the second part. Then a RBDO procedure considering design variables across two scales is developed using a modified particle swarm optimization and surrogate modeling techniques. The structure of the CFRP battery box achieved by the proposed multiscale optimization procedure realizes a weight loss of 22.14%, and the performance demands are satisfied with high reliability, which further reveals the advantages of using this methodology.