UC San Diego

A multiscale multi-fidelity robust topology Optimization (MFRTO) is proposed to design continuous fiber-reinforced composites (CFRC) components minimizing the structural compliance mean and variance subjected to material uncertainties within a prescribed volume constraint. The topology optimization problem employs the level set method, where the fiber orientation is continuously updated following principle stress approach. At the macro level, multi fidelity Monte Carlo (MFMC) is used to compute the mean and variance of the structural compliance utilizing a density filter creating lower mesh-related fidelity models from the highest fidelity models. Likewise, the mean and variance of the macroscopic elastic properties of a unidirectional fiber-reinforced composite are computed via different fidelity numerical homogenization approaches, specifically the generalized method of cells (GMC) and High-Fidelity GMC (HFGMC). Numerical simulations are performed to validate the proposed algorithm, showcasing peculiar features of the achieved optimal solutions with respect to the topology found in the case of deterministic materials. The paper also discusses the computational efficiency of the approach, particularly the speed-up achieved compared to the traditional Monte Carlo method.