Hybrid topology/shape optimization under uncertainty for actively-cooled nature-inspired microvascular composites

Abstract We present a computational framework for Hybrid Topology/Shape (HyTopS) optimization of actively-cooled microvascular composites under uncertainty. We developed a novel HyTopS optimization scheme for microvascular composites which can perform the topological change during the shape optimization process. This task has been done by introducing a new set of design parameters to add/remove microchannels and change the topology of the network. The profound advantage of the HyTopS method in contrast to a solely shape optimization approach is that the design space in HyTopS is not limited to the topology of the initial configuration. We integrate the HyTopS optimization method with the non-intrusive polynomial chaos expansion (PCE) approach to conduct a robust and reliable design. The non-intrusive nature of the proposed method allows for almost any source of uncertainty to be incorporated virtually in the design optimization process. The PCE representations of the response metrics enable the optimizer to efficiently and precisely approximate the statistical moments, failure probabilities, and their sensitivities with respect to the design variables. We solve several numerical examples to demonstrate the advantages of using the suggested optimization scheme over deterministic optimization method for microvascular composites. The results reveal that the optimized designs obtained from performing optimization process under uncertainty outperform the optimized configurations of the deterministic optimization approach in terms of reducing the sensitivity of the design performance to the various random variables.