Inverse design of constituent materials of 3D hybrid woven composites for tailored effective properties using reduced order model and genetic algorithm

3D hybrid woven composites can exhibit superior mechanical performance compared to non-hybrid counterparts, due to synergistic effects arising from the combination of fibres with different properties. However, the design space of such materials expands exponentially with the number of yarns and candidate fibre types, rendering trial-and-error approaches impractical for identifying a global optimum within this vast parameter space. In this work, we propose an efficient inverse design framework for tailoring the effective properties of 3D woven hybrid composites through the selection of constituent materials. The framework integrates a genetic algorithm as an optimiser to automatically search for optimal constituent material combinations, while employing a reduced order model generated by the proper generalised decomposition method as a rapid forward predictor of effective properties during iterations. The proposed inverse design framework was verified through a case study on 3D hybrid orthogonal woven composites, involving a design space of one million potential combinations of constituent materials. The computational time of material characterisation in the inverse design framework results in an acceleration of approximately 122 times compared to the conventional numerical homogenisation in the case study. The results confirm the framework’s ability to efficiently navigate vast design spaces to identify optimal solutions with high computational performance.