Design Optimization of Auxetic Composite Metamaterials Using Finite Element Simulations and Random Forest AI Models

The development of biomaterials with tunable mechanical properties is important for addressing stiffness mismatch between implants and bone, although the present study focuses on computational material design rather than implant-level performance. In this work, a novel composite design is proposed in which a 3D re-entrant honeycomb auxetic fiber is incorporated within a solid matrix. A series of 670 finite element models was analyzed by systematically manipulating the elastic modulus ratio of fiber to matrix (E_f/E_m, where E_f is the elastic modulus of the fiber and E_m is the elastic modulus of the matrix) and the fiber-to-matrix volume ratio (VR) to explore their impacts on effective mechanical properties. The simulations demonstrated that by changing the material and volume ratios, the effective Young's modulus and Poisson's ratio can be tuned from 617 to 53,380 MPa and from +0.3 to -0.35, respectively. The dataset was used to train machine learning models for both forward and inverse design. In the forward model, the effective elastic modulus and Poisson's ratio of the composite are predicted from input parameters (E_f, E_m, and VR). Conversely, the inverse model predicts feasible input combinations to achieve target mechanical properties. This AI-assisted framework enables rapid prediction of composite behavior within the investigated design space. It enables the computational design of auxetic composites with tunable stiffness and Poisson's ratio. Overall, this study presents a computational data-driven strategy for designing auxetic composite with tunable mechanical properties for potential future biomedical material applications.