Towards High-Fidelity Simulation of Fiber-Reinforced Thermoplastics Using AI-Driven Intelligence and Molding Simulation

Fiber-Reinforced Thermoplastics (FRTs), in both short- and long-fiber forms, are increasingly adopted in aerospace, automotive, and electronics industries due to their lightweight and high-performance characteristics. However, their broader use has been hindered by the limitations of conventional computer-aided engineering (CAE) tools, which often fail to capture the anisotropic and heterogeneous behavior of FRTs at the component scale. As a result, applications have been largely confined to non-critical structures. This study introduces a novel framework that integrates AI-driven intelligence with molding simulation to improve predictive fidelity while reducing computational cost. The approach emphasizes lighter, physically representative models, shorter runtimes, and more reliable results by directly linking microstructural predictions with component-level performance. Validation was conducted on two benchmark components with available experimental data: an aircraft armrest and a pole extender. Results demonstrate that the integrated framework significantly outperforms conventional finite element methods in accuracy, particularly in predicting fiber orientation and concentration across complex geometries. Computational runtimes were reduced by over 40%, illustrating its efficiency. Furthermore, analysis revealed the importance of in-plane averaging techniques when evaluating fiber concentration, as pointwise sampling may obscure meaningful variations influenced by fiber aspect ratio. Overall, the proposed methodology offers a robust and efficient solution for high-fidelity simulation of FRTs. It supports their wider adoption in primary load-bearing applications and contributes to advancing sustainable material design by enabling polymers to replace metals in demanding engineering contexts.