Bumper Impact Test Damage and Static Structural Characterization in Hybrid Composite Aided by Numerical Simulation and Machine Learning Analysis

Modern automotive design has increasingly embraced plastics for bumper construction; however, it can lead to material degradation. To overcome these limitations, the automotive industry is turning to fiber–resin material, namely carbon–epoxy composites. Our research focuses on determining the effects of fiber orientation and angle alignment on the structural stress of the car bumper, examining the hybrid material (carbon–epoxy reinforced by CFRP) in static structural tests, and performing dynamic impact tests at various speeds, applying the Tsai–Wu criterion as a basic failure model. However, Tsai–Wu’s failure in numerical analysis highlights the limitation of not being able to experimentally distinguish between failure modes and their interaction coefficients. To address this issue, we employ ANSYS® 2024 R1 with a Fortran program, which enables more accurate estimation of failure behavior, resulting in an average error of 13.19%. To identify research gaps, machine learning (ML) plays a vital role in predicting parameter values and assessing data normality using various algorithms. By combining ML and FEA simulations, the result shows strong data performance. Bridging from 2 mm mesh sizing of 50% carbon–epoxy woven/50% CFRP laminate in 6 mm thickness at 0° orientation shows the most distributed shear stresses and deformation, which converged toward stable values. For comprehensive research, total deformation was included in ML analysis as a second target to build a multivariate analysis. Overall, Random Forest (RF) is the best-performing model, indicating superior robustness for modeling shear stress and total deformation.