Numerical analysis and optimization of hemming processes

Abstract Prediction and minimization of hemming defects are critical to the final quality of automobile products. In this paper, finite element analysis (FEA) is employed to simulate flat surface-straight edge hemming operations. Numerical simulations are conducted on the most representative points selected using a computer experiment design method, which involves orthogonal Latin hypercube sampling and response model building. A novel approach of selecting combined parameters as input factors from engineering analysis is proposed. The response surface models (predictors) obtained can capture the fundamental mechanism of the hemming process. The optimization method is developed based on the predictors instead of on time-consuming FEA simulations.