Distortion Reduction in Roller Offset Forming Using Geometrical Optimization

Roller offsetting is an incremental forming technique used to generate offset stiffening or mating features in sheet metal parts. Compared to die forming, roller offsetting utilizes generic tooling to create versatile designs at a relatively lower forming speed, making it well-suited for low volume productions in automotive and other industries. However, more significant distortion can be generated from roller offset forming process resulting from springback after forming. In this work, we use particle swarm optimization to identify the tool path and resulting feature geometry that minimizes distortion. In our approach, time-dependent finite element simulations are adopted to predict the distortion of each candidate tool path using a quarter symmetry model of the part. A multi-objective fitness function is used to both minimize the distortion measure while constraining the minimal radius of curvature in the tool path. We find that the predicted reduction of distortion in a quarter model appears to be magnified when extended to the full-scale model with the full-scale model showing a greater reduction in part distortion than the quarter-model. Application of the method to a benchmark problem of a square plate with a central offset feature demonstrates that part distortion reduction by 27% can be achieved.