Reliability-based robust process optimization of multi-point dieless forming for product defect reduction

The product quality of multi-point dieless forming (MDF) is dependent on the process parameters. Moreover, variations in the friction and material properties may substantially worsen the final product quality. In this paper, a technique for compensating for MDF product defects by minimizing the variations in the noise parameters is proposed. This can be attained by a reliability-based robust optimization (RRO) technique to obtain the optimal process settings of the controllable parameters. Initially, two MDF finite-element (FE) simulations of an AA3003-H14 aluminum-alloy saddle shape exhibited a substantial amount of dimpling and wrinkling along with shape errors. FE analyses were consequently carried out using ABAQUS commercial software to obtain the correlation between the control process settings and the variation in the noise with regard to product defects. The best prediction models were chosen from a family of metamodels to swap the computationally expensive FE simulation. Further, a genetic algorithm (GA) was applied to determine the optimal process settings of the control parameters. A Monte Carlo analysis (MCA) was executed to determine how the variation in the noise parameter affects the final product quality. Finally, the RRO FE simulation and experimental results showed that the amendment of the control parameters in the final forming process leads to a considerably better-quality product.