Optimization of Deep Drawing Processes Using Statistical Design of Experiments

Abstract

Abstract For a deep drawing process some important controllable variables (factors) upon the maximum drawing force are analyzed to find a setting adjustment for these process factors that provides a very low force for the metal forming process. For this investigation an orthogonal array L18 with three-fold replication is used. To find the optimum of the process, the experimental results are analyzed in accordance with the robust-design-method according to Taguchi (Liesegang et. al., 1990). For this purpose, so-called Signal-to-Noise-ratios are calculated. The analysis of variance for this S/N-ratios leads to a mathematical model for the deep drawing process. This model allows to find the pressumed optimal settings of the investigated factors. In the following, a confirmation experiment is carried out by using these optimal settings. The maximum drawing force of the confirmation experiment does not correspond with the confidence interval, which was calculated by analysis of variance techniques. So the predicted optimum of the process does not lead to a metal forming process with very low deep drawing force. The comparison with a full factorial plan shows that there are interactions between the investigated factors. These interactions could not be discovered by the used orthogonal array. Thus the established mathematical model does not describe the relation between the factors and deep drawing force in accordance with the practical deep drawing conditions.