Numerical Optimization of an Industrial Multi-Steps Stamping Process by Using the Design of Experiment Method

摘要

The deep drawing process consists in realizing parts with complex shapes like different kind of boxes, cups, etc., from a metal sheet. These parts are obtained through one or several stamping steps. The tool setup motion of the stamping process is difficult to obtain. It requires practice and special knowledge on the process. The design is long and difficult to optimize. It is expensive in machining tool operations because it needs many trials and modifications on the tool before obtaining the right shape and the good working of the tool. The mains reasons of these difficulties come from the strain heterogeneity, the spring back after the tool removing and the decrease of thickness. There are many influent parameters for this kind of process. They modify directly the shape of the part. These parameters can be listed in three different categories: Firstly, the parameters linked to the tool geometry like the die enter radius, the punch diameter and the clearance between die and punch. Secondly, the parameters linked to the manufacturing conditions like the stamping speed, the lubricant and the blank holder force. And thirdly, the parameters linked to the flank geometry. This paper proposes a systematic method of stamping progression optimization, illustrated on an industrial five part made from five forming steps. By using empirical rules and industrial knowledge, the number of steps, the nominal dimensions and process conditions are defined. From this initial tool definition, the numerical simulation by the finite elements method of the stamping steps is carried out using Abaqus software. Then, the parameters of each stage, which have an influence on the shape of the final stamped part, have to be selected. A range of variation around their nominal values is defined. Experimental designs are used to test the influence of these parameters in the numerical simulations. This allows to establish a mathematical model between the geometrical variation on the part and selected influent parameters. With this model an optimization of these parameters can be realized to find their best values.