FEM optimisation of sheet metal forming

A methodology has been developed for the integration of Finite Element Modelling (FEM), and a gradient search optimisation method for the sheet metal forming process with an emphasis on the selection of an optimisation criterion. By taking into account the natural variation in process variables to be expected during a production run, it is possible to optimise all the input parameters available at the die design stage to ensure that the part has the greatest chance of being formed with the lowest possible defect rate. The variables involved in the optimisation include die geometry, material properties, and process variables. A rectangular cup deep drawing process is used to illustrate the method and criterion selection. Experimental work in the deep drawing of the parts was performed on a 75ton press at Ford Geelong Stamping Plant. The FE model was created in I-DEAS, and simulated using ABAQUS Explicit. The punch, die and blank holder were modelled as rigid bodies, with the blank as S4R shell elements. The total number of elements for the complete model varied between 4597 and 5797. In order to capture as much of the forming data possible, without over complicating the optimisation process, various optimisation criteria were extracted from the strain space FLD process signature. These criteria endeavour to capture the stress-strain state of the part, and uses an algorithm, for each defect mode of interest, to summarise into a single variable. The relative success of the optimisation process greatly depends on how well the criteria describes the part. The modes assessed currently include splitting and wrinkling.