Nonlinear Finite Element Modeling and Analysis of Metal Hot Forming for Automotive Weight Reduction
MetadataShow full item record
This research develops a methodology for three dimensional, multi-stage, large deformation, finite element contact analysis, which considers the effects of thermo-plasticity and austenite decomposition. The developed methodology is applied to the forming process of a transmission clutch hub which is constructed from UHSS 22MnB5 for the purpose of automotive lightweighting. The entire process chain is considered, including the hot stamping process, in which a high temperature blank enters a stamping press with actively cooled tooling. The press then forms the blank and remains closed, so as to cool the part rapidly enough to induce phase transformations in the material. The resulting part increases in strength by up to 250% during this process. This research focuses on simulating the entire multi-stage stamping process including cold formed stages but with the exception of piercing and trimming operations. The clutch hub is currently manufactured from 2.5mm thick HSLA and a transition to 1.5mm thick 22MnB5 is proposed. This would represent a 38% mass reduction if successful. Typically, 22MnB5 has been reserved for structural components, which have substantially less complex geometry than that of a typical transmission clutch hub, thereby increasing the complexity of the problem. Simulations including austenite decomposition were carried out using the non-linear finite element solver LS-DYNA. In order to validate the numerical models, three criteria were evaluated and compared to experimental data: material thickness, lubrication hole geometry, and Vickers hardness. It was found that the thickness distribution of the numerical model accurately represented the thickness distribution of the cold formed stage. The error in thickness estimation was 0.91% for the cold stage. Thickness change during the hot formed stage was only considered at the splines. The final geometry of the lubrication hole presented by the numerical model also closely resembled that of the experimental result: the error between the numerical and experimental model was 1.4% in the major diameter, where the primary stretching occurred. Finally, Vickers hardness values for both models were compared at 11 points distributed along the radial direction and the error was found to be an average of 13.5%.