Numerical mapping of the flow splitting process and development of a metamodel


Over the last decades, numerical analysis of forming processes has developed into an established method in industry and scientific practices. Many processes can be efficiently analyzed with sufficient accuracy. The finite element method (FEM) is predominantly used in a wide range of software environments. While the industrially established processes can be well represented, there is still further development and research requirements for new processes such as, flow splitting or bend splitting, which are intensively researched in the SFB 666. In the numerical mapping of these sheet metal forming processes, the challenge arises to control the strong mesh distortion in the splitting zone and to reduce the non-stationary effects at the start and end parts of the profiles.


For a numerical simulation of the flow splitting process, the underlying models are to be verified. For this purpose, methods are to be developed which make verification with experimental data possible. In the existing flow splitting process, components with constant cross-section over the length as well as components with variable cross-sections over the length are to be considered.

With the help of developed algorithms, the simulation time of the flow splitting process could be significantly reduced. However, it still takes several days to arrive at the final solution. Therefore, the product developer is unable to react quickly to parameter or material changes. In order to remedy this, metamodels are to be developed.


In order to verify the FE models, geometrical disturbances are to be introduced into the component as an extension of the previous verification methods, in order to be able to compare the occurring changes in the experiments effectively.

For the numerical mapping of the production route of the flexible cross sections, remeshing algorithms are to be developed. The contour as well as the forming zone in the component is known from the preceding simulation step. The geometrical envelope of the component of a forming stage is mapped and a new 2D network is generated in the cross-section and extended along the curve to a 3D mesh. Finally, the results are transferred to the new mesh. This approach increases the quality of the results as well as the computing efficiency.

With the help of analytical assessments and results so far, modelling approaches are to be found which adequately represent the flow splitting processes. These approaches are brought together in a metamodel.


This research project is funded by the German Research Foundation (DFG).