Springback compensation of industrial parts with the displacement adjustment and springforward methods: Milestone 1

One of the main problems for the deep-drawing process is springback. Generally, the deep-drawing tools are directly derived from the shape of the product. However, when the press is opened after forming, the product will spring back due to internal stresses. In order to produce a geometrically accurate product, the geometry of the tools is compensated. The goal of this project is to develop an algorithm to perform this task automatically, with the use of FE deep-drawing simulations. In this report the two main compensation algorithms are discussed that are mentioned in literature. Springback compensation is always the last step in the process planning of a deep drawn product. It is not always needed, in some cases the product can be made to fit the assembly by simply pushing it back into the right shape. To make the right decision, springback has to be measured and evaluated properly, which can be hard for complex shaped products. Today, manual springback compensation is applied, based on the measurements on prototype products. This is a time consuming process for which a lot of experience is needed. The use of FE simulations can speed up this process, but to fully use the detailed results of such a simulation, a compensation algorithm has to be used. The first algorithm is called the smooth displacement adjustment (SDA) method. The method is based on a direct comparison of the desired shape and the shape after springback. The idea is to compensate springback by reversing the springback deformation field and applying this to the product geometry. The method can also be applied iteratively by using the shape deviation field between the springback (actual) geometry and the reference (desired) geometry. A smoothing/extrapolation function has been added, so the new toolset can be derived directly from this compensated product geometry. The method has been tested on a real industrial part, a trunk-lid inner frame that has been adopted as a benchmark part for the NUMISHEET 2005 conference. The method was very successful, after compensation the mean shape deviation was lowered by 70%. However, some large shape deviation was still present in the product flanges, showing that there is still room for improvement. The second algorithm is called the springforward (SF) method. The principle of the SF method is to compensate springback with the internal stresses that cause it, instead of applying direct geometric optimization, such as the DA method. The method was applied to an academic process, the plastic bending of a strip. It is explained that with the current definition of the algorithm, iterative application is not useful. This problem is solved by adding a push-back stage, so the compensation is linked to the desired geometry again. However, the actual calculation of the compensated geometry suffers from large principal and numerical problems, as was demonstrated using a simple industrial product. 1