AutoForm Engineering’s contribution to the Numisheet 2022 conference held in Toronto, Ontario, Canada on July 10-14 was a success. At the conference, AutoForm participated in the first benchmark study on springback prediction of a twist die panel provided by the Auto/Steel Partnership (A/SP). The excellent results achieved demonstrate that AutoForm’s software holds the leading position among the software codes in use. The software excels in speed, accuracy and consistency, both when used by AutoForm and by its customers.
The objective of the benchmark study was to predict the springback of the formed panel stamped under two different processing conditions. The first condition was with a binder load to constrain metal flow, and the second condition was with a draw bead-like feature known as a stake bead, which acts like a miniature punch. As the steak bead contacts the metal in the last several millimeters of the punch stroke, the flow of metal from the binder into the die cavity is stopped. When this happens, the last moments of the forming process are changed from draw forming to stretch forming.
In addition to the challenge of predicting the springback under two very different processing conditions, the benchmark study included different materials: DP 980 and 6000 series aluminum alloy with thicknesses of 1.5 and 1.4 mm respectively. The binder gap was set at 1.6 mm, which is larger than the sheet thicknesses for both materials used in this study.
Twist Die panel with and without stake bead and the corresponding springback
The tooling for this benchmark study was provided by the Auto/Steel Partnership and is called Twist Die due to the tendency of the panel to twist during springback. In addition to twisting, the panels may demonstrate side wall curl. These two issues are the most difficult types of springback to control and compensate in die engineering. Consequently, Twist Die is an excellent tool for springback analysis.
The tooling provided by the Auto/Steel Partnership
The participants were required to report the coordinates (X,Y,Z) of the simulation results in five defined sections across the panel before and after springback at 19 predefined reference points labeled from P0 to P18.
Springback measurement sections and points
Five participants with four software codes took part in this benchmark study. The software codes used were LS-DYNA (participant 1), Inspire Form (participant 2), AutoForm Forming R10 (participants 3 and 4) and Stampack (participant 5). Only participant 5 (Stampack) carried out the simulation using volume elements while all the other software codes used shell elements. All of the participants presented their computation times as well as the results with and without stake beads, as well as for both materials, DP 980 and 6000 series aluminum alloy. AutoForm’s contribution, submitted by Dr. Bart Carleer, Dr. Kidambi Kannan, Theron Short and Dr. Igor Burchitz, is presented below (participant 4).
AutoForm Forming R10 computation time results were the best. Both AutoForm and its customer obtained the same computation time, which was 11 times faster than LS-DYNA, 20 times faster than Inspire Form, and even 137 times faster than Stampack, which was the only participant to use volume elements.
Experimental vs Predicted Results
The accuracy of AutoForm’s results presented in this benchmark study clearly underlines the leading position AutoForm holds in this field. The images below present the experimental versus predicted results of all of the participants in five selected cross sections. The results are presented for all of the defined cases, with and without stake beads, and for both materials, DP 980 and 6000 series aluminum alloy.
AutoForm’s contribution is presented under number 4 and AutoForm’s customer contribution is presented under number 3. In both cases, the results in each cross section are consistently very similar to the experimental results or even match them precisely. For many reference points along the selected cross sections, other participants’ results deviate greatly from the experimental results and do not come close to the level of accuracy of AutoForm’s results. The outstanding results and high level of consistency achieved in this benchmark study can be attributed to AutoForm software’s ease-of-use and effective user guidance. In all of the cases presented below, AutoForm received excellent scores.
DP 980 with stake bead
DP 980 without stake bead
AA 6xxx-T4 with stake bead
AA 6xxx-T4 without stake bead
Average Section Curvature Error
When comparing results, a more objective measure is to quantify the curvature of the side wall curl. In the graphs below, the deviation of the simulation results from the experimental results is represented as a section curvature error. The smaller the error on the vertical axis, the closer the simulation is to the experimental results.
Comparison of curvature prediction and measurement
Comparison of average section curvature errors
The software codes used were LS-DYNA (participant 1), Inspire Form (participant 2), AutoForm Forming R10 (participants 3 and 4) and Stampack (participant 5). The Numisheet benchmark study committee rated the submissions as follows:
At the Numisheet conference AutoForm demonstrated once again that its state-of-the-art software enables users to rapidly, accurately and consistently simulate the entire stamping process. The software excels in:
AutoForm software’s leading position among the software codes in use was confirmed by the outstanding results achieved in this benchmark study. The scores achieved by both AutoForm and its customer demonstrate the software’s exceptional speed, accuracy and consistency. We are pleased that our expertise and cutting-edge technology were so highly rated and acknowledged as the best at a conference as prestigious as Numisheet.
Customer Service+41 43 444 6161