Efficient process simulation of metallic bipolar plate stamping using a GPU-accelerated one-step solver
Year: 2024
Type: article
Abstract: The mass production of bipolar plates for fuel cells through stamping remains an economic and technical challenge. Specifically, the process design phase using the Finite Element Method is still time-consuming and costly due to the high resolution needed to study the feasibility of flow field design, which entails very large meshes. Up to now, incremental solution methods for elastic–plastic deformation prevail. While one-step solvers have proven to be efficient in analyzing feasibility in conventional stamping, no attempt has been made to adopt one-step solvers for stamping of bipolar plates. In this paper, the effectiveness of the one-step approach for sheet metal forming simulations is assessed in representing the strain and thinning that occur during the forming process. Also, a novel method to incorporate contact conditions in the one-step approach, using the final part’s curvature, is presented. To further reduce the computational cost, a Graphics Processing Unit (GPU) accelerated linear equation solver is implemented. For two sample bipolar plate geometries, it is demonstrated that this contact modeling methodology yields reasonable contact conditions. The predictions of strain and thinning exhibit the same qualitative distribution when compared to more precise incremental simulation setup, though some small quantitative differences can be observed. The computational cost associated with the one-step approach is an order of magnitude lower than that of the incremental method, and utilizing the GPU-accelerated equation solver further decreases the runtime by an additional order of magnitude. Therefore, the presented one-step approach appears to be highly useful and advantageous, especially in the early stages of the process design phase, helping to reduce its overall cost.
Authors: L. Gassler, M. Afrasiabi, M. Bambach
Institutions: ETH Zurich, Inspire
Language: English
Subfield: Mechanical Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
Estimated APC paid: $3,300
OpenAlex ID: W4403954977