WeldFormFEM
WeldFormFEM is the main finite element solver line for large deformation mechanics, with ongoing work in explicit dynamics, implicit forming, contact stability, tetra and hexa support, and remeshing under severe distortion.
Current focus
- Large strain elastoplasticity
- CPU/GPU solver consistency
- Reduced integration and locking control
- Contact algorithms for forming and compression
- Remeshing workflows for distorted meshes
Implicit forming line
The implicit branch is currently being used for quasi-static large-deformation forming benchmarks, including upsetting/compression and extrusion-style validation cases where contact stability, pressure behavior, and mesh robustness are critical.
In the upsetting/compression benchmark, the 2D formulation uses a mixed U-P approach with selective reduced integration, following the Paulo Martins scheme used in metal forming simulations.
References related to the Martins scheme
Nielsen and Martins, Metal Forming: Formability, Simulation, Tool Design.
Paulo Martins, metal forming reference in Portuguese.
For the 3D model, the implementation does not rely on hexahedra as described in Martins' book. Instead, it follows the MINI element methodology with bubble enrichment for mixed formulations.
References related to the 3D MINI element methodology
Arnold, Brezzi and Fortin, A stable finite element for the Stokes equation.
Andrea Cioncolini and Daniele Boffi, The MINI mixed finite element for the Stokes problem: An experimental investigation.
Compression benchmark summary
The 3D compression benchmark was intentionally performed using a very coarse tetrahedral mesh of about 3,000 elements to keep computational cost low for educational and demonstration purposes. Even with that coarse discretization, the overall load evolution remains in good agreement with Simufact Forming, with a relative RMSE below 5%.
The 3D case starts from an initial mesh of 700 nodes and 2950 elements, while the 2D axisymmetric model uses an approximate element size of 0.5 mm. During remeshing, the target is to preserve the original mesh density as much as possible instead of drifting toward systematic refinement or coarsening.
The quantitative comparison table and benchmark-specific error metrics are reported in the dedicated Upsetting / Compression validation page.