arXiv subjects

Fabian Teichmann

Publications and source records attributed to Fabian Teichmann.

2 recordsLinked to original sources

arXiv2026

Development and Experimental Validation of Novel Evaluation Criteria for Turbulent Two-Phase VOF Simulations in High-Pressure Die Casting

Air entrapment during mold filling critically affects porosity and overall casting quality in High Pressure Die Casting. This study assesses the feasibility of applying the vof method within OpenFOAM to simulate compressible, turbulent mold filling in a thin-walled geometry. Three-dimensional simulations with the "compressibleInterFoam" solver were carried out under ambient initial cavity conditions, using both laminar flow and the k-e turbulence model. The free surface dynamics were examined across a range of inlet velocities to evaluate their influence on interface morphology, cavity pressurization, and gas entrapment. To quantify these effects, three evaluation criteria were introduced: the TIFSA as a measure of oxidation risk, the TMVF as an indicator of filling continuity and air entrapment, and the TIVF as a proxy for surface loading. Results show that turbulence modeling accelerates pressurization and limits the persistence of entrapped gas, with velocity governing the balance between smooth filling, turbulent breakup, and exposure duration. Comparison with experimental casting trials, including CT based porosity analysis and photogrammetric surface evaluation, validated that the model captures key defect mechanisms and provides quantitative guidance for process optimization.

physics.flu-dyn↗

arXiv2025

Fourier Neural Operators for Two-Phase, 2D Mold-Filling Problems Related to Metal Casting

We formulate mold filling in metal casting as a 2D neural operator learning problem that maps geometry and boundary data on an unstructured mesh to time resolved flow quantities, replacing expensive transient CFD. In the proposed method, a graph based encoder aggregates local neighborhood information on the input mesh and encodes geometry and boundary data, a Fourier spectral core operates on a regular latent grid to capture global interactions across the domain, and a graph based decoder projects the latent fields to a target mesh. The model is trained to jointly predict velocity components, pressure, and liquid volume fraction over a fixed rollout horizon and generalizes across different ingate locations and process settings. On held out geometries and inlet conditions, it reproduces large scale advection and the fluid-air interface evolution with localized errors near steep gradients. The mean relative L2 error is about 5% across all fields, and inference is two to three orders of magnitude faster than conventional CFD, enabling design in the loop exploration. Ablation studies show monotonic accuracy degradation under stronger spatial subsampling of input vertices and a smoother decline under temporal subsampling. Halving the training set yields only a small increase in error. These results establish neural operators as accurate and data efficient surrogates for 2D mold filling and enable rapid optimization of gating systems in casting workflows.

cs.CE↗