AI & ML
N -phase flow simulations using X-Mesh and an explicit marker interface representation
arXiv cs.NA0 views
arXiv:2611.13921v1 Announce Type: cross Abstract: We present a marker-based X-Mesh framework for incompressible N-phase flow simulations with sharp, explicitly tracked interfaces. The interface is represented by Lagrangian marker chains, while the computational mesh is deformed so that selected mesh edges conform to the tracked geometry without changing mesh connectivity. This combination preserves the advantages of body-fitted finite element discretizations while allowing highly distorted and nearly degenerated elements to capture fine interfacial features on relatively coarse meshes without requiring mesh refinement. The arbitrary Lagrangian-Eulerian (ALE) finite element solver uses a pressure field that is continuous within each phase and discontinuous across the fitted interface. Surface tension is introduced through a variational membrane, so the pressure jump is resolved without prescribing a curvature-dependent correction; a Young wall energy provides the corresponding treatment of fluid-fluid-solid triple points. Topological changes are handled geometrically from the marker representation. Numerical experiments illustrate the ability of the approach to transport complex interfaces and to control topology changes through marker-based feature-size criteria.
Source:https://arxiv.org/abs/2611.13921PreviousMultifidelity Operator Inference for Non-intrusive Model Reduction under Data ScarcityNextLengthscale Estimation for the Gaussian Kernel and Deterministic Response Functions
Related posts
A useful vocal isolate is not a karaoke track
0dev.to AI
How to Pick the Right AI Development Partner for Your Business
0dev.to AI
DeepSeek R1 vs OpenAI o1-preview: Reasoning Accuracy, Token Economics, and Self-Hosting TCO
0dev.to AI

After the AI Writes, Part 1: Don't Carry Someone Else's Work Into Your Commit
0dev.to AI
My deploy script got my site removed from Bing and ChatGPT search. Seven words got a human to fix it.
0dev.to WebDev
Diluting Poisoned Data Made the Backdoor Stronger: Anthropic's 'Homeopathic Learning' Explained
0dev.to Security