Common‐refinement‐based data transfer between non‐matching meshes in multiphysics simulations

TLDR

Multiphysics simulations with partitioned components use separate meshes whose interfaces are generally non‑matching, requiring data exchange that must be numerically accurate and physically conservative. The study introduces accurate, conservative, and efficient data‑transfer algorithms that use a common refinement of two non‑matching surface meshes. The algorithms employ a common refinement of the two meshes to minimize transfer errors in a chosen norm while enforcing strict conservation, and they are benchmarked against traditional data‑transfer techniques. Numerical experiments show that common‑refinement methods outperform alternatives, particularly for repeated transfers, and the approach also succeeds in complex 3‑D non‑matching mesh scenarios. Simulation data (e.g.

Abstract

Abstract In multiphysics simulations using a partitioned approach, each physics component solves on its own mesh, and the interfaces between these meshes are in general non‐matching. Simulation data (e.g. jump conditions) must be exchanged across the interface meshes between physics components. It is highly desirable for such data transfers to be both numerically accurate and physically conservative. This paper presents accurate, conservative, and efficient data transfer algorithms utilizing a common refinement of two non‐matching surface meshes. Our methods minimize errors in a certain norm while achieving strict conservation. Some traditional methods for data transfer and related problems are also reviewed and compared with our methods. Numerical results demonstrate significant advantages of common‐refinement based methods, especially for repeated transfers. While the comparisons are performed with matching geometries, this paper also addresses additional complexities associated with non‐matching surface meshes and presents some experimental results from 3‐D simulations using our methods. Copyright © 2004 John Wiley & Sons, Ltd.