Ansys Fluent Parallel Solver: Re-Partitioning Triggers When Dynamic Mesh Updates Shift Load Balance

Re-Partitioning During Dynamic Mesh Updates

Ansys Fluent's parallel solver distributes the mesh across compute cores, using MPI for inter-node communication and threading within each node, while ghost cells keep boundary data consistent between adjacent partitions. When dynamic mesh events deform or remesh cells in a transient run, per-core workload can drift until communication overhead erodes the benefit of the added cores.

The unresolved point is what actually initiates a re-partitioning event in that situation. Assuming a recent release such as Fluent 2023 R1 or later, the default imbalance threshold, the minimum interval between redistributions, and whether either is exposed as a user-settable parameter are not stated in the standard workflow documentation. The behavior is expected to be release-dependent, so any answer should be checked against the Parallel section of the Fluent Theory Guide for the version in use.

• What default imbalance threshold governs automatic load balancing during dynamic mesh updates?
• Can the trigger threshold or re-partitioning interval be tuned to favor stability over frequent redistribution?
• Does the trigger behave differently in CPU-only parallel runs versus GPU-accelerated solver configurations?