Refine regions because they control discretization error in the quantity of interest. A visually dense mesh is not evidence of resolution if the cells are concentrated in the wrong places.
Start from expected flow features
Identify wakes, jets, shear layers, recirculation, narrow gaps, thermal plumes, free surfaces, shocks or rotating interfaces before choosing local controls. Connect each refinement zone to a physical reason.
Build a consistent size hierarchy
Keep surface resolution, near-wall prisms and nearby volume cells compatible. Very abrupt transitions can degrade accuracy and mesh quality even when the finest local cell size looks adequate.
Use volume shapes as controlled refinement envelopes
Box, cylinder and other volumetric controls are efficient when the important region can be bounded geometrically. Make the control large enough to include the feature as it develops downstream, not only its origin.
Watch cell-count multiplication
Halving a three-dimensional cell size can multiply the cell count dramatically. Estimate the cost before applying global refinement and prefer targeted controls where the sensitivity is localized.
Finish with mesh sensitivity
A local control is justified only when the engineering output becomes less sensitive to further refinement. Use at least a structured coarse/medium/fine comparison for the regions that dominate the result.
Run the engineering check
Use the linked Abecator calculator or workflow to turn the setup decision into a quantitative check.
Continue the Abecator path
- CFD Verification Workflow
- Technical Case: Grid Independence
- STAR-CCM+ Professional Workflow Training
- CFD Troubleshooting & Model Review
Primary Siemens reference
STAR-CCM+ is a Siemens product name. Abecator is independent and this article is original engineering guidance; it does not reproduce Siemens documentation or third-party tutorial text.