Mesh quality and grid convergence

Quality before the solve

Check positive cell volume, skewness, non-orthogonality, growth and geometry conformity. Interpret thresholds using the solver’s definitions. Inspect the worst cells in context, especially near the leading edge, tips, gaps, wakes and interfaces.

Verify the quantities you need

Build a systematic coarse, medium and fine grid sequence where feasible. Keep boundary conditions and model choices comparable, and reduce iterative and time-discretization errors before attributing differences to the mesh. Compare $C_L$, $C_D$, pitching moment and local quantities such as surface pressure or wake deficit. Domain extent and boundary treatment are separate checks.

Richardson extrapolation and the Grid Convergence Index can help estimate discretization uncertainty when a suitable convergence regime and refinement sequence exist. Report nonmonotonic changes or lack of convergence. For LES and hybrid methods, also assess model activity, resolved flow content, sampling and how the effective filter or grid length changes.

Keep validation distinct

Grid sensitivity tests show how the numerical solution changes with resolution. Validation compares against physical measurements under comparable conditions, including Reynolds and Mach numbers, transition or trip, and geometry.

Sources

• NASA Glenn · Examining Spatial (Grid) Convergence
• NASA Glenn · Verification and Validation tutorial
• NASA TMR · benchmark cases and experimental data