How should prism layers be designed in CFD?

The four numbers that define the prism stack

For a geometric progression of layer thicknesses, these quantities are linked:

Because the parameters are coupled, changing growth rate without checking total thickness or final-layer size can unintentionally change the entire mesh transition.

First layer: connect geometry to the wall model

The y+ relation uses the wall-normal distance from the wall to the near-wall solution location. A mesher commonly asks for the geometric thickness of the first prism or inflation layer. In a simple orthogonal cell, the cell center may lie roughly halfway through the layer, but actual centroid location and software conventions vary.

Start by choosing wall treatment and turbulence strategy, then choose target y+, estimate the physical wall distance, translate that into the mesher's geometric convention, and verify the achieved y+ after solving.

Growth rate: gradual expansion is usually easier to resolve

A growth rate greater than one expands cells away from the wall. Very aggressive growth reduces cell count but can create large jumps in wall-normal resolution and a poor transition to the core mesh.

There is no universal “best” growth rate. The acceptable value depends on the solver, cell type, wall treatment, total thickness, local curvature and how rapidly the physical gradients change. Instead of optimizing a single number, inspect:

• the ratio between adjacent layers,
• the thickness of the final prism cell,
• the size of nearby core cells,
• cell quality where prism layers terminate or collide,
• the solved velocity/temperature gradient across the stack.

A moderate, smooth progression is generally easier to justify than a stack that starts extremely fine and then expands abruptly.

Total thickness: cover the important near-wall region

The prism stack should extend far enough to represent the wall-normal gradients that matter. If it ends inside a steep velocity or thermal gradient, the core mesh must suddenly carry that gradient with a different cell topology and size.

For simple external flow, a boundary-layer correlation can provide an order-of-magnitude estimate of physical layer thickness. For complex CFD—separation, curved passages, rotating machinery, jets, buoyancy or strong pressure gradients—the solved flow is the more reliable guide.

Layer count is a consequence of resolution and thickness

A fixed rule such as “always use 10 prism layers” ignores first-layer size, growth rate and total thickness. Ten layers with r = 1.1 produce a very different stack from ten layers with r = 1.5.

Layer count should be large enough to produce a controlled wall-normal progression while reaching the target total thickness. For wall-resolved heat transfer or skin-friction work, more wall-normal resolution may be justified than for a cost-limited wall-function calculation.

The last prism layer has to meet the core mesh

A high-quality wall mesh can still create numerical problems if the final prism layer meets a much larger core cell abruptly. The transition region deserves its own inspection.

• Compare final prism-layer thickness with adjacent core-cell dimensions.
• Watch for prism collapse at sharp corners, gaps and high curvature.
• Use local surface and volume refinement when the core mesh is too coarse to accept the prism stack smoothly.
• Check solver-specific quality metrics after mesh generation, not just layer count.
• Inspect where prism layers terminate; termination lines can create localized quality problems.

In turbomachinery or narrow passages, gap size can constrain total prism thickness from both walls, so the layer strategy may need local controls instead of one global value.

A practical prism-layer workflow

1. Choose wall treatment and target y+.

   Near-wall modelling comes before geometric layer settings.

2. Estimate first-cell wall distance.

   Use representative fluid properties and local velocity/wall-shear conditions.

3. Translate wall distance into the mesher's first-layer convention.

   Confirm whether the input represents full layer thickness or another quantity.

4. Estimate the boundary-layer scale or required total thickness.

   Use correlations only as starting estimates for simple flow.

5. Choose moderate growth and solve for layer count.

   Check final-layer size and transition into the volume mesh.

6. Generate the mesh and inspect local quality.

   Pay attention to corners, gaps, curvature and termination regions.

7. Run CFD and verify y+ plus wall-normal gradients.

   Remesh based on solved evidence, not solely on the pre-mesh target.

8. Perform a result-oriented mesh sensitivity study.

   Compare the outputs that drive the engineering decision.

Common prism-layer mistakes

• Using the y+ wall distance as first-layer thickness without checking the software convention.
• Choosing layer count by habit while ignoring total thickness.
• Using aggressive growth to save cells and creating a poor core-mesh transition.
• Allowing prism layers to collapse in critical gaps or leading-edge regions.
• Applying one global stack to surfaces with very different physics and length scales.
• Checking average y+ instead of the distribution on critical walls.
• Declaring mesh independence from total cell count while the near-wall mesh remains unchanged.