ashfordeOU/aero-agent-skills / skills/aerodynamics/cfd/cfd-mesh-generation
cfd-mesh-generation
VerifiedUse when the task is CFD mesh generation, grid type selection, prism layer setup, near-wall resolution, domain sizing, or cell quality checking for a solver run. Generate a CFD mesh for an aerospace flow case: choose between structured, unstructured, and hybrid grids, size the near-wall first cell height from a y plus target and skin friction coefficient, build boundary-layer prism layers with a growth ratio, flag cell quality with skewness, orthogonality, and aspect ratio checks, size the far-field domain, and plan refinement levels. Produces the grid type recommendation, first cell height, prism layer count, quality verdict, cell count estimate, and refinement plan. Trigger: mesh generation, grid types, prism layers, first cell height, skewness, orthogonality, aspect ratio, domain size, mesh refinement.
CFD Mesh Generation (aerodynamics/cfd/cfd-mesh-generation)
Use when the task is CFD mesh generation for an aerospace flow case: choosing the grid type, sizing the near-wall spacing from a y plus target, building boundary-layer prism layers, checking cell quality, sizing the far-field domain, or planning refinement.
Domain quick reference
- Grid types: a structured grid is hexahedral with implicit point ordering, efficient for simple geometries such as airfoil sections, cascades, and ducts; an unstructured grid of tetrahedra or polyhedra fits complex geometries such as a full aircraft; a hybrid grid stacks prism layers on the walls and fills the rest with unstructured cells, the standard for wall-bounded external aerodynamics.
- Near-wall resolution: the first cell height comes from the dimensionless wall distance y+ = y * u_tau / nu, so y = y+ * nu / u_tau; u_tau = v_inf * sqrt(cf / 2) follows from the skin friction coefficient. A target y+ near 1 resolves the viscous sublayer, values up to about 30 blend wall treatment, and wall functions tolerate y+ up to about 300.
- Boundary-layer prism layers grow the first cell outward with a geometric growth ratio, commonly 1.1 to 1.3, until the boundary-layer thickness is covered; the layer count follows the geometric series sum h1 * (r^n - 1) / (r - 1).
- Cell quality metrics: skewness near 0 is good and above about 0.9 is bad, the orthogonality angle should stay above about 20 degrees, and the aspect ratio should stay small for general cells while boundary-layer cells legitimately run high.
- Domain sizing: far-field boundaries sit tens of chord or body lengths away, commonly 20 to 50 chords for external aero, so the boundary condition effect stays small; the cell count scales with domain volume over spacing cubed.
- Mesh refinement: each refinement level halves the cell size, so level n gives size_n = base / 2^n; refine the wake, shock, and separation regions, and adapt where the error indicator is large.
Workflow
- Choose the grid type with grid_type_recommendation from the geometry complexity and the boundary-layer resolution need.
- Estimate u_tau from cf and v_inf, then size the first cell height from the y+ target with first_cell_height_from_cf (or first_cell_height when u_tau is known).
- Count the prism layers with prism_layer_count from the first cell height, boundary-layer thickness, and growth ratio.
- Check cell quality with quality_flags on skewness, orthogonality angle, and aspect ratio; pass boundary_layer_cell=True for near-wall cells so their high aspect ratio is not flagged.
- Size the far-field domain and estimate the cell count with estimate_cell_count; plan refinement levels with refinement_sizes.
- Verify the achieved y plus of the produced first cell with achieved_y_plus before generating the mesh.
Pitfalls
- Sizing the first cell from the wrong y+ band: a y+ near 1 cell paired with a wall-function treatment wastes cells; the y+ target must match the wall treatment.
- Confusing the first cell height with twice it: y+ uses the cell center distance from the wall, not the cell width.
- Growth ratios near 1 explode the prism layer count; ratios above about 1.5 degrade the boundary-layer profile resolution.
- Ignoring quality metrics: skewed or low-orthogonality cells stall the solver even when the near-wall spacing is correct.
- A domain too small contaminates the solution from the far-field boundary; a domain too large wastes cells.
- Treating boundary-layer aspect ratio like general-cell aspect ratio: high aspect ratio is expected in prisms, not in the free stream.
Behavior contract (gate 3)
The grid type, y plus, prism layer, quality, domain sizing, and refinement logic is exercised by the gate 3 contract test: scripts/test_cfd_mesh_generation.py against scripts/cfd_mesh_generation_logic.py (stdlib unittest, offline). Run: python3 scripts/test_cfd_mesh_generation.py
Compliance
- NACA Report 824 is US government work (public domain); summary and reference data only, per standards-map.yaml. The mesh generation methodology here is common CFD practice, not reproduced text.
- compliance: STANDARDS-REF, gated: false.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |