NACA 0012 Airfoil CFD Validation Study Against XFOIL with Mesh Convergence

NACA 0012 Airfoil CFD Validation Study Against XFOIL with Mesh Convergence is an individual M.Tech research study that answers the question every CFD report should be forced to answer: how do you know your mesh is good enough, and how do you know your solver is right? It does so on the best-documented external-aerodynamics case in the open literature — the NACA 0012 aerofoil — using entirely free, scriptable tools.

Two-dimensional RANS solutions are computed in OpenFOAM v12 with the Spalart–Allmaras and k-ω SST turbulence models at the NASA Turbulence Modeling Resource's canonical conditions — Mach 0.15, Reynolds number 6 million per chord — for angles of attack from −5° to +15°. A family of nested structured C-grids is generated in Gmsh so that a formal Grid Convergence Index (GCI) analysis per Celik et al. runs on the integrated forces, with the observed order of accuracy reported rather than assumed. Results are validated on three levels: against NASA TMR reference solutions and the Ladson experimental dataset (NASA TM 4074), against a second, completely different method — XFOIL 6.99, the panel-plus-integral-boundary-layer code — and against each other through the GCI bands.

The study is deliberately scoped as a validation protocol demonstration: it produces not just lift and drag curves but the evidence structure — convergence tables, y+ distributions, error bands and a written RANS-vs-XFOIL comparison — that a UAV wing-design office or a subsequent dissertation can inherit as a template.

Aerofoil CFD results are routinely reported without quantified discretisation error, and aerofoil design in student and small-industry settings relies on XFOIL without a quantified understanding of where its assumptions fail. Before CFD can responsibly support a UAV wing or turbine-blade design, the solver setup must be shown to converge and to agree with trusted data — and the design tool must have known limits.

There is a need for a documented validation study that (i) establishes grid convergence for NACA 0012 RANS solutions at the canonical M = 0.15, Re = 6 × 10⁶ conditions using a nested mesh family and formal GCI analysis, (ii) validates lift, drag and pressure distributions against NASA TMR reference solutions and the Ladson experimental dataset, and (iii) quantifies the agreement and divergence between RANS CFD and XFOIL across the attached-flow and pre-stall range — using only open software so that the entire protocol is reproducible by any student or small design team without licence costs.

Abstract

This dissertation performs a systematic CFD validation of the NACA 0012 aerofoil at M = 0.15 and Re = 6 × 10⁶ in OpenFOAM v12 with SA and k-ω SST models, establishes grid convergence through a nested mesh family with Grid Convergence Index analysis, and validates integrated forces and pressure distributions against NASA TMR reference solutions, the Ladson experimental data, and XFOIL 6.99.

Introduction

CFD is now the primary aerofoil design tool in Indian UAV and wind-energy programmes, but its credibility rests on validation discipline that undergraduate teaching rarely demonstrates. The NACA 0012 at the NASA TMR's canonical conditions is the community's benchmark precisely because high-quality experimental data (Ladson, NASA TM 4074), reference RANS solutions and downloadable mesh families exist — making it the honest test of a solver setup before it is applied to a new aerofoil.

Literature Gap and Motivation

• Published aerofoil CFD studies frequently report single-mesh results with no discretisation-error estimate; the ASME editorial by Celik et al. exists precisely because this practice produced unreliable literature.
• XFOIL remains the workhorse of conceptual aerofoil design, yet its stated limits — attached flow, integral boundary-layer closure, weak performance near stall — are rarely quantified against modern RANS by students who use it daily.
• Indian UAV-sector teams need a documented, zero-licence-cost validation workflow on open tools, since commercial CFD seats are scarce in small organisations.
• Gap addressed: a complete, reproducible mesh-convergence-plus-validation protocol on open software, benchmarked to NASA reference data, packaged as a dissertation that future students can extend to other aerofoils.

Proposed Work

• Case definition: NACA 0012, M = 0.15, Re = 6 × 10⁶ (chord-based), fully turbulent assumption; α from −5° to +15° in 2.5° steps; far field at ≈500 chords.
• Meshing: four nested structured C-grids with refinement ratio near √2, generated in Gmsh with wall spacing set for y+ ≈ 1 (wall-resolved), and documented grid-quality metrics.
• Solver studies: OpenFOAM steady incompressible RANS with SA and k-ω SST; second-order schemes; iterative convergence judged on residuals and force stability; each angle run on every grid.
• Convergence analysis: Richardson extrapolation and GCI per Celik et al. on CL and CD, observed order of accuracy reported, fine-grid GCI bands carried into all validation plots.
• Validation: Cp distributions and integrated CL/CD compared against NASA TMR reference data, Ladson experiments, and XFOIL 6.99 runs at matching conditions; divergence near stall analysed physically (separation, turbulence-model limits).

Feasibility

• Technical: OpenFOAM, XFOIL, Gmsh, ParaView and Python are free and scriptable; the NASA TMR publishes grids and data openly; every step is automatable so the full matrix (angles × grids × models) is feasible on a workstation.
• Economic: zero licence cost — significant for a study whose contribution is precisely that small Indian teams can reproduce it.
• Operational: the 24-week span maps onto VTU's Phase-1 (literature, formulation, XFOIL baseline) and Phase-2 (mesh campaign, GCI, validation, dissertation), with the mesh family built early so solver runs proceed in parallel.

In scope

• 2D steady incompressible RANS of the NACA 0012 at the NASA TMR canonical conditions, fully turbulent, wall-resolved.
• Two turbulence models (SA, k-ω SST), four nested grids, fourteen angles of attack.
• GCI analysis on integrated forces; validation against TMR data, Ladson experiments and XFOIL.
• Workflow packaging (scripts, case files, notebooks) for reuse.

Out of scope

• 3D wing effects, compressibility corrections beyond Mach 0.15, and transition modelling (γ-Reθ) — the fully turbulent comparison is the study's declared frame.
• Unsteady RANS, DES/LES and deep-stall post-stall behaviour beyond α = 15°.
• Aerofoil redesign or optimisation — this study validates the toolchain, it does not design with it.
• Wind-tunnel experimentation; all experimental data is archival.

The dissertation follows a validation-hierarchy research design over 24 weeks, split across VTU M.Tech Project Work Phase-1 and Phase-2.

Rigour controls: iterative convergence is declared only when scaled residuals fall below 10⁻⁶ and integrated forces drift less than 0.1% over the last 500 iterations; every reported force carries its fine-grid GCI band; all post-processing is scripted in Python so any plot can be regenerated from the archived case directory; a validation log records each comparison's discrepancy against the acceptance bands agreed with the guide in Phase-1.

The study is organised as a validation pipeline — formulation, meshing, solution, convergence, validation — in which each stage only accepts inputs that passed the previous stage's checks. The flowchart below is the dissertation's methodology diagram.

flowchart TD
  FORM{"Case formulation: M 0.15, Re 6e6, alpha -5 to 15 deg"} --> MESH["Gmsh C-grid family: coarse / medium / fine / very fine, r ~ sqrt(2)"]
  MESH --> Q{"Grid quality: orthogonality, growth, y+ ~ 1"}
  Q -->|fail| MESH
  Q -->|pass| SOL["OpenFOAM steady RANS: SA and k-omega SST"]
  XF["XFOIL 6.99 baseline at matching conditions"] --> CMP
  SOL --> C{"Converged? residuals < 1e-6 and forces stable < 0.1 percent"}
  C -->|no| SOL
  C -->|yes| ARC[("Solution archive per case")]
  ARC --> GCI["Richardson extrapolation + GCI per Celik et al. on CL and CD"]
  GCI --> CMP{"Three-way validation"}
  CMP -->|"CL, CD, Cp"| TMR["NASA TMR reference solutions"]
  CMP --> EXP["Ladson experiments, NASA TM 4074"]
  CMP --> BAND["Fine-grid GCI bands on every plot"]
  TMR --> DIS{"Discrepancy analysis: attached flow vs pre-stall"}
  EXP --> DIS
  BAND --> DIS
  DIS --> DIS2["Dissertation: results, error bands, protocol packaging"]

Prose description. Two feedback loops carry the study's rigour: the grid-quality loop rejects meshes that fail orthogonality or y+ targets before any solver time is spent, and the convergence loop refuses to archive a solution whose forces are still drifting. The three-way validation at the right of the pipeline is the dissertation's core contribution — RANS against reference CFD establishes solver correctness, RANS against experiment establishes physical fidelity, and XFOIL against both quantifies the conceptual-design tool's envelope, each with GCI-derived error bands so every claim about agreement is a measurable statement rather than an eyeball judgement.

• Transition modelling: add the γ-Reθ transition model and map the attached-flow region where it changes CD materially, sharpening the XFOIL comparison.
• Second aerofoil: repeat the protocol on a cambered section (NASA SC(2)-0714 or an NACA 4-digit cambered family) to test protocol transferability.
• Reynolds-number sweep: extend to Re 10⁵–10⁶, the small-UAV regime, where XFOIL's boundary-layer assumptions and fully turbulent RANS diverge most.
• Uncertainty quantification: propagate mesh and model-form uncertainty into an interval-based aerofoil-design margin, connecting to research-methodology coursework.
• Publication: the validation protocol itself, benchmarked at zero licence cost, is a candidate for a national-conference paper on accessible CFD practice.