# MS1101 global layer-grid protocol v2

Registered before any v2 meshing or solver run on 27 September 2026. This protocol replaces the failed N=24 construction candidate; it does not relax any gate.

## Question

Can a three-level, whole-domain mesh family with near-wall layers be constructed and solved consistently enough to assess spatial convergence of thrust and torque?

## Fixed inputs and varied parameter

- Source: the accepted medium MS1101 case recorded in `layer-provenance.json`; geometry, domain, operating point, flow properties, rotor/wake refinement levels (5/3/2), mesh-quality controls, layer algorithm, solver, schemes, and boundary conditions remain fixed.
- Base cells per axis: N=28, 32, 44 in the same 0.84 m cubic background domain. Nominal spacing is h=0.84/N, giving unequal refinement ratios that must be handled explicitly in any later GCI calculation.
- Layer construction: `displacementMotionSolver`, eight requested blade layers, expansion ratio 1.2, first height `50 µm × 32/N`, and minimum thickness 10% of first height.
- Each case is generated fresh from the same source. No old mesh, solved fields, or per-case tuning is carried forward. Preserve source/input hashes, meshing logs, and solver provenance.

## Gates fixed before running

Construction passes only if all conditions hold: `checkMesh` reports `Mesh OK.`, no more than 3,000,000 final cells, and at least 95% of requested blade faces receive some layer extrusion. Coverage is face-count based, not area based; also report the requested and achieved layer counts and thicknesses. A failed case remains recorded and is not repaired by relaxing controls or thresholds. If fewer than three candidates pass, stop and register a new protocol before trying another family.

Only if all three construction cases pass, solve all three with the unchanged common solver protocol and matching stopping rules. Check mesh quality, all residual criteria over the predeclared terminal window, thrust and torque plateau/range, and y+ from the final converged field at the same iteration. Keep failed or incomplete solver runs in the evidence set and exclude them from convergence analysis.

GCI is not automatic after three successful solves. First inspect ordered thrust and torque sequences for monotonicity and evidence of an asymptotic range. If they do not support those assumptions, report grid sensitivity and do not calculate or claim Richardson extrapolation/GCI. This work does not establish physical validation against the ENOLA experiment.

## Rationale and references

N=28 is the first new coarser candidate above the failed N=24 mesh; N=32 and N=44 retain previously screened resolutions. Including all three in a fresh registered family keeps the source and controls common while avoiding post-result threshold changes. The unequal h ratios are recorded for proper treatment, not assumed equal.

- [NASA NPARC: Examining Spatial (Grid) Convergence](https://www.grc.nasa.gov/www/wind/valid/tutorial/spatconv.html)
- [Celik et al. (2008), Procedure for Estimation and Reporting of Uncertainty Due to Discretization in CFD Applications](https://doi.org/10.1115/1.2960953)
- [OpenCFD: Layer addition in snappyHexMesh](https://doc.openfoam.com/2212/tools/pre-processing/mesh/generation/snappyhexmesh/layers/)