# LW33 R14 — stricter mesh-quality control

## Result

R14 removes the failed low-interpolation-weight check and reduces poor-determinant cells from 42 (R13) to 18. Minimum face interpolation weight is 0.050716008, above the 0.05 check threshold. It still fails two detailed checks: poor cell determinants and concave cells (20,708). It is **not an accepted performance mesh**.

The fixed-hull flow run saved results through **0.009897634 s**. Maximum saved air/water velocities: **29.03 / 16.45 m/s**. Stop reason: **450 second wall limit**. Solver exit: 0. All final velocity values finite: True.

Air selection: water fraction <0.01; water selection: water fraction >0.99. The website's maximum-velocity history spans all fluid cells. Neither metric is a boat-speed prediction. Full force and velocity histories are retained, without interpreting startup forces as converged resistance.

## Controlled change and resolution

Only snappyHexMesh minFaceWeight changes from 0.02 (R13) to 0.05. Hull surface, three-layer proportions, layer-termination buffer and other mesh controls are retained. Fluid cells: 330610. Wetted area with at least one layer: 81.89%; with at least three layers: 35.21%. R13 had 82.42% and 35.46%, respectively. Coverage is area-weighted using face-centroid selection below the initial waterline; it is not proof of adequate y+ or wall resolution.

The 18 residual poor-determinant cell locations are recorded in audit/r14-residual-defects.json. Seventeen parametric cell centers lie below the initial Z=0 waterline; one is above it. This locates the next repair targets but does not establish that these cells cause any observed flow instability.

OpenFOAM v2412, serial interFoam, prescribed towing speed 30 kn, fixed hull. Requested endpoint 0.01 s, 450-second wall budget, guard above 100 m/s global speed or below 5-microsecond time step. Adaptive stepping and checkpoint response mean actual saved endpoints differ from requested limits. The bounded run remains much shorter than one hull transit (~0.64 s).

## Boat configuration and validation status

The new user reference is **4,100 kg loaded; one Volvo Penta D6-340; reported Volvo Penta H6; H8 only an unverified candidate; CG 2,750 mm with datum and direction unconfirmed**. Drive/ratio and vertical CG remain missing. This reference is published separately from historical CFD evidence.

R14 retains the old 3,500 kg reference, original initial waterline and inferred LCG 2.9405231959611386 m forward of the aft-keel datum to isolate the mesh change. The new loading and CG have NOT been applied to this flow case. Do not infer the new CG datum from the historical one. No resolved engine or propeller is simulated.

Step 2 remains open. Next: repair residual mesh defects, check wall resolution, extend the accepted candidate, and demonstrate temporal and mesh convergence. Then establish equilibrium using confirmed loaded CG and compare with actual sea-trial measurements. No validated resistance, attainable speed, fuel use, running trim or wave response is available.

## Checkpoint reproduction

The archive contains complete R14 mesh inputs and generated mesh; initialized flow fields and the final two saved field directories; complete post-processing/logs; and audit and preparation scripts. Each mesh is stored once. After extracting, copy `meshes/r14-strict-weight/constant/polyMesh` to `cases/LW33-r14-strict-weight/constant/polyMesh` if absent, preserving the original evidence. Work in a separate copy to rerun with OpenFOAM v2412. The mesh/run scripts retain historical-source dependencies as provenance; the archived initialized case can be used directly with OpenFOAM. Python, NumPy, VTK and OpenFOAM are not bundled. A successful process exit for checkMesh is not a clean-mesh verdict: read its textual failures.