cfMesh for OpenFOAM Foundation 13

A heavily modified and actively developed fork of cfMesh for OpenFOAM Foundation 13, focused on robust Cartesian/polyhedral meshing and boundary-layer generation for difficult CFD geometry.

This fork is aimed particularly at geometries where conventional boundary-layer extrusion becomes fragile: turbomachinery, propellers, very thin near-wall layers, periodic-sector boundaries, sharp features, multi-patch junctions, and strong refinement transitions.

Upcoming work (to do list):

• Improve deterministic behavior during parallel meshing.
• Port to OpenFOAM Foundation 14.
• Eliminate remaining advanced checkMesh geometry failures.
• Continue improving high-aspect-ratio boundary-layer and multi-patch junction quality.
• Generalize automatic gap and tip-clearance detection.
• Further improve periodic-interface and cross-patch boundary-layer handling.
• Consolidate experimental repair logic into a cleaner production architecture.
• Build automated regression and mesh-quality acceptance testing.
• Improve performance and scalability on large meshes.
• Expand documentation, example cases, and CFD validation.

Current release: v1.0.0

Target platform: OpenFOAM Foundation 13

CFD validation of the v1.0.0 meshes is now in progress.

The original cfMesh was developed by Franjo Juretić / Creative Fields. This fork preserves the original GPL licensing and authorship while adding extensive OpenFOAM Foundation 13 compatibility, robustness, diagnostics, topology repair, and boundary-layer development.

---

v1.0.0 milestone

Release 1 marks the transition from primarily mesher development to CFD validation.

The current code has demonstrated robust meshing on two particularly difficult geometries:

NASA Rotor37

The current Rotor37 regression contains approximately 4.75 million cells with 15 requested boundary layers and a first-layer height of approximately 3 micrometres.

Final ordinary OpenFOAM Foundation 13 checkMesh result:

Metric	Result
Points	5,291,690
Faces	14,786,584
Cells	4,747,717
Hexahedra	4,444,069
Prisms	54,165
Polyhedra	249,483
Negative-volume cells	0
Incorrect face pyramids	0
Maximum non-orthogonality	88.678 degrees
Faces >= 90 degrees	0
Severe non-orthogonal faces (>70 degrees)	881
Maximum skewness	3.992
Ordinary checkMesh	Mesh OK.

Boundary-layer coverage is approximately:

Patch	Coverage
Hub	99.63%
Shroud	89.57%
Periodic 1	99.04%
Periodic 2	99.24%
Blade 1	98.78%
Blade 2	98.99%
Blade 3	99.47%
Blade 4	99.78%

Rotor37 exercises blade/hub/shroud junctions, periodic boundaries, sharp trailing-edge geometry, BL/no-BL transitions, cross-patch BL seams, and very small first-layer heights.

---

NASA Rotor37 regression mesh showing the blade passages, endwalls, periodic-sector boundaries, and near-wall boundary-layer structure.

## Propeller regression The second major regression is a complex marine-propeller geometry with a very thin wall-normal boundary layer.

Metric	Result
Points	8,061,652
Faces	23,804,465
Cells	7,869,273
Hexahedra	7,776,252
Prisms	3,597
Polyhedra	89,424
Requested propeller BL layers	14
Minimum first-layer height	approximately 7.5 micrometres
Nominal first-layer height	7.5-10 micrometres
Propeller BL coverage	96.33%
Negative-volume cells	0
Incorrect face pyramids	0
Maximum aspect ratio	440.46
Maximum non-orthogonality	88.21 degrees
Faces >= 90 degrees	0
Severe non-orthogonal faces (>70 degrees)	36
Maximum skewness	3.847

The very thin first layer is intentional for wall-resolved marine CFD.

Advanced geometry diagnostics

Marine-propeller regression mesh showing the overall mesh and detailed near-wall boundary-layer structure.

The propeller still reports several failures under:

checkMesh -allTopology -allGeometry

Current populations include:

Diagnostic	Count
Face-tet decomposition failures	138,732
Cells with determinant < 0.001	684,625
Concave cells	10,612
Faces with interpolation weight < 0.05	1,378
Faces with volume ratio < 0.01	1,176

These remain active development targets.

They are disclosed deliberately. A successful ordinary mesh-quality result does not by itself prove CFD accuracy or solver robustness.

---

CFD validation status

CFD validation is now underway.

An earlier development mesh from this project successfully advanced an OpenFOAM CFD solution and produced velocity, pressure, turbulence, and yPlus fields.

The exact v1.0.0 Rotor37 and propeller meshes are now entering solver validation.

At this stage the project does not claim validated quantitative:

• propeller thrust or torque;
• cavitation extent;
• compressor pressure ratio;
• compressor efficiency;
• compressor mass flow;
• aerodynamic or hydrodynamic performance accuracy.

Meshing validation and CFD validation are separate milestones.

---

Boundary-layer architecture

The primary development architecture is:

boundaryLayerArchitecture constraintPlanner;

The constraintPlanner path treats boundary-layer creation as a constrained topology problem instead of independently extruding every surface.

Major capabilities include:

• patch-specific layer counts and first-layer heights;
• neighboring-face layer-count compatibility;
• configurable maximum layer-count steps;
• cross-patch BL seam synchronization;
• stable BL provenance;
• BL/no-BL transition handling;
• topology validation;
• transactional repair and rollback;
• proof-gated topology reconstruction;
• OpenFOAM-aligned geometry diagnostics.

The goal is to preserve useful near-wall resolution whenever possible instead of simply deleting difficult layers.

---

Automatic boundary-junction discovery

v1.0.0 includes automatic detection of multi-patch boundary junctions from actual surface topology and resolved boundary-layer intent.

The mesher can identify relationships such as:

• BL / BL seams;
• BL / no-BL termination seams;
• multi-patch junctions;
• BL/BL/neutral junctions;
• BL/neutral/neutral junctions.

This reduces reliance on hard-coded patch names.

Current clearance limitation

Real non-touching clearances, such as a compressor blade tip separated from a shroud, are a different geometric problem.

The current gap system still uses explicit gapPatchPairs and proximity detection.

Automatic opposing-surface clearance detection and BL-envelope negotiation are planned future work.

---

Major v1.0.0 improvements

Release 1 includes extensive work in:

• OpenFOAM Foundation 13 compatibility;
• boundary-layer topology robustness;
• cross-patch layer synchronization;
• periodic face-fan repair;
• multi-patch junction handling;
• stable boundary-layer provenance;
• prospective quality checks;
• transactional topology repair;
• lossless 17-digit mesh serialization;
• large polyhedral topology handling;
• OpenMP race hardening;
• surface projection and feature handling;
• OpenFOAM geometry parity diagnostics.

Several general correctness issues were also fixed, including:

• meshSubset equality logic;
• processorBoundaryPatch type comparison;
• quadricFitting eigenvector normal extraction;
• initializer ordering;
• compiler-warning cleanup.

The current tested libmeshLibrary.so build completes with:

WARNING COUNT: 0

---

Build

This project targets OpenFOAM Foundation 13.

Load OpenFOAM:

source /opt/openfoam13/etc/bashrc

Clone:

git clone https://github.com/Ultravis66/cfMesh-OpenFOAM13.git cfMesh
cd cfMesh

For the v1.0.0 release branch:

git switch release/v1.0.0

Build:

./Allwclean
./Allwmake

---

Running cartesianMesh

The primary supported mesher in this fork is cartesianMesh.

For deterministic development and regression testing:

export OMP_NUM_THREADS=1
cartesianMesh
checkMesh

For deeper geometry diagnostics:

checkMesh -allTopology -allGeometry

Highly anisotropic boundary-layer meshes should not be judged from a single quality metric.

Relevant checks include:

• positive cell volume;
• face-pyramid orientation;
• non-orthogonality;
• skewness;
• interpolation weight;
• volume ratio;
• determinant;
• face-tet quality;
• BL continuity and coverage;
• actual CFD solver behavior.

---

Current development priorities

Following v1.0.0, the main priorities are:

1. CFD validation of the Rotor37 and propeller meshes.
2. High-thread-count/OpenMP regression testing.
3. Reduction of advanced propeller geometry failures without destroying near-wall resolution.
4. Further automatic boundary-relation discovery.
5. Automatic non-touching clearance detection for turbomachinery blade-tip/shroud gaps.
6. Continued reduction of manual patch-role configuration.

---

Project history

This project targets OpenFOAM Foundation 13.

Earlier work involved bringing cfMesh forward through newer OpenFOAM versions, but active mesher development and validation are now focused on Foundation 13.

OpenFOAM 12 is not an actively developed target of this repository.

---

License and original authorship

cfMesh is free software distributed under the GNU General Public License.

The original cfMesh project was developed by Franjo Juretić and Creative Fields.

This repository is an OpenFOAM Foundation 13 port and extensive engineering fork. Original copyright, licensing, and authorship are preserved in accordance with the existing source headers and GPL requirements.

---

Release status

v1.0.0 — mesher validation milestone reached; CFD validation in progress.