[robincfd]

Hi everyone

My name is Robin, and I’m a beginner in CFD, especially when it comes to meshing with snappyHexMesh.

I’m currently working on a study of pressure losses in a suction orifice, and I’m trying to build a clean, consistent mesh of the fluid domain before running any CFD calculations.

At this stage, I’m not yet trying to achieve y+ ≈ 1.
I’m simply trying to generate boundary layers (inflations) along all Wall surfaces, and make sure they project correctly near the Inlet and Outlet intersections.
Once this works properly, I’ll focus on getting a mesh that respects wall-resolved turbulence requirements and mesh quality controls, so any advice for that next step is welcome too!

Current situation

I can generate layers on the Wall surfaces.

However, they don’t project correctly near the Inlet and Outlet boundaries — they stop or collapse at the sharp connection between Wall and Inlet/Outlet (see screenshots).

The only way I can currently get layers to appear at all is by setting:

nLayerIter = 1;
nRelaxIter = 1;


which disables the meshQualityControls.
With normal values, no many layers are generated.

Geometry setup

The fluid domain is defined by three STL surfaces:

Wall.stl
Inlet.stl
Outlet.stl

Each surface is defined separately as a triSurfaceMesh in the geometry block.

What I’m trying to understand

How to make the boundary layers project smoothly up to the Inlet and Outlet faces, without being deleted near the junctions.

Later, how to re-enable proper mesh quality controls while keeping layer generation stable.

And finally, once the layers behave correctly, how to adjust them to reach y+ ≈ 1.

Attached

My current snappyHexMeshDict

A few screenshots of the Wall–Inlet and Wall–Outlet transitions where the layers fail to project

Any help or advice on how to correctly set up snappy for this kind of geometry (closed pipe domain built from separate Wall/Inlet/Outlet STL surfaces) would be greatly appreciated.

Thanks a lot in advance for your time and guidance,
Robin

snappyHexMeshDict :

/*--------------------------------*- C++ -*----------------------------------*\
| =========                 |                                                 |
| \\      /  F ield         | OpenFOAM: The Open Source CFD Toolbox           |
|  \\    /   O peration     | Version:  v2012                                 |
|   \\  /    A nd           | Website:  www.openfoam.com                      |
|    \\/     M anipulation  |                                                 |
\*---------------------------------------------------------------------------*/
FoamFile
{
    version     2.0;
    format      ascii;
    class       dictionary;
    object      topoSetDict;
}

// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

castellatedMesh false;
snap            false;
addLayers       true;

mergeTolerance 1e-6;

geometry
{
    Inlet
    {
        type triSurfaceMesh;
        file "Inlet.stl";   // dans constant/triSurface
    }
    Outlet
    {
        type triSurfaceMesh;
        file "Outlet.stl";   // dans constant/triSurface
    }
    Wall
    {
        type triSurfaceMesh;
        file "Wall.stl";   // dans constant/triSurface
    }
}

castellatedMeshControls
{
    // bornes pour contenir la taille tout en autorisant un raffinement poussé
    maxLocalCells        2000000;
    maxGlobalCells       10000000;
    minRefinementCells   10;
    nCellsBetweenLevels  1;          // transitions serrées → suit mieux les détails
    maxLoadUnbalance 	 0.1;

    // Utilise l’eMesh pour forcer les arêtes
    features
    (
        { file "Inlet.eMesh"; level 3; }
	{ file "Outlet.eMesh"; level 3; }
	{ file "Wall.eMesh"; level 3; }  
    );

    // angle à partir duquel une arête du STL est “résolue” par raffinement
    resolveFeatureAngle  1;            // 30–40 pour bien accrocher les arêtes

    // Raffinement surfacique par région STL
    refinementSurfaces
    {
        Inlet { level (0 2);}
        Outlet { level (0 2);}
        Wall { level (0 2);}
    }
    refinementRegions
    {
	Wall {mode distance; levels ( (0.003 3) );}
    }

    insidePoint (-0.03 0 0.03);
    allowFreeStandingZoneFaces true;
}

snapControls
{
    nSmoothPatch   		5;
    nSmoothInternal 		$nSmoothPatch;
    tolerance      		3;     // plus strict → colle mieux (1.5–2.0)
    nSolveIter     		50;
    nRelaxIter     		6;

    nFeatureSnapIter 		10;
    nFaceSplitInterval 		5;

    implicitFeatureSnap    	false;
    explicitFeatureSnap    	true;     
    multiRegionFeatureSnap 	false;

}

addLayersControls
{
    relativeSizes 		false;
    expansionRatio		1.1;
    firstLayerThickness		0.00017;
    minThickness		0.00003;

    layers
    {
        Wall   			{ nSurfaceLayers 5; projectToPatch (Wall); }
        Inlet  			{ nSurfaceLayers 0; }
        Outlet 			{ nSurfaceLayers 0; }
    }

    nGrow                 	0;
    featureAngle        	179;
    slipFeatureAngle    	179;

    nBufferCellsNoExtrude	0;
    nLayerIter			1; //50
    nRelaxIter			1; //30
    // nRelaxedIter		30;
    additionalReparting		true;

    maxFaceThicknessRatio     	1;
    mergePatchFacesAngle	179;
    nSmoothSurfaceNormals 	10;
    nSmoothThickness     	10;
    minMedianAxisAngle        	179;
    maxThicknessToMedialRatio 	1;

    nSmoothNormals        	15;

    mergePatchFaces		true;
    mergeAcrossPatches		false;

    // projectToSurface		true;

    correctLayers		true;
}

meshQualityControls
{
    maxNonOrtho          	180;
    maxBoundarySkewness  	20;
    maxInternalSkewness  	4;

    maxConcave           	90;
    minVol               	-1e30;
    minTetQuality        	-1e30;
    // minArea              	-1;
    minTwist             	0.02;

    minDeterminant       	0.001;

    minFaceWeight        	0.05;

    minVolRatio          	0.01;
    // minTriangleTwist     	1e-15;

    nSmoothScale         	4;
    errorReduction       	0.75;
}

debug 0;

// (Facultatif) encore moins de sorties
writeFlags ();

// ************************************************************************* //

[AtoHM]

It seems you messed quite alot with the settings in the addLayers part. Here are the documented keywords https://www.openfoam.com/documentati...sh-layers.html and I fail to find some stuff you specified there and some settings seem to be off of the "recommended" settings, e.g. slipFeatureAngle should be half the featureAngle, whereas it is the same in your case. I recommend you set the default stuff for all the weird angles and iterations etc.

The reason why it collapses the layers toward the inlet/outlet is, that you specified 0 layers for these patches. It freezes the cells on the surface and dis-allows to move according the the growth of the adjacent walls. Completely delete these entries with 0 layers from the dict, and it should work.

[robincfd]

Thanks for your reply!

Originally Posted by AtoHM

It seems you messed quite alot with the settings in the addLayers part. Here are the documented keywords https://www.openfoam.com/documentati...sh-layers.html and I fail to find some stuff you specified there and some settings seem to be off of the "recommended" settings, e.g. slipFeatureAngle should be half the featureAngle, whereas it is the same in your case. I recommend you set the default stuff for all the weird angles and iterations etc.

The reason why it collapses the layers toward the inlet/outlet is, that you specified 0 layers for these patches. It freezes the cells on the surface and dis-allows to move according the the growth of the adjacent walls. Completely delete these entries with 0 layers from the dict, and it should work.

I managed to fix the inflation issue that was projected onto the inlet and outlet — not sure if that was exactly the root cause, but I rebuilt my geometry from scratch following the WolfDynamics tutorials (here’s the link in case it helps someone else).

http://https://www.youtube.com/watch...z8YxC&index=34

Now, I’m struggling to generate inflations on all the walls due to the geometric complexity. I’ve been tweaking the different parameters, but I can’t seem to get anything close to uniform layers.

If anyone has advice on which key parameters are most relevant to improve this, I’d really appreciate it.

Thanks in advance,
Robin

[AtoHM]

... and welcome to the world of pain that is adding layers with sHM


Jokes aside, snappyHexMesh does not shine for adding layers to complex geometries. Settings that work for one case might worsen it in another. You will find hundreds of threads here in the OF-Meshing subforum talking about strategies to resolve it ... and many without a resolution.


In order to give more hints, we would need the settings you currently use.

[Alczem]

Hey,


It is a pain, but it is getting better! I am sharing tkeskita's work on this, hope they won't mind


https://github.com/tkeskita/snappyLa...nappy.template


These settings work quite well for layer coverage. I would just increase the featureAngle up to 180 (or even 350), I have found it worked better for most of my geometries. The other setting that really improves coverage is nOuterIter.