[JensD]

I started using cfMesh only recently, but based on the tutorials I'm already a big fan. Using the user guide I was able to mesh a few geometries and can control the sizing of the cells, however I'm stuck on the boundary layer addition.

The mesher seems to only split up the last cell at the boundary into the number of boundary layers I specified in the meshDict. This leaves me with a big jump in size between the last boundary layer and the next cell towards the center. I believe this will badly influence my simulations, is there any way to make the transition between the boundary layer and the rest of the grid smoother? Am I missing a keyword in the dictionary?

/*--------------------------------*- C++ -*----------------------------------*\
| =========                 |                                                |
| \\      /  F ield         | cfMesh: A library for mesh generation          | 
|  \\    /   O peration     |                                                |
|   \\  /    A nd           | Author: Franjo Juretic                         | 
|    \\/     M anipulation  | E-mail: franjo.juretic@c-fields.com            |
\*---------------------------------------------------------------------------*/

FoamFile
{
    version   2.0;
    format    ascii;
    class     dictionary;
    location  "system";
    object    meshDict;
}

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

maxCellSize				0.0006;		// (MANDATORY) Maximum cell size in the mesh [m]
//boundaryCellSize			0.0003;		// (optional) size of the cells at the boundary [m]
//boundaryCellSizeRefinementThickness	0.0003;		// (optional) distance from the boundary at which boundary cell size shall be used

surfaceFile				"constant/triSurface/cylinder.stl";

boundaryLayers
{
    nLayers				3;		// (optional) global number of layers 

    thicknessRatio 			1.2;		// (optional) thickness ratio

    //maxFirstLayerThickness 		0.5;		// (optional) max thickness of the first layer [m]

    patchBoundaryLayers					// local settings for individual patches
    {
        wall
        {
            nLayers			20;
            thicknessRatio		1.1;
	    //maxFirstLayerThickness 	0.5; 		// [m]
            //allowDiscontinuity 	0;
        }
    }

    untangleLayers 			1;		// (optional) deactivate untangling of boundary layers - activated by default

    optimiseLayer 			1;		// (optional) activates smoothing of boundary layers - deactivated by default

    optimisationParameters
    {
        nSmoothNormals 			5;		// (optional) number of iterations in the procedure for reducing normal variation
        relThicknessTol 		0.1; 		// (optional) maximum allowed thickness variation of thickness between two neigbouring points, divided by the distance between the points
        featureSizeFactor 		0.4;		// (optional) ratio between the maximum layer thickness and the estimated feature size
        reCalculateNormals 		1; 		// (optional) activate 1 or deactivate 0 calculation of normal
        maxNumIterations 		5;		// (optional) maximum number of iterations of the whole procedure
    }
}

keepCellsIntersectingBoundary		1;		// (optional) keep template cells intersecting boundary (1) - use only internal cells (0) 

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

[franjo_j]

Hi,

Your setup is Ok, and it should not cause problems if the gradient is not significant in the transition zone. My run the simuation and find out if it causes any problems.

Are you interested in doing an experiment regarding this? I can quickly re-organise the procedure that might improve this, but at the expense of longer meshing time. We can open a new branch in the repository and evaluate various options.

[JensD]

Hi Franjo, thanks for your reply!

The temperature gradient is quite large (few 10 degrees over a cm approximately). When running the simulation, pressure starts to fluctuate and then explodes. I had no problems with this with other meshes.
EDIT: I've found that my turbulent kinetic energy starts increasing prior to pressure difficulties. I'm doing LES with the dynamic one equation eddy viscosity model as implemented in OF 2.2.x.

What kind of experiments are you refering to? Right now, I'm running simulations on a simple cylinder, using each time a different mesh obtained with different (automated) meshers.

I don't mind increasing the meshing time, as it is still very fast compared to manual meshing and its performance in terms of boundary layer addition is better than automated meshers like snappyHexMesh (which I used most of the time)

[Fredvol]

Originally Posted by JensD

I started using cfMesh only recently, but based on the tutorials I'm already a big fan. Using the user guide I was able to mesh a few geometries and can control the sizing of the cells, however I'm stuck on the boundary layer addition.

The mesher seems to only split up the last cell at the boundary into the number of boundary layers I specified in the meshDict. This leaves me with a big jump in size between the last boundary layer and the next cell towards the center. I believe this will badly influence my simulations, is there any way to make the transition between the boundary layer and the rest of the grid smoother? Am I missing a keyword in the dictionary?

Hello,
I'am experiencing the same problem you had.
i.e: a big gap in thichness between the last boudary cell the the next cell.
Did you solve it ? Do I also miss a dictionary keyword?


Thanks,
Fred

[JensD]

Hi Fred

I did not solve the issue, I switched from automated meshers to Pointwise. It takes a bit more time, but gives you much more control + can also be used to script the meshing procedure.

Sorry!

[Fredvol]

Originally Posted by JensD

Hi Fred

I did not solve the issue, I switched from automated meshers to Pointwise. It takes a bit more time, but gives you much more control + can also be used to script the meshing procedure.

Sorry!

Ok thanks for the prompt answer.

[CFDavatar]

I am facing the same issue,

Does anyone solved it?

Kind regards,

Marcelo