[fra76]

Hi!
I've just noticed that checkMesh utility (or whenever mesh.checkMesh(true) is called) returns an error even on a correct mesh.

1.3 release output:

################################################## ######################
Checking geometry...
Boundary openness in x-direction = -1.13243e-14
Boundary openness in y-direction = 1.91513e-15
Boundary openness in z-direction = 1.4988e-15
Boundary closed (OK).
Max cell openness = 5.55112e-17 Max aspect ratio = 2.16706. All cells OK.

Minumum face area = 0.000412266. Maximum face area = 0.242651. Face area magnitudes OK.

Min volume = 4.41069e-06. Max volume = 0.037127. Total volume = 149. Cell volumes OK.

Mesh non-orthogonality Max: 63.0194 average: 19.0074
Non-orthogonality check OK.

Face pyramids OK.

Max skewness = 128.837 percent. Face skewness OK.

Minumum edge length = 0.0241059. Maximum edge length = 0.874393.

All angles in faces are convex or less than 10 degrees concave.

All faces are flat in that the ratio between projected and actual area is > 0.8

Geometry check done.

Number of cells by type:
hexahedra: 0
prisms: 0
wedges: 0
pyramids: 0
tet wedges: 0
tetrahedra: 56824
polyhedra: 0
Number of regions: 1 (OK).
Mesh OK.
################################################## ######################


1.3 develop 01_05_2007 output:

################################################## ######################
Checking geometry...
Boundary openness in x-direction = -1.13243e-14
Boundary openness in y-direction = 1.91513e-15
Boundary openness in z-direction = 1.4988e-15
Boundary closed (OK).
--> FOAM Serious Error :
From function bool primitiveMesh::checkClosedCells(const bool report, labelHashSet*) const
in file meshes/primitiveMesh/primitiveMeshCheck.C at line 346
100 singly connected cells found. Min connections: 2
Minumum face area = 0.000412266. Maximum face area = 0.242651. Face area magnitudes OK.

Min volume = 4.41069e-06. Max volume = 0.037127. Total volume = 149. Cell volumes OK.

Mesh non-orthogonality Max: 63.0194 average: 19.0074
Non-orthogonality check OK.

Face pyramids OK.

Max skewness = 128.837 percent. Face skewness OK.

Minumum edge length = 0.0241059. Maximum edge length = 0.874393.

All angles in faces are convex or less than 10 degrees concave.

All faces are flat in that the ratio between projected and actual area is > 0.8

Geometry check done.

Number of cells by type:
hexahedra: 0
prisms: 0
wedges: 0
pyramids: 0
tet wedges: 0
tetrahedra: 56824
polyhedra: 0
Number of regions: 1 (OK).
Failed 1 mesh checks.
################################################## ######################



The curious thing is that it always find " 100 singly connected cells"...

Bye!
francesco

[lr103476]

I've got exactly the same error message using that development OF version. In OF 1.4 this message dissapears.

Still, I am curious what it means....Btw, Eugene, this is the error which I briefly mentioned.

Frank

[eugene]

I can only guess since I do not have the development version to hand. I have checks in some of my applications which count the number of faces a cell or cell cluster has between itself and the rest of the mesh.

If a cell or cell cluster only shares a single face with the rest of the domain and there are no fixed pressure boundaries or cells in its extent, then the pressure in that cell or cell-cluster will tend to float and eventually cause the solution to diverge. To understand why this happens, consider a tet with all but one of its faces on a wall boundary. Now calculate the pressure gradient in the cell. Notice that the pressure gradient is essentially an extrapolation of the single face and hence ill defined. Similarly, if you consider that the velocity in the cell is a function of the single face pressure gradient, you can see how the system can easily become unstable because according to continuity, the cell velocity should be zero at convergence. I am not sure how Fluent/STAR/CFX handle this issue, because they seem to run without any problems on such cells. If I had to hazard a guess, I would say they either apply gradient limiters to the problem faces or use large momentum sinks to force the cell velocities to zero.

This kind of situation is particularly prevalent on tet meshes. If anyone knows a good way to address it please let me know.

[eugene]

Oh I forgot to mention, this is a serious problem. Your solution is likely to blow up if such errors are present.

[lr103476]

Strange, I have no problems on these meshes. My Hex meshes were perfectly fine with the original release of OF 1.3 and with the original 1.4 release.

Only with the development version, I get these errors using checkMesh, but no problems with convergence at all....

Frank

[eugene]

Then my guess as to the message's meaning was probably wrong, since it would be very difficult to get the kind of errors I described on a block-structured hex mesh.