High aspect ratio

oort · September 14, 2011, 12:58

Hello.
When we have a big domain and want to do a first estimative using a rough mesh there is any problem using a mesh that gives a "fail" in high aspect ratio in checkmesh?
I have a script to create the mesh and with my usual parameters it gives no "fails" but I had to decrease the number of cells in each block to be computational affordable in this first "test"...
Thanks

// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //
Create time

Create polyMesh for time = 0

Time = 0

Mesh stats
points: 353073
faces: 990200
internal faces: 923800
cells: 319000
boundary patches: 7
point zones: 0
face zones: 0
cell zones: 0

Overall number of cells of each type:
hexahedra: 319000
prisms: 0
wedges: 0
pyramids: 0
tet wedges: 0
tetrahedra: 0
polyhedra: 0

Checking topology...
Boundary definition OK.
Point usage OK.
Upper triangular ordering OK.
Face vertices OK.
Number of regions: 1 (OK).

Checking patch topology for multiply connected surfaces ...
Patch Faces Points Surface topology
inlet 440 483 ok (non-closed singly connected)
outlet 440 483 ok (non-closed singly connected)
topo 15180 15893 ok (non-closed singly connected)
base 15400 16353 ok (non-closed singly connected)
filamentos 5940 6440 ok (non-closed singly connected)
face_proxima 14500 15351 ok (non-closed singly connected)
face_afastada 14500 15351 ok (non-closed singly connected)

Checking geometry...
Overall domain bounding box (0 0 0) (0.1989 0.03 0.0015)
Mesh (non-empty, non-wedge) directions (1 1 1)
Mesh (non-empty) directions (1 1 1)
Boundary openness (4.10234e-19 -3.21156e-20 5.99409e-18) OK.
***High aspect ratio cells found, Max aspect ratio: 6393.14, number of cells 3680
<<Writing 3680 cells with high aspect ratio to set highAspectRatioCells
Minumum face area = 6.73115e-12. Maximum face area = 1.55208e-05. Face area magnitudes OK.
Min volume = 1.13096e-16. Max volume = 1.54538e-09. Total volume = 8.83461e-06. Cell volumes OK.
Mesh non-orthogonality Max: 56.9438 average: 10.4423
Non-orthogonality check OK.
Face pyramids OK.
Max skewness = 3.04746 OK.

Failed 1 mesh checks.

End

romant · September 15, 2011, 03:50

the results can be unreliable for very aspect ratios, it depends on the turbulence models that you use and if the cells are elongated in the direction of flow or perpendicular to the flow

oort · September 15, 2011, 08:03

Thank you for your answer.
The flow is laminar, and high aspect ratio cell are mainly in flow direction (i think).
I want simulate the flow in a channel filled with transversal cylinders.
3D simulations over a "cylinder module" with cyclic and symmetry conditions were done before.
Also 2D simulations over full length of the channel.
However 3D simulations gives a strange pattern in streamlines. 3D simulations were performed with a solver similar to channelFoam. 2D simulations with icoFoam.
So now i would like to simulate the flow in a 3D geometry in the entire channel using icoFoam to see if the "strage" streamlines appears or the problem is the solver...

2D:

3D with cyclic/symmetry:

3D full channel:

Well. Inserting some more cells and reducing grading the mesh are OK, but as increased from 319000 to 522000 cells...

Checking geometry...
Overall domain bounding box (0 0 0) (0.1989 0.03 0.0015)
Mesh (non-empty, non-wedge) directions (1 1 1)
Mesh (non-empty) directions (1 1 1)
Boundary openness (-4.25725e-20 -3.21156e-20 -9.28017e-18) OK.
Max cell openness = 2.19302e-16 OK.
Max aspect ratio = 312.091 OK.
Minumum face area = 1.73529e-10. Maximum face area = 1.14583e-05. Face area magnitudes OK.
Min volume = 4.36933e-15. Max volume = 1.14089e-09. Total volume = 8.83461e-06. Cell volumes OK.
Mesh non-orthogonality Max: 56.9438 average: 10.4105
Non-orthogonality check OK.
Face pyramids OK.
Max skewness = 1.12258 OK.

Mesh OK.

romant · September 15, 2011, 08:34

from your pictures it is hard to see in which direction your flow has and if the cells are high aspect in flow direction.

your maximum aspect ration at the moment is 312, you can go higher i think, but make sure the stretching is in flow direction and not in any other direction. (OF will complain at 1000)

since you are talking about a first test, decrease cells in all directions, then the stretching does not change.

when making the mesh with blockMesh, watch out in which direction you specify your cells, it only be xyz if your face definition was xyz, so the first direction on your block becomes x, the second y and the last z. you can see this as a coordinate system specific to every block that you create.

Chris Lucas · September 16, 2011, 02:59

Hi,

the aspect ratio shouldn't be the problem, checkMesh was written when OpenFOAM wasn't as stable as it's now. OpenFOAM still works when checkMesh return an error.

However, the sudden change in cell size in your grid (figure: 3D full channel) is the real problem I see. The cells size should increase smoothly (e.g. with an ratio of 1.3)

Regards,
Christian

oort · September 16, 2011, 14:06

Hi Chris,
Thanks for your reply.
Well i can put some grading, but that will create very big cells in the beginning of entrance channel, but probably is the best.
Anyway my 2D results even with this big drop in cell size are good (probably not so big because in 2D I have more cells in the entrance channel).

September 14, 2011, 12:58	High aspect ratio	#1
oort New Member alex Join Date: Jun 2009 Posts: 17 Rep Power: 16	Hello. When we have a big domain and want to do a first estimative using a rough mesh there is any problem using a mesh that gives a "fail" in high aspect ratio in checkmesh? I have a script to create the mesh and with my usual parameters it gives no "fails" but I had to decrease the number of cells in each block to be computational affordable in this first "test"... Thanks // * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * // Create time Create polyMesh for time = 0 Time = 0 Mesh stats points: 353073 faces: 990200 internal faces: 923800 cells: 319000 boundary patches: 7 point zones: 0 face zones: 0 cell zones: 0 Overall number of cells of each type: hexahedra: 319000 prisms: 0 wedges: 0 pyramids: 0 tet wedges: 0 tetrahedra: 0 polyhedra: 0 Checking topology... Boundary definition OK. Point usage OK. Upper triangular ordering OK. Face vertices OK. Number of regions: 1 (OK). Checking patch topology for multiply connected surfaces ... Patch Faces Points Surface topology inlet 440 483 ok (non-closed singly connected) outlet 440 483 ok (non-closed singly connected) topo 15180 15893 ok (non-closed singly connected) base 15400 16353 ok (non-closed singly connected) filamentos 5940 6440 ok (non-closed singly connected) face_proxima 14500 15351 ok (non-closed singly connected) face_afastada 14500 15351 ok (non-closed singly connected) Checking geometry... Overall domain bounding box (0 0 0) (0.1989 0.03 0.0015) Mesh (non-empty, non-wedge) directions (1 1 1) Mesh (non-empty) directions (1 1 1) Boundary openness (4.10234e-19 -3.21156e-20 5.99409e-18) OK. ***High aspect ratio cells found, Max aspect ratio: 6393.14, number of cells 3680 <<Writing 3680 cells with high aspect ratio to set highAspectRatioCells Minumum face area = 6.73115e-12. Maximum face area = 1.55208e-05. Face area magnitudes OK. Min volume = 1.13096e-16. Max volume = 1.54538e-09. Total volume = 8.83461e-06. Cell volumes OK. Mesh non-orthogonality Max: 56.9438 average: 10.4423 Non-orthogonality check OK. Face pyramids OK. Max skewness = 3.04746 OK. Failed 1 mesh checks. End

September 15, 2011, 03:50		#2
romant Senior Member Roman Thiele Join Date: Aug 2009 Location: Eindhoven, NL Posts: 374 Rep Power: 20	the results can be unreliable for very aspect ratios, it depends on the turbulence models that you use and if the cells are elongated in the direction of flow or perpendicular to the flow mm.abdollahzadeh and saeed_hb like this. __________________ ~roman

September 15, 2011, 08:34		#4
romant Senior Member Roman Thiele Join Date: Aug 2009 Location: Eindhoven, NL Posts: 374 Rep Power: 20	from your pictures it is hard to see in which direction your flow has and if the cells are high aspect in flow direction. your maximum aspect ration at the moment is 312, you can go higher i think, but make sure the stretching is in flow direction and not in any other direction. (OF will complain at 1000) since you are talking about a first test, decrease cells in all directions, then the stretching does not change. when making the mesh with blockMesh, watch out in which direction you specify your cells, it only be xyz if your face definition was xyz, so the first direction on your block becomes x, the second y and the last z. you can see this as a coordinate system specific to every block that you create. mm.abdollahzadeh and amolrajan like this. __________________ ~roman

September 16, 2011, 02:59		#5
Chris Lucas Senior Member Christian Lucas Join Date: Aug 2009 Location: Braunschweig, Germany Posts: 202 Rep Power: 17	Hi, the aspect ratio shouldn't be the problem, checkMesh was written when OpenFOAM wasn't as stable as it's now. OpenFOAM still works when checkMesh return an error. However, the sudden change in cell size in your grid (figure: 3D full channel) is the real problem I see. The cells size should increase smoothly (e.g. with an ratio of 1.3) Regards, Christian mm.abdollahzadeh likes this.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
pressure eq. "converges" after few time steps	maddalena	OpenFOAM Running, Solving & CFD	69	July 21, 2011 07:42
High Aspect Ratio elements	Flavio	CFX	2	November 24, 2006 12:01
mesh for a high aspect ratio geometry	matthias	CFX	3	October 20, 2006 02:55
Multicomponent fluid	Andrea	CFX	2	October 11, 2004 05:12
Aspect Ratio at high turbulent and swirled flows	zago	FLUENT	0	May 19, 2004 02:37

September 15, 2011, 08:03		#3
oort New Member alex Join Date: Jun 2009 Posts: 17 Rep Power: 16	Thank you for your answer. The flow is laminar, and high aspect ratio cell are mainly in flow direction (i think). I want simulate the flow in a channel filled with transversal cylinders. 3D simulations over a "cylinder module" with cyclic and symmetry conditions were done before. Also 2D simulations over full length of the channel. However 3D simulations gives a strange pattern in streamlines. 3D simulations were performed with a solver similar to channelFoam. 2D simulations with icoFoam. So now i would like to simulate the flow in a 3D geometry in the entire channel using icoFoam to see if the "strage" streamlines appears or the problem is the solver... 2D: 3D with cyclic/symmetry: 3D full channel: Well. Inserting some more cells and reducing grading the mesh are OK, but as increased from 319000 to 522000 cells... Checking geometry... Overall domain bounding box (0 0 0) (0.1989 0.03 0.0015) Mesh (non-empty, non-wedge) directions (1 1 1) Mesh (non-empty) directions (1 1 1) Boundary openness (-4.25725e-20 -3.21156e-20 -9.28017e-18) OK. Max cell openness = 2.19302e-16 OK. Max aspect ratio = 312.091 OK. Minumum face area = 1.73529e-10. Maximum face area = 1.14583e-05. Face area magnitudes OK. Min volume = 4.36933e-15. Max volume = 1.14089e-09. Total volume = 8.83461e-06. Cell volumes OK. Mesh non-orthogonality Max: 56.9438 average: 10.4105 Non-orthogonality check OK. Face pyramids OK. Max skewness = 1.12258 OK. Mesh OK.

September 16, 2011, 14:06		#6
oort New Member alex Join Date: Jun 2009 Posts: 17 Rep Power: 16	Hi Chris, Thanks for your reply. Well i can put some grading, but that will create very big cells in the beginning of entrance channel, but probably is the best. Anyway my 2D results even with this big drop in cell size are good (probably not so big because in 2D I have more cells in the entrance channel).