Extremely Small Time step to Maintain Courant Number of 10

willindsey · June 5, 2017, 23:50

I am running a komegasst simulation in parallel containing 4 million elements. I made my mesh using ansys meshing software and below is the checkMesh results:
Mesh stats
points: 2076394
internal points: 1949842
faces: 12198649
internal faces: 11945553
cells: 5177068
faces per cell: 4.66368
boundary patches: 4
point zones: 0
face zones: 1
cell zones: 1

Overall number of cells of each type:
hexahedra: 0
prisms: 3435930
wedges: 0
pyramids: 0
tet wedges: 0
tetrahedra: 1741138
polyhedra: 0

Checking topology...
Boundary definition OK.
Cell to face addressing 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 2238 1186 ok (non-closed singly connected)
outlet 2238 1186 ok (non-closed singly connected)
wall 19558 9911 ok (non-closed singly connected)
cube 229062 114533 ok (closed singly connected)

Checking geometry...
Overall domain bounding box (-1.09 -1.01797 -2.75) (1.09 1.0272 1.75)
Mesh has 3 geometric (non-empty/wedge) directions (1 1 1)
Mesh has 3 solution (non-empty) directions (1 1 1)
Boundary openness (-3.05175e-16 6.10809e-16 -3.99538e-17) OK.
Max cell openness = 6.63526e-16 OK.
Max aspect ratio = 20.153 OK.
Minimum face area = 3.21818e-08. Maximum face area = 0.0145499. Face area magnitudes OK.
Min volume = 1.00001e-11. Max volume = 0.000502156. Total volume = 20.0617. Cell volumes OK.
Mesh non-orthogonality Max: 62.8113 average: 12.8053
Non-orthogonality check OK.
Face pyramids OK.
Max skewness = 2.47308 OK.
Coupled point location match (average 0) OK.

Mesh OK.

End

In controlDict I specified that I need a courant number below 10 and openFoam corrects it. Except to have a courant number of 10, my time steps are around 1e-53 and then the last one is 1e-108 and the program throws a floating point error.
If someone minds taking a look that would be great. Thanks!
I can upload my files to if needed.

floquation · June 6, 2017, 04:22

Why are you going for 10?
Your simulation is probably just diverging, with velocities over 1e50.
Then, the dynamic Courant criterion demands that your timestep is less than 1e-53 (of course, depending on your cell sizes).

Try maxCo=0.5, and see what happens.

Co>1 is unstable. The only way to perform those kind of simulations is with heavy under-relaxation and possibly some other tricks.

jherb · June 6, 2017, 12:14

Which solver are you using? Do you use underrelaxation?

willindsey · June 6, 2017, 14:09

I am using pimple foam and my underelaxation values are .7 on U and .3 on everything else.

Mohammed_D · December 27, 2018, 18:11

Hello Everyone!

My calculation is converging evey time, and my courant number are as follows:

Global Courant Number [Variable Time Step Criteria] : 0.07

Global Courant Number [Explicit VOF Criteria] : 0.01

I am running my simulation on HPC cluster. Are those value good to have a flow ??

Thank you in advance!

June 5, 2017, 23:50	Extremely Small Time step to Maintain Courant Number of 10	#1
willindsey New Member wil Join Date: Oct 2016 Posts: 6 Rep Power: 10	I am running a komegasst simulation in parallel containing 4 million elements. I made my mesh using ansys meshing software and below is the checkMesh results: Mesh stats points: 2076394 internal points: 1949842 faces: 12198649 internal faces: 11945553 cells: 5177068 faces per cell: 4.66368 boundary patches: 4 point zones: 0 face zones: 1 cell zones: 1 Overall number of cells of each type: hexahedra: 0 prisms: 3435930 wedges: 0 pyramids: 0 tet wedges: 0 tetrahedra: 1741138 polyhedra: 0 Checking topology... Boundary definition OK. Cell to face addressing 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 2238 1186 ok (non-closed singly connected) outlet 2238 1186 ok (non-closed singly connected) wall 19558 9911 ok (non-closed singly connected) cube 229062 114533 ok (closed singly connected) Checking geometry... Overall domain bounding box (-1.09 -1.01797 -2.75) (1.09 1.0272 1.75) Mesh has 3 geometric (non-empty/wedge) directions (1 1 1) Mesh has 3 solution (non-empty) directions (1 1 1) Boundary openness (-3.05175e-16 6.10809e-16 -3.99538e-17) OK. Max cell openness = 6.63526e-16 OK. Max aspect ratio = 20.153 OK. Minimum face area = 3.21818e-08. Maximum face area = 0.0145499. Face area magnitudes OK. Min volume = 1.00001e-11. Max volume = 0.000502156. Total volume = 20.0617. Cell volumes OK. Mesh non-orthogonality Max: 62.8113 average: 12.8053 Non-orthogonality check OK. Face pyramids OK. Max skewness = 2.47308 OK. Coupled point location match (average 0) OK. Mesh OK. End In controlDict I specified that I need a courant number below 10 and openFoam corrects it. Except to have a courant number of 10, my time steps are around 1e-53 and then the last one is 1e-108 and the program throws a floating point error. If someone minds taking a look that would be great. Thanks! I can upload my files to if needed.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[snappyHexMesh] Error snappyhexmesh - Multiple outside loops	avinashjagdale	OpenFOAM Meshing & Mesh Conversion	53	March 8, 2019 10:42
High Courant Number @ icoFoam	Artex85	OpenFOAM Running, Solving & CFD	11	February 16, 2017 14:40
[mesh manipulation] Importing Multiple Meshes	thomasnwalshiii	OpenFOAM Meshing & Mesh Conversion	18	December 19, 2015 19:57
foam-extend_3.1 decompose and pyfoam warning	shipman	OpenFOAM	3	July 24, 2014 09:14
dynamic Mesh is faster than MRF????	sharonyue	OpenFOAM Running, Solving & CFD	14	August 26, 2013 08:47

June 6, 2017, 04:22		#2
floquation Senior Member Kevin van As Join Date: Sep 2014 Location: TU Delft, The Netherlands Posts: 252 Rep Power: 21	Why are you going for 10? Your simulation is probably just diverging, with velocities over 1e50. Then, the dynamic Courant criterion demands that your timestep is less than 1e-53 (of course, depending on your cell sizes). Try maxCo=0.5, and see what happens. Co>1 is unstable. The only way to perform those kind of simulations is with heavy under-relaxation and possibly some other tricks.

June 6, 2017, 12:14		#3
jherb Senior Member Joachim Herb Join Date: Sep 2010 Posts: 650 Rep Power: 22	Which solver are you using? Do you use underrelaxation?

June 6, 2017, 14:09		#4
willindsey New Member wil Join Date: Oct 2016 Posts: 6 Rep Power: 10	I am using pimple foam and my underelaxation values are .7 on U and .3 on everything else.

December 27, 2018, 18:11		#5
Mohammed_D New Member Mustapha Mukhtar Join Date: Dec 2018 Posts: 7 Rep Power: 8	Hello Everyone! My calculation is converging evey time, and my courant number are as follows: Global Courant Number [Variable Time Step Criteria] : 0.07 Global Courant Number [Explicit VOF Criteria] : 0.01 I am running my simulation on HPC cluster. Are those value good to have a flow ?? Thank you in advance!