What is the smallest mesh I can have?

musahossein · July 6, 2014, 18:14

Dear all:
I have a very small tank (0.1mx0.510m). Mesh size is 50X200. Water depth is 0.021m. When I apply a sinusoidal displacement to it, I see wave breaking, but the graphics does not seem right. Can anyone tell me if this is because of the tiny mesh size? I am using paraview 4.1.0 and openfoam 2.2.1. I have attached the paraview image. The input waveform is also attached.

Thankyou,
Musa

siamakghh2000 · July 7, 2014, 04:42

I am strongly doubtful that a small mesh is posing a problem, the smaller the mesh the safer the solution. Even from a certain point on, solution should be independent of mesh size.
If you like to know the the smallest mesh you can have, it is found in a trial and error procedure. That is, your y+ should be at most 2 next to the wall for correct shear stress calculation (running the simulation and checking the y+, changing the mesh and running it again).

musahossein · July 7, 2014, 07:31

Quote:

Originally Posted by siamakghh2000

I am strongly doubtful that a small mesh is posing a problem, the smaller the mesh the safer the solution. Even from a certain point on, solution should be independent of mesh size.
If you like to know the the smallest mesh you can have, it is found in a trial and error procedure. That is, your y+ should be at most 2 next to the wall for correct shear stress calculation (running the simulation and checking the y+, changing the mesh and running it again).

Understood. Thankyou. However what do you mean by at most 2? Do you mean the number of cells in the vertical direction (wave elevation axis -- z in my case) should be 2 times more than the number of cells in the horizontal direction (direction of wave motion +- y in my case?). I look forward to your reply.

siamakghh2000 · July 7, 2014, 08:08

Anytime.
No, the point is to stay in the viscus sub-layer which happens at y+<5 (y+=yu*/viscosity), when I say y+ at most 2. This is because wall shear stress is calculated from laminar shear stress formula ,Wall shear stress= Mu*du/dy (Mu laminar viscosity).
However, 5 is too big for y+, therefore they say that 2 is better.
You will need to calculate y+ at the end of a solution convergence for a given mesh configuration. That is, u* which is an always constant value beside the wall and y value which points to the location of first grid point next to the wall to be able to calculate y+=yu*/viscosity
I should actually have asked whether you are developing a code by yourself. Because if you make use of commercial codes, I think they use wall functions. In that case you have nothing to worry about.

musahossein · July 7, 2014, 09:15

I am using OpenFoam. Your discussion was most insightful -- I should have done some diligence before posing this question. So thankyou again.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[snappyHexMesh] Add Mesh Layers doesnt work on the whole surface	Kryo	OpenFOAM Meshing & Mesh Conversion	13	February 17, 2022 07:34
[snappyHexMesh] snappyHexMesh & Mesh around a missile	sasanghomi	OpenFOAM Meshing & Mesh Conversion	2	October 15, 2014 19:54
engrid -> save as .stl with boundarie codes	Zymon	enGrid	31	August 29, 2011 13:40
Smallest mesh size	cfd_newbie	FLUENT	2	December 5, 2008 05:29
basic of mesh refinement	arya	CFX	4	June 19, 2007 12:21

July 7, 2014, 04:42		#2
siamakghh2000 New Member Siamak Gharahjeh Join Date: Aug 2012 Posts: 27 Rep Power: 13	I am strongly doubtful that a small mesh is posing a problem, the smaller the mesh the safer the solution. Even from a certain point on, solution should be independent of mesh size. If you like to know the the smallest mesh you can have, it is found in a trial and error procedure. That is, your y+ should be at most 2 next to the wall for correct shear stress calculation (running the simulation and checking the y+, changing the mesh and running it again).

July 7, 2014, 08:08		#4
siamakghh2000 New Member Siamak Gharahjeh Join Date: Aug 2012 Posts: 27 Rep Power: 13	Anytime. No, the point is to stay in the viscus sub-layer which happens at y+<5 (y+=yu/viscosity), when I say y+ at most 2. This is because wall shear stress is calculated from laminar shear stress formula ,Wall shear stress= Mudu/dy (Mu laminar viscosity). However, 5 is too big for y+, therefore they say that 2 is better. You will need to calculate y+ at the end of a solution convergence for a given mesh configuration. That is, u* which is an always constant value beside the wall and y value which points to the location of first grid point next to the wall to be able to calculate y+=yu*/viscosity I should actually have asked whether you are developing a code by yourself. Because if you make use of commercial codes, I think they use wall functions. In that case you have nothing to worry about.

July 7, 2014, 09:15		#5
musahossein Senior Member musaddeque hossein Join Date: Mar 2009 Posts: 309 Rep Power: 18	I am using OpenFoam. Your discussion was most insightful -- I should have done some diligence before posing this question. So thankyou again.