Mesh independence issues with wall shear stress

jw22 · June 28, 2011, 11:35

Hello,

I'm simulating steady flow through a rigid tube, so it should be quite straightforward to get mesh independent wall shear stress results, but I've been getting about a 10% change for each successive mesh where I cut the base size in half. I've gone up to a mesh with 600,000 cells, but the solution still isn't converging, and my laptop can't handle larger meshes. Perhaps my mesh isn't optimally set up? I'm using a polyhedral mesh with 10 prism layers. I'd appreciate it if someone familiar with this problem could help, thanks!

- Jay

abdul099 · July 10, 2011, 06:21

To get a mesh independent solution, a very, very fine mesh is needed. For example, with 10 prims layers you will obviously not be able to resolve the boundary layer as good as with 20 prism layers - and with that low number, it makes a difference whether you are using 10 more or less.

So you might have done all right but expecting the wrong result. Get rid of the idea, you could get a mesh independend solution with 600,000 cells. Most cfd-simulations are perfomed with a mesh which isn't fine enough to get a mesh independend solution. It up to the performing engineer to judge whether the errors are acceptable or not.

But I don't understand why your laptop can't handle bigger meshes. How old is it? You need approximately 1GB / million cells, and even my IBM T43 from 2005 has 2GB memory. And today, memory cost's nearly nothing!

June 28, 2011, 11:35	Mesh independence issues with wall shear stress	#1
jw22 New Member Jay Wang Join Date: Jun 2011 Posts: 1 Rep Power: 0	Hello, I'm simulating steady flow through a rigid tube, so it should be quite straightforward to get mesh independent wall shear stress results, but I've been getting about a 10% change for each successive mesh where I cut the base size in half. I've gone up to a mesh with 600,000 cells, but the solution still isn't converging, and my laptop can't handle larger meshes. Perhaps my mesh isn't optimally set up? I'm using a polyhedral mesh with 10 prism layers. I'd appreciate it if someone familiar with this problem could help, thanks! - Jay

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Natural convection in a closed domain STILL NEEDING help!	Yr0gErG	FLUENT	4	December 2, 2019 00:04
[Paraview] Wall Shear Stress	-mAx-	OpenFOAM Post-Processing	5	August 30, 2019 10:30
Water subcooled boiling	Attesz	CFX	7	January 5, 2013 03:32
fluent add additional zones for the mesh file	SSL	FLUENT	2	January 26, 2008 11:55
Flux depending on wall shear stress in Fluent 5.2	Andreas Borg	FLUENT	0	June 7, 2000 06:27

July 10, 2011, 06:21		#2
abdul099 Senior Member Join Date: Oct 2009 Location: Germany Posts: 636 Rep Power: 21	To get a mesh independent solution, a very, very fine mesh is needed. For example, with 10 prims layers you will obviously not be able to resolve the boundary layer as good as with 20 prism layers - and with that low number, it makes a difference whether you are using 10 more or less. So you might have done all right but expecting the wrong result. Get rid of the idea, you could get a mesh independend solution with 600,000 cells. Most cfd-simulations are perfomed with a mesh which isn't fine enough to get a mesh independend solution. It up to the performing engineer to judge whether the errors are acceptable or not. But I don't understand why your laptop can't handle bigger meshes. How old is it? You need approximately 1GB / million cells, and even my IBM T43 from 2005 has 2GB memory. And today, memory cost's nearly nothing!