solution not converging for fine mesh..

saurabh.deshpande88 · February 1, 2010, 09:18

Hello everyone,I am doing analysis of flow around a NACA 0008 airfoil. Using gambit and fluent for the purpose.
When i use a coarser mesh (i.e. nodes=1,500) my solution is converging and i am getting the results.
But for a fine mesh (nodes=12,000) even after 1,60,000 iterations my solution is not converging and there is not a slightest sign that it will come down! In fact the "residual vs iterations" curve follows a zig-zag pattern. Why is it so?? please help. thanks..

IMP DATA:

residual: 1e-06

inviscid flow, angle of attack=20 degree

Discretization schemes: pressure= presto
momentum= second order upwind

Pressure velocity coupling: simple

underrelaxation factors have been reduced to the extent of 0.01 for pressure and momentum.

DoHander · February 1, 2010, 15:50

Using a small mesh starts to smell like a DNS (Direct Navier Stokes) but if it is 2D and steady is not DNS

. The moral is - if you use an extremely fine mesh this must be fine enough for representing the physics of your case. A fine mesh will usually catch small zones of separations and reattachments "invisible" for a coarse mesh - this is probably why you can't obtain a convergent solution (steady ?).

Do you use a turbulence model in your simulation ?

Do

saurabh.deshpande88 · February 2, 2010, 10:23

Hey thanks for the reply.

I got what you mean to say. due to "seperation and reattachment" of flow at micro level velocity changes continiously and solution is not getting converged.

This also explains why solution was converging for lower angles of attack (< 20 degree) for the same fine mesh. seperation starts as angle of attack increases and solution convergence stops.

i am not using a turbulent model. i am a begineer in the field of CFD. i am using inviscid model. should i change to turbulent models?

(But

, actually if i am considering inviscid model then viscosity is zero and hence there should be no skin friction, no boundry layer formation and hence no seperation.. then why is it not converging? )

February 1, 2010, 09:18	solution not converging for fine mesh..	#1
saurabh.deshpande88 New Member saurabh deshpande Join Date: Jan 2010 Posts: 4 Rep Power: 17	Hello everyone,I am doing analysis of flow around a NACA 0008 airfoil. Using gambit and fluent for the purpose. When i use a coarser mesh (i.e. nodes=1,500) my solution is converging and i am getting the results. But for a fine mesh (nodes=12,000) even after 1,60,000 iterations my solution is not converging and there is not a slightest sign that it will come down! In fact the "residual vs iterations" curve follows a zig-zag pattern. Why is it so?? please help. thanks.. IMP DATA: residual: 1e-06 inviscid flow, angle of attack=20 degree Discretization schemes: pressure= presto momentum= second order upwind Pressure velocity coupling: simple underrelaxation factors have been reduced to the extent of 0.01 for pressure and momentum.

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
grid dependancy	gueynard a.	Main CFD Forum	19	June 27, 2014 21:22
[ICEM] Unstructure Meshing Around Imported Plot3D Structured Mesh ICEM	kawamatt2	ANSYS Meshing & Geometry	17	December 20, 2011 11:45
basic of mesh refinement	arya	CFX	4	June 19, 2007 12:21
Solution mesh independent	Armin Gips	FLUENT	2	August 13, 2001 10:35
Mesh	Mignard	FLUENT	2	March 22, 2000 05:12

February 1, 2010, 15:50		#2
DoHander Senior Member Join Date: Nov 2009 Posts: 411 Rep Power: 20	Using a small mesh starts to smell like a DNS (Direct Navier Stokes) but if it is 2D and steady is not DNS . The moral is - if you use an extremely fine mesh this must be fine enough for representing the physics of your case. A fine mesh will usually catch small zones of separations and reattachments "invisible" for a coarse mesh - this is probably why you can't obtain a convergent solution (steady ?). Do you use a turbulence model in your simulation ? Do

February 2, 2010, 10:23		#3
saurabh.deshpande88 New Member saurabh deshpande Join Date: Jan 2010 Posts: 4 Rep Power: 17	Hey thanks for the reply. I got what you mean to say. due to "seperation and reattachment" of flow at micro level velocity changes continiously and solution is not getting converged. This also explains why solution was converging for lower angles of attack (< 20 degree) for the same fine mesh. seperation starts as angle of attack increases and solution convergence stops. i am not using a turbulent model. i am a begineer in the field of CFD. i am using inviscid model. should i change to turbulent models? (But , actually if i am considering inviscid model then viscosity is zero and hence there should be no skin friction, no boundry layer formation and hence no seperation.. then why is it not converging? )