Strange Result for Versteeg Testcase (2D, convection only, steady state)

caramelo · August 16, 2011, 04:42

Hi everybody,

I'm new to OpenFOAM and tried to implement a testcase from the Versteeg and Malalasekera book[1]. GoogleBooks

To solve the problem I used the scalarTransportFoam solver and set DT = 0 as it is a convection only problem. I used the boundary condition from the book except for the two outlet patches, where I used zeroGradient for phi.

The result I get are quite normal as long as I don't reach a certain number of cells. For example if my domain is 1 x 1 m and I use more than 21 cells in the x_1 and the x_2 direction the result is completely instable and has nothing to do with the analytic solution. I solved the div(phi,T) term with Gauss linear. But the solution doesn't converge either if I use QUICK or vanLeer to compute the gradient. Only upwind always give a reasonable but not really accurate solution.

Does anybody have an idea what problem is?

Thanks in advance.

caramelo

santiagomarquezd · August 16, 2011, 15:16

Can you post an sketch of the problem and BC's used?

Regards.

caramelo · August 16, 2011, 16:45

You can see a sketch in the GoogleBooks links i posted. I just added the whole case as an attachment. With this setting you will get a reasonable result. If you now increase the numbers of cells in the blockMeshDict, no reasonable result will be received any more.

l_r_mcglashan · August 16, 2011, 18:17

When you get finer grid cells, I think there's an issue with the Courant number for non-upwind schemes. Try changing ddtSchemes to Euler and make the endTime in controlDict something like 2; I get 'better' results with that.

I note that:

"At the first and last nodes where the diagonal intersects the boundary a value of 50 is assigned to \phi."

Might that be an issue?

alberto · August 17, 2011, 18:55

It is simply a question of conditioning of the linear problem you are trying to solve. You would need under-relaxation (have to modified the code adding T.relax()) to use the code as steady state.

Alternatively, simply perform a pseudo-transient simulation, as suggested by Laurence, since the effect is the same as under-relaxing.

P.S. It is not a problem in the schemes ;-)

Best,

August 16, 2011, 04:42	Strange Result for Versteeg Testcase (2D, convection only, steady state)	#1
caramelo New Member Chris Join Date: Jun 2011 Posts: 12 Rep Power: 15	Hi everybody, I'm new to OpenFOAM and tried to implement a testcase from the Versteeg and Malalasekera book[1]. GoogleBooks To solve the problem I used the scalarTransportFoam solver and set DT = 0 as it is a convection only problem. I used the boundary condition from the book except for the two outlet patches, where I used zeroGradient for phi. The result I get are quite normal as long as I don't reach a certain number of cells. For example if my domain is 1 x 1 m and I use more than 21 cells in the x_1 and the x_2 direction the result is completely instable and has nothing to do with the analytic solution. I solved the div(phi,T) term with Gauss linear. But the solution doesn't converge either if I use QUICK or vanLeer to compute the gradient. Only upwind always give a reasonable but not really accurate solution. Does anybody have an idea what problem is? Thanks in advance. caramelo

August 16, 2011, 15:16		#2
santiagomarquezd Senior Member Santiago Marquez Damian Join Date: Aug 2009 Location: Santa Fe, Santa Fe, Argentina Posts: 452 Rep Power: 24	Can you post an sketch of the problem and BC's used? Regards. __________________ Santiago MÁRQUEZ DAMIÁN, Ph.D. Research Scientist Research Center for Computational Methods (CIMEC) - CONICET/UNL Tel: 54-342-4511594 Int. 7032 Colectora Ruta Nac. 168 / Paraje El Pozo (3000) Santa Fe - Argentina. http://www.cimec.org.ar

August 16, 2011, 18:17		#4
l_r_mcglashan Senior Member Laurence R. McGlashan Join Date: Mar 2009 Posts: 370 Rep Power: 23	When you get finer grid cells, I think there's an issue with the Courant number for non-upwind schemes. Try changing ddtSchemes to Euler and make the endTime in controlDict something like 2; I get 'better' results with that. I note that: "At the first and last nodes where the diagonal intersects the boundary a value of 50 is assigned to \phi." Might that be an issue? __________________ Laurence R. McGlashan :: Website Last edited by l_r_mcglashan; August 16, 2011 at 18:56.

August 17, 2011, 18:55		#5
alberto Senior Member Alberto Passalacqua Join Date: Mar 2009 Location: Ames, Iowa, United States Posts: 1,912 Rep Power: 36	It is simply a question of conditioning of the linear problem you are trying to solve. You would need under-relaxation (have to modified the code adding T.relax()) to use the code as steady state. Alternatively, simply perform a pseudo-transient simulation, as suggested by Laurence, since the effect is the same as under-relaxing. P.S. It is not a problem in the schemes ;-) Best, __________________ Alberto Passalacqua GeekoCFD - A free distribution based on openSUSE 64 bit with CFD tools, including OpenFOAM. Available as in both physical and virtual formats (current status: http://albertopassalacqua.com/?p=1541) OpenQBMM - An open-source implementation of quadrature-based moment methods. To obtain more accurate answers, please specify the version of OpenFOAM you are using.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Time step dependence of convergence behavior of steady state simulations in CFX	Chander	Main CFD Forum	5	December 23, 2013 06:31
Constant velocity of the material	Sas	CFX	15	July 13, 2010 09:56
regarding steady state discrete phase calculations	hajszan_gyula	FLUENT	1	February 28, 2006 02:32
About the difference between steady and unsteady problems	Lisa	Main CFD Forum	11	July 5, 2000 15:37
Steady state formulation of turbulence ??	Jitendra	Main CFD Forum	1	June 27, 2000 19:49