Numerical inaccuracy at domain interfaces in CFX

Chander · March 23, 2011, 06:12

Hi,

I have a computational domain which I I have divided into different parts and I have meshes these parts seperately in ICEM. I did so because I wanted to keep my model modular for future parametric studies on different parts.
Next, I have imported meshes for these parts into CFX (taking adequate care of the spatial relation between them). Then I define a domain for each of these meshes and I also define interfaces between these meshes (i.e domain interfaces) to ensure mass, momentum and energy flux conservation. The type of interface used is GGI.
My query is that if one has many such interfaces due to inclusion of many meshes in a CFX case, then can these interfaces lead to numerical inaccuracy and thus cause convergence problem?

ghorrocks · March 23, 2011, 17:30

In general no. 1 to 1 interfaces are direct links and create no inaccuracy or convergence issue. GGI interfaces do involve some interpolation and so can create problems, but for most applications are fine and the accuracy of the interpolation is more than adequate.

Chander · March 24, 2011, 07:52

Thanks for your reply Glenn!
I have many interfaces in my fluid domain and I will try to simplify my analysis by removing some of these interfaces.
I would be greatly helped if you could also have a look and reply to my thread 'Turbulence modeling in cfx'
http://www.cfd-online.com/Forums/cfx...eling-cfx.html

March 23, 2011, 06:12	Numerical inaccuracy at domain interfaces in CFX	#1
Chander Senior Member Join Date: Oct 2010 Location: Zurich Posts: 176 Rep Power: 15	Hi, I have a computational domain which I I have divided into different parts and I have meshes these parts seperately in ICEM. I did so because I wanted to keep my model modular for future parametric studies on different parts. Next, I have imported meshes for these parts into CFX (taking adequate care of the spatial relation between them). Then I define a domain for each of these meshes and I also define interfaces between these meshes (i.e domain interfaces) to ensure mass, momentum and energy flux conservation. The type of interface used is GGI. My query is that if one has many such interfaces due to inclusion of many meshes in a CFX case, then can these interfaces lead to numerical inaccuracy and thus cause convergence problem?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
CFX: How to save variables of boundary domain of transient run and use it as input	papanef	CFX	5	November 10, 2010 16:31
CFX: Transient sim with fluid-fluid domain interface	farzola	CFX	4	April 7, 2010 15:08
Numerical Diffusion in CFX	John S.	CFX	4	August 17, 2008 19:47
Interfaces 1:1 vs numerical error.	JP	CFX	1	April 29, 2008 18:48
boundary conditions in numerical domain	Bounecer	FLUENT	0	October 20, 2005 07:15

March 23, 2011, 17:30		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	In general no. 1 to 1 interfaces are direct links and create no inaccuracy or convergence issue. GGI interfaces do involve some interpolation and so can create problems, but for most applications are fine and the accuracy of the interpolation is more than adequate.

March 24, 2011, 07:52		#3
Chander Senior Member Join Date: Oct 2010 Location: Zurich Posts: 176 Rep Power: 15	Thanks for your reply Glenn! I have many interfaces in my fluid domain and I will try to simplify my analysis by removing some of these interfaces. I would be greatly helped if you could also have a look and reply to my thread 'Turbulence modeling in cfx' http://www.cfd-online.com/Forums/cfx...eling-cfx.html