CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   residuals still going down! (http://www.cfd-online.com/Forums/fluent/109755-residuals-still-going-down.html)

yorelchr November 26, 2012 04:27

residuals still going down!
 
Hello,

I'm running a case of turbogas. When I work on the "cold" case, to test the mesh, the residuals are never "plane". What could be the problem? same problem when switching on combustion.
I send you the pictures of these residuals.
http://cjoint.com/data/0KAjAQL9Y2T.htm

on which parameters should I "play" ?

thank you very much for any help

yorelchr

RodriguezFatz November 26, 2012 05:36

Residuals look quite normal to me. What exactly do you think the problem is? That some residuals start jittering at 1.0e-11 or that some of the others (continuity) are not "plain"?
How many cells does your grid have? 20,000 iterations seem to be pretty much... Maybe you can get faster convergence with a different solution method.

ghost82 November 26, 2012 09:42

Quote:

Originally Posted by yorelchr (Post 394143)
Hello,

I'm running a case of turbogas. When I work on the "cold" case, to test the mesh, the residuals are never "plane". What could be the problem? same problem when switching on combustion.
I send you the pictures of these residuals.
http://cjoint.com/data/0KAjAQL9Y2T.htm

on which parameters should I "play" ?

thank you very much for any help

yorelchr

As RodriguezFatz writes, residuals are fine: some small variation can appear, but it is quite usual.
To see if you have a converged solution monitor some parameters (species concentration, velocity, pressure, other..) on some planes into your domain; if they don't change with successive iterations, you have a converged solution.

yorelchr November 28, 2012 17:13

residuals and combustion
 
thank you Rodriguez Fatz and Ghost 82 for your help and sorry to reply so late.
I thought residuals should be plane, but I monitored some variables and it seems ok.
I have another question : my geometry is extracted from a bigger one. Since the bigger one has too many cells I have cut the domain and I put profiles coming from the entire turbogas simulations and I've kept the same mesh (it has been possible since the entire geometry has been meshed by block and I kept the original blocks (except of course the ones that I have eliminated). So, the strange thing is that the first 3000 iterations, my residuals are going down, really low...then the monitoring of the variables tell me that everything is changing and after many new iterations, I obtain the same results as with the entire domain. But with the entire domain, there is no need to wait for these first 3000 iterations, the "final results" are coming quicker....so if had not the full domain, I would have stopped my simulations during the first 3000 iterations since nothing anymore was happening...so...why this difference between the full domain and the extracted one?
(at this stage, there is no combustion, it's just for testing the mesh and checking the velocity profiles)
Thanks for any clue...
yorelchr

RodriguezFatz November 29, 2012 08:09

Maybe:
The equations of the reduced case differ from the complete case. Not only the full case has a larger system of equations, but also the boundary conditions are different, i.e. the bc of your reduced case lie somewhere inside your complete domain. So in one case the bc are "hard wired" to some value, whereas in the other (complete) case the same cells have to be calculated.
Also the solver will behave differently with two different sets of equations.
Then, the initialization differs and so on...
I would not be too suprised by this behavior.


All times are GMT -4. The time now is 18:25.