CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   convergence problem (http://www.cfd-online.com/Forums/cfx/97920-convergence-problem.html)

pato February 28, 2012 11:00

convergence problem
 
hi, i am trying to modelate a difusser with porous media instead of perforated plate, with an steady state model. My problem is that i donīt get any convergence. I know i have to look at the residuals since a read cfx-documentation. Here is the location of max residualas:

--------------------------------------------------------------------+
| Locations of Maximum Residuals |
+--------------------------------------------------------------------+
| Equation | Node # | X | Y | Z |
+--------------------------------------------------------------------+
| U-Mom-fluid media | 1481 | 1.926E-02 | 1.780E-01 | 0.000E+00 |
| V-Mom-fluid media | 35131 | 4.058E-03 | 1.780E-01 |-8.202E-03 |
| W-Mom-fluid media | 654443 | 4.795E-02 | 1.880E-01 |-2.501E-03 |
| P-Mass-fluid media | 33090 | 7.858E-03 | 1.780E-01 |-4.402E-03 |
| U-Mom-porous media | 70 | 0.000E+00 | 1.880E-01 |-8.202E-03 |
| V-Mom-porous media | 9790 | 4.058E-03 | 1.780E-01 |-8.202E-03 |
| W-Mom-porous media | 1033 | 1.926E-02 | 1.780E-01 | 0.000E+00 |
| P-Mass-porous media | 1039 | 0.000E+00 | 1.780E-01 |-3.740E-03 |
| K-TurbKE-fluid media | 654443 | 4.795E-02 | 1.880E-01 |-2.501E-03 |
| E-Diss.K-fluid media | 1481 | 1.926E-02 | 1.780E-01 | 0.000E+00 |
| K-TurbKE-porous medi | 12455 | 4.031E-01 | 1.780E-01 |-8.202E-03 |
| E-Diss.K-porous medi | 2253 | 4.012E-01 | 1.780E-01 |-1.005E-02 |
+--------------------------------------------------------------------+



I always get the max residuals in those surfaces which are perpendicular to the given moment type (X;Y,Z) i.e, for max U-Mom, i have max residuals in surfacers perpendicular to x direction. I donīt know whether that is very important or not. I already tried changing timescale factor and others, but my model does not reach convergence. COULD YOU PLEASE HELP ME TO REACH CONVERGENCE????


thank you

Chander February 28, 2012 13:36

You should explain your problem in more detail. Location of the residuals does not say much unless you also show the domain.
Try the following quick checks..
check mesh quality in the .out...make sure everything is fine
what is the Reynolds number of your flow...are you using the correct fluid model (laminar/ turbulent)?
also look at the following link for guidance..
http://www.cfd-online.com/Wiki/Ansys...gence_criteria

pato February 28, 2012 15:27

actually i had good convergence when i had a normal fluid media for a perforated diffuser, but when i tried to change this perforated plate into a porous media, is when i cannot get convergence. Is it more difficult to reach convergence in porous models?
my model is quite simple and with the perforated plate worked well with steady state simulations. Should it also work with steady state simulations when changing it into a porous media??

ghorrocks February 28, 2012 18:23

There has been some modifications in V14 to make porous regions converge better. What version of CFX are you running?

pato February 28, 2012 19:20

ansys 12.1 cfx-5

ghorrocks February 29, 2012 06:51

I think you will find the convergence behaviour has improved significantly with the latest release. Time to upgrade to V14.

pato March 1, 2012 05:52

But is it normal that exactly with same geom and boundary conditions as in a normal fluid media, i don`t reach convergenece?? could the problem have changed to transient???
I am quite stucked and can not go any further. My monitor points indicating velocity(the ones at the beginning and at the end of the diffuser) begin by 20 it to oscilate and reach high velocities.

Could anyone help meŋ

Chander March 1, 2012 05:59

Yes, it is possible. Flow impingement on porous medium can cause problems in convergence. You asked wether your problem is transient. You can check this by following a simple check given in CFX manual under tips for convergence.
Also, try increasing the timestep and grid size..that may smear out the transients.
Just go through once the CFX manual part on convergence problems..that should give you some clue.
Also, as I asked you earlier, did you check if you are using the correct fluid model as per the flow reynolds number?

Chander March 1, 2012 06:01

And do try with V14 as per Glen's suggestion. I had similar problems but unfortunately no access to CFX 14.

pato March 1, 2012 06:54

Unfortunately i also do not have acces to v14.
quick summary,

with 0Pa static pressure as opening boundary condition i got floating point problem, overflow. After seaching on the forum, i changed this pressure for a small one, 8Pa, and also changed to SST turbulence model, and timestep 0.1, beeing those boundaries maybe the cause of the floating point.

Well, my mass and momentum plots do not oscilate anymore, but residuals do have oscilatory behaviour. Velocities in monitor points oscilate a lot(think of function sin(x), my velocities oscilates similar to this function, up and down).
i checked and tried things in the manual like:

decreasing timestep size,
using physical timescale

but nothing seems to work

should i try transient?


All times are GMT -4. The time now is 06:28.