CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CD-adapco (http://www.cfd-online.com/Forums/cd-adapco/)
-   -   Convergence problems with different mesh size (http://www.cfd-online.com/Forums/cd-adapco/55298-convergence-problems-different-mesh-size.html)

echo April 13, 2006 04:58

Convergence problems with different mesh size
 
Hi,everyone! I'm planning to simulate a steady-state incompressiable turbulent flow of catalyst converter, but calculations don't seem to give a stable solution with my model.

I've set up the model in stardesign. Let me discribe my model and wish someone would help me. There are four cylinders and two lofts from circle in my medel.The cylinders' coordinate,radius and height are (0,0,0,) r=72 h=150 ,(0,0,150) r=75 h=240 ,(0,0,445) r=20 h=300 ,(0,0,-130) r=20 h=-125 ,and the four circles is (0,0,0) r=75 (0,0,-130) r=20 , (0,0,390) r=75 ,(0,0,445) r=20. And I have defined the surface of cylinder (0,0,-130) r=20 h=-125 as the inlet, the surface of cylinder (0,0,445) r=20 h=300 as the outlet and the cylinder (0,0,0,) r=72 h=150 as the porous. In stardesign when I generate the tetrahetral mesh as the 3% of the medel, the model would run and converge. But I think the result is wrong. So I want to change the mesh size. But when I change the mesh size to 4% or 2%, it can't converge! why does this happen? I also import the .dbs file to proam and generate the tetrahedral mesh in it as about the mesh size of 3% in stardesign. But it diverges in prostar.

Since I'm a newbie with Star, I'm afraid I could have made many mistakes setting the values of the various parameters (e.g.: initialization; monitoring and reference; turbulence intensity and mixing lenght; etc.). I have chosen the SIMPLE algoritm and the UD diff. sch.; is it ok ? As you can see I still have to learn many things, but if anybody could answer to some of these questions it would be of great help for me. Thanks in advance,

Echo


echo April 13, 2006 06:04

more information
 
I have reduced the relaxation fator:u from 0.7 to 0.2 and pressure from 0.3 to 0.1. But this can't converge too. And the mass is worse.

kathir April 15, 2006 02:46

Re: more information
 
please try to use PISO ALGORITHM. IT IS VERT USEFUK FOR TRANSIENT PROBLEM. SOME TYPES STEADY STATE PROBLEMS WE SOLVE BY TRASIENT METHOD.

echo April 15, 2006 04:38

Re: more information
 
OK!I will try!Thank you for your reply!

echo April 15, 2006 07:36

Re: more information
 
Hi,kathir. I have run the medel with PISO. But the residual of velocity is worse and it diverge more rapidly. And have many other warnings as: warning #013 maximum specified corrector stages 20 reached before convergence criterion is satisfied, warning #001 fluid inflow detected at some outflow boudaries where mass fluxes are forced to zero, warning #065 out of bounds value -8.5077E+06 for scalar variable ENTH at cell 214002. corrective

These warning didn't appear when using simple. Could I increase the maximum number of corrector stages? Thank you for your attention! Wish your reply!

Anton Lyaskin April 16, 2006 09:22

Re: more information
 
First, try to change outlet boundary to something else, e.g. pressure. As I can guess from #065 warning you're solving for compressible fluid and usually outlet type boundary is not the best choice for these problems.

Second, instead of increasing the number of corrector stages (you can do it in Analysis Controls->Solution Controls->Solution Method) try to lower the time step. It's a good idea to check .info file for the mean value of Courant number - though you use implicit scheme (which is the default option) it's still a good idea to keep at least the mean value of Courant number below 1.0.

echo April 17, 2006 07:11

Re: more information
 
Hi,anton! I'm solving imcompressible fluid. When I adopt the simple algorithm, there isn't this warning. But just when I change the simple algorithm to PISO, this warning #065 appears. And when I use the simple algorithm and define the pressure boundary as the outlet, this warning appears also. And it couldn't converge too.

How to lower the time step? Thank you! Wishing!

Tom April 17, 2006 13:04

Re: Convergence problems with different mesh size
 
Let me preface my response by saying I have never used tet mesh, I always use hex. In a hex mesh if the mass won't converge, that usually means that the inlet doesn't see the outlet. You did not say how you refined the mesh. I have found that if I start with a hex mesh and use refine from the cell tool GUI, I always have to create couples after I am through. Of course, some couples created will not be correct, so assure that you merge couples, check couples, etc. Do a quick check by using the check tool and connectivity.

Tom

echo April 17, 2006 20:29

Re: Convergence problems with different mesh size
 
Thank you for your reply.I find that I don't need to create the couples when using tet. I have checked everything through check model setup and there isn't any problem. I also do just as you tell me to check the connectivity and the output window tells ALL CELLS IN THE SELECTED RANGE APPEAR TO BE CONNECTED IN ONE GROUP. I think the tet mesh doesn't need the couple. Am i right?

Anton Lyaskin April 18, 2006 05:52

Re: more information
 
To lower the time step in ProSTAR StarGUIde go to Analysis Preparation/Running->Set Runtime Controls and change the value in 'Time Step for Period'

DenizeN April 19, 2006 04:13

Re: more information
 
Anton Laskin write to den_umd@ukr.net if you want to exchange information about STAR-CD on russion.


All times are GMT -4. The time now is 18:07.