CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   CFX iteration problem. (http://www.cfd-online.com/Forums/cfx/91216-cfx-iteration-problem.html)

windhair August 3, 2011 21:26

CFX iteration problem.
 
when I run a multi-phase (continuous fluid - dispersed fluid) simulation, the RMS seems very nice at the beginning, until to step 197, it suddently crashed with a overflow error, what may be the reason of such strange situiation ?

(BTW, the PC running the simulation has no internet connection, I can not upload the RMS pictures here.)

thank for you reply in advance.

ghorrocks August 4, 2011 01:11

Could be anything - but in general something has caused a gross numerical divergence despite being previously stable.

It could be physical - ie a shock wave has hit something or the first bubble has reached the exit boundry condition, or it could be numerical - ie a start-up smoothing parameter has turned off and now your simulation is unstable.

windhair August 4, 2011 10:37

I'm running a fluid-solid Euler-Euler simulation, with an artificial viscosity as discribed in document. Another supporter recommand me to use EDEM model in fluent, but we did not buy this license, will that special model give a more stable simulation? what is the best feature of EDEM?

ghorrocks August 4, 2011 18:44

As you have not descirbed your application I have not idea if your eular-eular approach is appropriate. EDEM is a very different approach as it is lagrangian particle tracking. Consider EDEM if you want to include more advanced particle interactions like maximum packing, particle collision, particle friction etc. Of course to use these advanced particle interactions you need to know what the maximum packing, collision physics and/or friction coefficients are for your particles!

I have no experience in coupling EDEM with CFX so cannot vouch for its stability. But I have no reason to suspect it would not be stable.

But I would not go to EDEM simply because your current approach is unstable. There is something you are doing with the current approach which is causing the problem, so work out what it is and fix it.

windhair August 8, 2011 12:06

thanks, i 'm still working on that. My application is about the mixing processing of very small particle, 0.1mm diameter, with a average density around 2000kg.m^-3, blowing by air. the gravity is also included. In the mixing , VF of the particle is about 50%, is this suitable to choice Euler-Euler model with a assumption to set the particle phase as a dispersed fluid, and set a artificial viscosity for it? I find some comment on such condition in CFX document. Under such situition, will the reference density for the buoyancy model play a sensentive rule in the simulation?

ghorrocks August 8, 2011 18:29

I do not have much experience with that exact setup so cannot say. All I can say is make sure you carefully validate your models and compare against quality experimental data.

altano August 10, 2011 04:20

Windhair,

I got same problem on my previous work.

How many particles represent your solid phase? Have you tried changing number of positions?

windhair September 5, 2011 03:50

Quote:

Originally Posted by altano (Post 319584)
Windhair,

I got same problem on my previous work.

How many particles represent your solid phase? Have you tried changing number of positions?

actually, using particle model is not really proper for my case, where the flo domain is about 2mX2mX2m, and the particle size is about 0.1mm, the total volume fraction of the particle is 70%, particle simulation is not suitable for such case as described in the CFX document. Thus a virtual fluid need to be configured as suggested by the CFX document.


All times are GMT -4. The time now is 01:29.