CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   MULTIPHASE & Mesh Resolution (http://www.cfd-online.com/Forums/fluent/107863-multiphase-mesh-resolution.html)

UDS_rambler October 8, 2012 11:31

MULTIPHASE & Mesh Resolution
 
Hi Everybody,

probably I'm going to ask you a stupid question, but recently I'm quite confused about it.:confused:

I'm coping with a multiphase model. The model itself is quite simple:

- a 3D domain (50 mls of volume) which I reconstructed and meshed in Gambit;
- a hydrotransport model system: disperse phase is transported within an acqueous primary phase. The dimension of the disperse phase spheres is 200 microns.

My doubt is: does the resolution of my mesh have to take into account the dimension of the spheres? do I have to have a mesh in which the average edgelength of the elements is at least 200 microns?
If I do so, my mesh will consist of 6 millions elements and this looks a bit odd in my opinion (considered the small volume of my control domain).

I'm pretty tempted to drastically reduce my mesh resolution, but I'm not sure about the consequences.

I thank you so much for your kind help

Best regards

UDS_rambler

RodriguezFatz October 9, 2012 03:50

Hi,

I think the DPM model implies the idea that particles move through cells and do not fill a complete cell. If your cells are smaller than your particles (or about the same size), the model can't work properly. As far as my experience goes, you get numerical problems, when too many particles move through these small cells. I think the Fluent manual also says more or less the same.

But what, if you need a small grid - e.g. when you can't use wall functions? I don't know any trick, maybe someone can help.

UDS_rambler October 9, 2012 10:41

Thank you so much RodriguezFats!

It is definitely what I was thinking about! And actually, momentum equations didn't converge!

Respect to the problem you asked about, are you refering to DPM or multiphase? Tell me more, I'll try to be useful if I can!

RodriguezFatz October 9, 2012 10:57

* to your second post: I am not sure about the terminology, but isn't DPM a multiphase model? Anyway, I ment DPM. What if I have particles of size 0.1mm and my grid has to be 0.01mm at the wall to get y+ = 1 ?

* to your first post: I think I don't understand your argumentation: If your average edge length is at least 200 microns - this idea results in some maximum number of cells your grid can have. Now you say "If you do so..." your grid consist of some enormous large number of cells. That doesn't fit together. Maybe language problems... What means "a bit odd" in this context. = "too much"?

UDS_rambler October 9, 2012 11:32

Ok, you're right we probably misunderstood each other: I used a bad terminology.

Thanks to your answer about my first post I solved my problem: I yielded a too much fine mesh and momentum equation didn't converge due to this. Now I remeshed my model and it works perfectly.

About your question: you're right DPM is a multiphase physics, but if you check fluent user's manual DPM adopts Eulerian-Lagrangian approach, while multiphase models Eulerian-Eulerian. By the way, have you already tried to build the model and run it? I would try because, in your case, although particles are greater than mesh elements, they don't have comparable dimensions and I figure out that with DPM model momentum will converge.

RodriguezFatz October 9, 2012 11:36

Yes, I used DPM and the model worked fine (or at least converged) for a particle size smaller than the small boundary sheath cells. However it diverged every time I had smaller cells in the boundary layer.

UDS_rambler October 9, 2012 12:00

Ok, if you've already tried I can't help you ... Unfortunately I've used DPM for particles smaller than mesh size too! The only thing I can tell you is to check the volume fraction that your despersed phase occupies: DPM doesn't work for volume fraction greater than 10%. Obviously, I don't know how exactly your model is, but maybe it can help you...

nkme2007 October 10, 2012 06:45

Hello All,

I want to do analysis of heat transfer from water flowing through pipes submerged inside concrete. I am modelling in GAMBIT and wish to analyse it on Ansys FLUENT.

Can anybody help me out, how to model and simulate?

Does any tutorials exist?

UDS_rambler October 10, 2012 06:55

Hi nkme2007,

just take a look here:

http://confluence.cornell.edu/displa...arning+Modules

Maybe there's something suitable for you.
Anyway, just read the manual in "modelling heat transfer "chapter, it's pretty exhaustive!

Best regards
G

nkme2007 October 10, 2012 06:58

Thank you G,

I have practiced the manual, but i felt that there is no chapter that caters my need. Thanks and I will read it once again.

nkme2007 October 10, 2012 07:01

Hi UDS_rambler,

Where does the "modelling heat transfer" chapter exist?

UDS_rambler October 10, 2012 07:07

Well,

I use Fluent 6.3 and I find it in the User's guide. You can find the guide on the web too, just check "fluent user's guide" on google...

Best regards
G

nkme2007 October 10, 2012 07:10

yah... thank you UDS_rambler

nkme2007 October 10, 2012 07:26

Hi UDS_rambler,

Do you have "periodic_flow_heat.zip"? Can you send please?

UDS_rambler October 10, 2012 07:29

I'm sorry, I don't have it...


All times are GMT -4. The time now is 14:00.