CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   [HELP] mass fraction in multi-component granular flow? (http://www.cfd-online.com/Forums/fluent/98649-help-mass-fraction-multi-component-granular-flow.html)

lihuang March 15, 2012 10:14

[HELP] mass fraction in multi-component granular flow?
 
Hi all,

I am trying out FLUENT Mixture multiphase model to solve multi-component granular flow. I have two velocity inlets and one pressure outlet. The fluid is air, and the secondary phases are multi-component particles of different sizes. At these two inlets, the mass fraction of each component (species) is different. For example, (0.2A+0.8B) at inlet 1 and (0.5A+0.5B) at inlet 2. Even with no chemical reaction or mass transfer between phases, it seems the mass fractions of secondary phases are not solved by FLUENT. Any ideas?

BTW, I tried Eulerian model. It does solve for mass fraction of the secondary granular phases, but it did not converge well. And it takes much longer time so I dont want to consider Eulerian model for now (unless absolutely necessary:().

doronzo March 15, 2012 12:36

hi

how big are the particles? how concentrated is the initial flow? do the particles enter the domain in (quasi) mechanical equilibrium?

regards

lihuang March 15, 2012 12:56

Quote:

Originally Posted by doronzo (Post 349664)
hi

how big are the particles? how concentrated is the initial flow? do the particles enter the domain in (quasi) mechanical equilibrium?

regards

Thanks for your reply. :)
Particles are very small with diameters in the range of nm to micrometers. The volume fraction of particles from inlet is around 1E-10. But what does it mean by being in mechanical equilibrium? How do I specify it?

doronzo March 15, 2012 13:05

...if the particles enter the domain with the same, or similar, velocity.

why not to use the DPM (discrete phase model)? has your grid a high resolution?

regards

lihuang March 15, 2012 13:20

Quote:

Originally Posted by doronzo (Post 349674)
...if the particles enter the domain with the same, or similar, velocity.

why not to use the DPM (discrete phase model)? has your grid a high resolution?

regards

particles are entering with the same velocity. I am not using DPM because I am having a great number of small particles. I am not interested in the actual tracks of each particles but the volume fraction and composition of particles far away from the inlets. In the future I have to expand my open domain to about several hundred meters. When I read the introductions to FLUENT multiphase model, I figured to go with the mixture model.
I dont have a very high resolution grid right now. But I plan to do so.

doronzo March 15, 2012 13:43

it indeed depends on the goals, but in general a particle concentration like that isn't high. also, the number of tracked particles depends a lot on the grid, because they are treated as cluster of particles, so I don't expect you would have a prohibitive number of trajectories.

in a first approximation you could consider your small particles as fully coupled to the fluid, and treat the mixture as multispecies (Fluent multispecies model), i.e. air (species 1) + pseudofluid (species 2, with its properties modified by the particles, e.g. increased density and viscosity; Marble 1970, Annu. Rev. Fluid Mech.).

in this second case, you could directly follow the species mass fraction, but of course you cannot have the particles leave the fluid transport system

lihuang March 15, 2012 14:03

Quote:

Originally Posted by doronzo (Post 349683)
it indeed depends on the goals, but in general a particle concentration like that isn't high. also, the number of tracked particles depends a lot on the grid, because they are treated as cluster of particles, so I don't expect you would have a prohibitive number of trajectories.

in a first approximation you could consider your small particles as fully coupled to the fluid, and treat the mixture as multispecies (Fluent multispecies model), i.e. air (species 1) + pseudofluid (species 2, with its properties modified by the particles, e.g. increased density and viscosity; Marble 1970, Annu. Rev. Fluid Mech.).

in this second case, you could directly follow the species mass fraction, but of course you cannot have the particles leave the fluid transport system

Thanks. This is definitely a good approach, but I need to sort it out. For example, I do need to remove some particles from the system due to deposition. It physically does happen during the transport over several hundred meters.
As I have already done quite some work based on the mixture model, I would like to figure out how to solve mass fraction of the secondary granular phases. I don't mind creating UDFs either.

doronzo March 15, 2012 14:37

I'm not too much familiar with the mixture model, but I think it's basically similar to a pseudofluid approach, with the difference it solves for the relative velocities, as well as the volume fraction of the secondary phase. This is done with some algebric expressions for the velocities, and a volume fraction equation for the secondary phase

lihuang March 16, 2012 10:54

Quote:

Originally Posted by doronzo (Post 349674)
...if the particles enter the domain with the same, or similar, velocity.

why not to use the DPM (discrete phase model)? has your grid a high resolution?

regards

I checked the FLUENT theory guide again, and it says DPM is not compatible with multiphase species transport.

lihuang March 26, 2012 14:59

I found what causes this is probably i am using too low volume concentration of the secondary phase. In my case I have to deal with volume fractions as low as 1e-20 or lower. But it seems FLUENT double precision solver works fine on mass fraction for secondary phase volume fractions greater then 1e-08. Anything smaller than that becomes 0. :(

Any idea to work around it?


All times are GMT -4. The time now is 21:42.