[alberto]

Why should it be better with 1e-15 instead than 1e-4 from a numerical point of view? This limitation is useful when alpha goes to zero, because you divide by it, so 1e-4 should have a better "stabilizing effect".

Moreover, you can set the inferior limit of alpha, in the case of a particle flow to something under 1e-4 (1e-6 is what is used in the kinetic theory framework, and it's conservative) without problems from a physical point of view, because the flow is so dilute that you can't consider the two-fluid equations valid in those conditions, and, as a consequence, you should not use them anyway.

Regards,
Alberto

[juho]

What I meant is that max(alpha,anumber) only affects alpha values smaller than anumber (clips the scale at the bottom) while (alpha+anumber) affects all the alpha values. (moves the whole scale upwards)

To me, the clipping seems like a more elegant solution. Assuming the max() function doesn't introduce new problems - like the dimension mentioned by Rasmus.

Thank you all very much for your responses on the whole issue!

[alberto]

Hello,

yes, sorry. I was not discussing about clipping. I agree with you on that.

I was just pointing out that you can happily clip alpha at a numerically safe value around (like 10^-6) instead than to a very small value, that in a division would return a huge coefficient.

Regards,
Alberto

[henry]

Hi Rasmus,

I have implemented your changes in our 1.5 release candidate and all seems well.

Thanks

Henry

[hemph]

Thanks Henry!

[danielle]

I think that in "liftDragCoeffs.H" file for blended case
K should be calculated as:
K = (alpha*Ka + beta*Kb);
not:
K = (beta*Ka + alpha*Kb);

[danielle]

are you agree with me !!!

[niklas]

assume phase a to be the discrete one and b to be the continous.

standard approach would be to calculate the drag as K = Ka, right?

alpha = 0 -> pure gas.

so...
as alpha -> 0 we should have K -> Ka

but with your approach K -> Kb

so no, i dont agree.

[Prash]

Hey Guys,

I think I am on the right thread. But I am trying to write a new drag law for my calculations.

I found this :

http://openfoamwiki.net/index.php/Ad...lerFoam_solver

I really cannot understand why -


The Syamlal and O'Brien drag model can be implemented in OpenFOAM by dividing the β' < math > bytheproductofthetwovolumefractions < math > αβ to obtain K.
N.B. Every drag formulation has to be divided by the product αβ before beeing implemented in twoPhaseEulerFoam because this product has been extracted for numerical reasons.



Why do we devide by the product of voulme fractions and where in the source code do we multiply them back ???





Please help me with this.





Best Wishes

Prashant

[alberto]

Hi Prashant,

you divide by the product of the phase fractions because of how the momentum equation is treated (phase-intensive form). You might want to check Henrik Rusche's Ph.D. thesis for some more detail on the implementation.

P.S. Notice that all the models listed there are already part of OpenFOAM now.

Best,

[Prash]

Hey,

Thank you very much for useful info, I had another question on calculations and use of solid volume fractions for different species. I am trying to do a segregation study on fluidized bed , hence I require total volume fraction(alpha) and volume fraction of each of two species of particles ( different sizes).

I am not sure where to look for them. I am trying to access the species volume fraction in order to calculate drag on individual species.

Waiting eagerly for a reply

Best Wishes
Prashant

[alberto]

The code deals with mono-disperse flows. If you need more than one dispersed phase, you'll have to extend it.

Best,

[pramod308k]

hi,
I want to simulate binary fluidized bed problem with OpenFoam, In my case more than one discrete phase are present. My question is can I use twoPhaseEulerFoam for more than one discrete phase.???? If yes than please give me some hint.

thank you
Pramod Kumar

[alberto]

Hi Pramod,

twoPhaseEulerFoam can deal with one particle phase.

Best,

[pramod308k]

hi Alberto,
can you please tell me how to set a case for more than one disperse phase in openFoam. I'm very new with this software.

thank you
Pramod

[alberto]

There is no Euler-Euler multi-fluid solver in OF at the moment.