CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (http://www.cfd-online.com/Forums/openfoam-solving/)
-   -   Steady spray simulation (http://www.cfd-online.com/Forums/openfoam-solving/58543-steady-spray-simulation.html)

didomenico September 17, 2008 12:07

Dear Foamers, I would like
 
Dear Foamers,

I would like to implement a steady version of dieselFoam (as library), and to couple it to a steady solver (like simpleFoam). Main assumptions I am going to take are:

- each particle path has to be integrated until it goes out of the domain or evaporates completely. This means that the integration time has to be determined from the library itself and not inherited from the flow solver (as it is currently done in dieselFoam, I suppose)

- droplet-droplet interaction and wall treatment are discarded (at least in this phase).

My questions are:

- can the dieselFoam library be coupled directly to a steady solver or does it need a physical timestep to work? I have seen several calls to runTime(), therefore I suppose that such a (mis)use has not been foreseen...

- if not, should I start from the lagrangian library or can I re-use most of dieselFoam routines? I am trying to figure out where to start, but a hint would be appreciated...

Thanks,
Max

stephan September 17, 2008 12:23

Hi Max, i am pretty confuse
 
Hi Max,

i am pretty confused: the whole process is pretty unsteady. when you speak about "integrated until it goes out of the domain" - dont you speak about intergration over are finite timestep?

regards
stephan
p.s. dieselFoam library does need a true timestep.

didomenico September 17, 2008 12:40

Dear Stephan, yes, you are
 
Dear Stephan,

yes, you are right, you have to perform the integration over physical times steps, but you can do it regardless to the actual time in your flow solver.
What I have in mind can be described as follows:
for(1, numberCouplingSteps)
<blockquote>for(1, numberOfFlowSolverIters || isSimpleConverged())
<blockquote>doSimpleStep(U, p, T, Y, spraySources);
endfor

doSprayIntegration(U, p, T, Y);

spraySources=getSpraySources();

spraySources.relax();</blockquote>
endfor</blockquote>

where

- doSprayIntegration() should perform particle integration as I have explained above (until particles disappear)

- numberCouplingSteps is the number of spray-flow iterations, in my experience few (from 3 to 10, according to the testcase) should be sufficient

- spraySources implement the coupling between phases and they enter in the solution of U, p, T and species equations. A relaxation could be necessary, in case of strong evaporation and/or combustion.

I know all limitation of such an algorithm, but I have not seen any example of steady spray with OpenFOAM.

Regards,
Max

stephan September 17, 2008 14:24

Hi Max, i am still confused
 
Hi Max,

i am still confused but anyway: the calls of runtime in the dieselspray library are as far i remember only to get the actual timestep.
if you set something appropriate for your simulation it should work the way you want.
regards
stephan
btw: what about injection and the corresponding timeperiod?

didomenico September 18, 2008 03:07

btw: what about injection and
 
Quote:

btw: what about injection and the corresponding timeperiod?
This is another problem I have to face, since I would like to set a constant mass flow rate and not a constant amount of mass.

Actually I am realising that I have to rewrite the whole stuff, but I may re-use many things from the dieselFoam library...

Regards,
Max

Rachel June 24, 2009 10:27

Dear all,

I am also interested in a steady state dieselFoam solver. In my case there is a constant mass flow rate & hence the idea of steady dieselFoam.

Can anybody please guide me, how to proceed with the task of stead dieselFoam?

Max, did you have success with steady state implementation?

ALC August 14, 2010 16:37

Steady Spray Follow-up
 
I realize this is an old thread, but I wanted to see if anyone has made any progress on developing a steady-state spray solver with combustion. I know that Fluent has this capability in its Discrete Phase Model, so I assume it is not too crazy of an idea.

Any information would be much appreciated.

Thanks.

Adam


All times are GMT -4. The time now is 11:50.