[waku2005]

Dear all

I'm reading source files of the BuoyantBoussinesqPisoFoam and two questions arose as below.
(Please refer at http://openfoamwiki.net/index.php/Bu...sinesqPisoFoam)

1) What is done in the TEqn.relax() before TEqn.solve() ?
I recognized that implicitly discretized temerature equation was solved
in the TEqn.solve(). Is the TEqn.relax() a preparation for the solver ???

2) How effects the fvc::ddtPhiCorr in the phiU calculation of corrector loop
to the flux phi ? (In other sources, it is commented out sometimes.)

Any comments will be welcomed ....

Best regards,
waku2005

[fabian_roesler]

Hi

the TEqn.relax() is used in SIMPLE solvers for underrelaxation. Since you use PISO you must not underrelax. And typicale you give no relaxation factors in fvSolution dictinary. So TEqn.relax() in PISO has no meaning.

Regards Fabian

[waku2005]

Hi fabian

Thanks to your comment.
I know that PISO scheme has a greater stability than other schemes like SMAC method, but in larger Co number cases such as Co=1～5, PISO is applicable or not ?

To tell the truth, curently I simulate a strong buoyancy Boussinesq unsteady flow with longwave and showtwave (SUN) radiations in a enclosure (a room with windows) using my own SMAC fortran code.
Time step size should be too small of 0.002 s for stablity, although total simulation time is 2 hours or so.
Thus I consider to use OpenFOAM.

Best regards
waku2005

Originally Posted by fabian_roesler

Hi

the TEqn.relax() is used in SIMPLE solvers for underrelaxation. Since you use PISO you must not underrelax. And typicale you give no relaxation factors in fvSolution dictinary. So TEqn.relax() in PISO has no meaning.

Regards Fabian

[waku2005]

Sorry I replied twice and deleted the second one.

[fabian_roesler]

Dear waku2005

when using PISO you should keep the courant number below 1. PISO does only one outer calculation step by calculating U and then correcting p in several corrector steps.
In OF 1.7.1 the buoyancy solver uses the PIMPLE algorithm which has better stability when using larger time steps. PIMPLE performs a PISO corrector loop which is repeated several times, starting with the results of the previous corrector step. Thus you are allowed to under relax the solver. The solver is faster and more stabel. Have a look here:

http://www.cfd-online.com/Forums/ope...algorithm.html

Regards Fabian

[waku2005]

Dear Fabian

Thank you very much for reply.
I'll check and try it.

Best regards,
waku2005

Originally Posted by fabian_roesler

Dear waku2005

when using PISO you should keep the courant number below 1. PISO does only one outer calculation step by calculating U and then correcting p in several corrector steps.
In OF 1.7.1 the buoyancy solver uses the PIMPLE algorithm which has better stability when using larger time steps. PIMPLE performs a PISO corrector loop which is repeated several times, starting with the results of the previous corrector step. Thus you are allowed to under relax the solver. The solver is faster and more stabel. Have a look here:

http://www.cfd-online.com/Forums/ope...algorithm.html

Regards Fabian

[feldy77]

Dear all,
Does anybody body can explain/guide/give a ready file/ for calculation of total heat transfer from a given surface by buoyantBoussinesqPimple solver. Until know I tried to do post processing in fluent and the values I got are not exact enough. Namely, I simulate a transient buoyant flow in the spherical gap. For a very low Ra number, let say 100 I perform time integration for a long time to obtain a steady solution, which should exist for such low Ra number. Then when I compare heat fluxes from external and internal surfaces in fluent they are not equal, differs by 10% which is not acceptable for such simple geometry. So I think that there are discrepancies when transforming from OpenfFoam to fluent and I want to be able to compute it directly by openFoam. Thank you very much for help