CFD Online Discussion Forums

CFD Online Discussion Forums (
-   OpenFOAM Running, Solving & CFD (
-   -   Solving systems of equations (

Marco Kupiainen (Kupiainen) February 2, 2005 08:14

Hi, I'm wondering how one ca
I'm wondering how one can solve systems of equations in OpenFOAM. I am not interested in a segregated solver of the PISO, SIMLE type, since I have source terms.
Does one have to construct a vector containing the unknowns, e.g. U=(rho, rho*u, rho*v, rho*w, rho*E) for the NS-equations. And then do the time integration with some scheme, e.g. Runge-Kutta or similar. Will there be problems with the physical units? What about the operators, div, grad, etc. can they handle this?
What other difficulties can you foresee? Is this a feasible way at all with OpenFOAM?
best regards

Eugene de Villiers (Eugene) February 2, 2005 13:28

Why cant you use a segregated
Why cant you use a segregated solver with source terms like the rest of us? OpenFOAM does not at present incorporate a block solver, so you are limited to a segregated approach.

To see how equations are defined and solved, please look at any solver's top level code, e.g. RANS based incompressible solver, OpenFOAM-1.0.2/applications/solvers/incompressible/turbFoam/turbFoam.C

Hrvoje Jasak (Hjasak) February 2, 2005 13:31

For this case, the segregated
For this case, the segregated solver is not appropriate - you get killed in cross-equation coupling because source-coupling is very strong. We basically need a block solver and the sooner the better.

If you try to solve this segregated, you will end up violating the entrophy condition because of lagged source interaction.


Henry Weller (Henry) February 2, 2005 13:55

We have very successful segre
We have very successful segregated solvers for this kind of problem, it's just a matter of choosing the best variables to solve for and being consistent with the discretisation and boundary conditions.

All times are GMT -4. The time now is 19:18.