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/)
-   -   MassMomentum sources (http://www.cfd-online.com/Forums/openfoam-solving/60447-massmomentum-sources.html)

rbw September 26, 2005 13:30

Is there a way to setup mass/m
 
Is there a way to setup mass/momentum point sources? Alternatively, can boundary conditions be assigned to single nodes?

I'm asking because I have a very high speed jet flowing into a very large volume. If I model the inlet as a real inlet, I have to use very small elements (obviously), therefore I have to use itsy bitsy teeny weeny time steps to achieve stability (since there is no local time stepping, as I understand). We are talking element edge sizes from 0.020m to 10m here. (Obviously the mesh is nicely graded.)

If I'm just interested in the global features of the flow, this is not so good, because I have to wait for thousands and thousands of time steps just for the flow information to propagate throughout the domain. I think I could get an adequate solution much faster if I could use a more uniform (with larger elements) mesh.

Would it work to use a non-uniform input profile, with all nodes but one with the velocity I want?

Many thanks!

olivier November 1, 2005 11:46

Hi, I want to solve an equati
 
Hi,
I want to solve an equation of momentum with an exterior force F (F doesn't depend on U, it is an volVectorField).

should I make F like a source term (with fvm::Su)?
because when I make :

tmp<fvvectormatrix> UEqn
(
fvm::div(phi, U)
- fvm::laplacian(mu,U)
);
UEqn().relax();

solve(UEqn() == -fvc::grad(p))-fvm::Su(F,U) )
I have a bad result.
thanks for your help

niklas November 1, 2005 15:20

no, you just add -F solve(UEq
 
no, you just add -F
solve(UEqn() == -fvc::grad(p))-F)

did you try that?


All times are GMT -4. The time now is 20:43.