[fshak92]

Hi FOAMers,

I'm implementing an explicit 3rd order Runge-Kutta scheme(for temporal discretization) to solve incompressible Navier-Stokes eq.

---------------
These are main steps solved 3 times in each time step (the complete steps can be found in the attachment):
a) Solve momentum eq. explicitly. U' is obtained.
b) Laplacian(p') = div(U') / (alpha+beta)*DeltaT
c) Velocity is corrected using p'
d) Pressure is corrected by : p=p+p'
----------------
Implementing the steps above in OpenFOAM (simple cavity test case, and comparing with icoFoam) provides me correct result for the velocity field, however the pressure solver is not converged (even after 1000iterations), consequently i don't get any reasonable value for pressure.(Two different algebraic solvers also have been tested for pressure).

If i remove step (d), still i get correct results for velocity, but the pressure solver now is converged. Also, i get reasonable result for pressure distribution (different with icoFoam, but same distribution).

This is the line for solving the pressure eq:

solve( fvm::laplacian(p1) == fvc::div(U)/((alpha+beta)*runTime.deltaT()) );

alpha and beta are some coefficients.

Does anybody know what can be the problem?
I would appreciate any hint.

[chriss85]

You might want to look into this thread:
http://www.cfd-online.com/Forums/ope...nge-kutta.html

[fshak92]

Originally Posted by chriss85

You might want to look into this thread:
http://www.cfd-online.com/Forums/ope...nge-kutta.html

Thank you for the link, however i'm trying to write incompressible solver, which requires solving the Poisson equation for pressure. Unfortunately, only compressible solvers(same as your link) can be found on the net, in which the pressure equation is not solved.

I've already found the problem, which lies on the accumulation of pressure by increasing the time step. If i set the pressure to zero before each time step, the pressure solver is converged, and i get nearly reasonable results. But still i wonder if setting pressure to zero is the correct way!

[chriss85]

I have no experience with incompressible solvers unfortunately.

[syavash]

Originally Posted by fshak92

Thank you for the link, however i'm trying to write incompressible solver, which requires solving the Poisson equation for pressure. Unfortunately, only compressible solvers(same as your link) can be found on the net, in which the pressure equation is not solved.

I've already found the problem, which lies on the accumulation of pressure by increasing the time step. If i set the pressure to zero before each time step, the pressure solver is converged, and i get nearly reasonable results. But still i wonder if setting pressure to zero is the correct way!

Hi,

Could you finish developing your solver?!
I am working on a LES simulation and I think a runge-kutta solver is preffered compared to an implicit solver like pisoFoam.
Can you share it with me?!

Thanks,
Syavash