[lam]

Hi all,

I want to use an other way to solve a steady Stokes equation for incompressible flow:

div(U)=0
laplacian(nu,U)=Grad(p) (2)

by

div(U)=epsi*p, where epsi is very small
i put this expression in (2), and I have:

Laplacian(nu,U)=grad(div(U)/epsi) (2')

But my problem is I have not managed to discretize "grad(div(U)/epsi)"

I've tried :

tmp<fvvectormatrix> UEqn (fvm::laplacian(nuEff(), U));
solve(UEqn()==fvc::grad(div(U)));


But it tells me someting like in compilation:
"
error: no matching function for call to 'div(Foam::volVectorField&)
"

I think it a intressing way to test, even if it could be expensive for the memory.

Hope taht anybody will give me some advice to make a good discretisation for (2')

Thanks in advance,

Lam

[lam]

nobody has an idea?

[lam]

I have tried this:

tmp<fvvectormatrix> UEqn (turbulence->divR(U));
solve(UEqn()==-fvc::grad((fvc::div(U))/epsi));
where epsi is a dimensioned scalar [0 -2 1 0 0 0 0]

But my solution doesn't obey the continuity!

Can somebody tell me where I am wrong?