Hi all

I am trying to implement this method with FVM, in an unstructured cell centered grid, for a 2D viscous fluid. The inviscid flux is calculated with the Roe Method.

The problem is that my code doesn't converge, even when in the first iterations(5000) the residue seems to get smaller. Any idea?

Thanks in advance,

Ruben

Hi,

Have a look at NSCKE it might help.

Regards

A.S.

Sorry but where can I find this?

Hi,

1) Can you please describe your resolution method ?

2) what is the value of the artificial compressibility parameter you choose ?

Regards

Buch

Hi,

1) Can you please describe your resolution method ?

I am trying to solve the dimensionless N-S Eq. for an incompressible flow using the Artificial Compressibility concept with an triangular unstructured grid. In order to discretizate this Eqs. I used the Finite Volume method, and I get the following system:

dq/dt + Suma(E-Ev)

where E is the inviscid flux normal to the face Ev is the viscous flux normal to the face q the vector with (p,u,v) variables

The inviscid flux is calculated using the Roe method The viscous flux is computed by applying Gauss Theorem For the time integration it is used four-stage Runge-Kutta.

Boundary Conditions: I use ghost cells inflow : u=1.0 v=0.0 Pghost= P neig outflow: Ughost= Uneig , Vghost= V neig, P = 0.0 wall : Ughost= -Uneig , Vghost= -V neig, Pghost= Pneig

2) what is the value of the artificial compressibility parameter you choose ?

First, I tried to implement a formula proposed by Hino, Betha = (u2+v2), but now I am trying to keep this parameter constant.

Thanks in advance,

Ruben

in Roe scheme, Q_L and Q_R at the cell face have to be reconstructed. It may be the ways of reconstruction that affect the convergence. How do you obtain Q_L and Q_R?

QL is just the value of the neighbour cell and QR is the cell analized. is that right?