CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (https://www.cfd-online.com/Forums/openfoam-solving/)

 lam July 30, 2007 12:25

Hi, As I have to make a lit

Hi,

As I have to make a little report about the way Foam use to solve the Navier-Stokes, I have some questions to make it clear:

1/ In the SIMPLE algo, why phi is calculated before the correction of U? If phi is calulated after the correction of U, isn't better?

2/i/ I understand that the coefficient aP and aN depend on U. BUT how are they guessed a the FIRST iteration in the SIMPLE algo???? I can not imagine the A matrix actually!
ii/ And the most important: what " /aP" means???
I mean HOW CAN WE DIVIDE by aP which is not a scalar (if I'm not wrong). What does it mean?

3/ To confirm: the previous U is just needed in
the relaxation of UEqn, isn't it?

Any clues or advice, or attempts of answer will be welcome. (and Hrv's reply will be more than appreciated)

Lam

 hjasak July 30, 2007 12:41

OK, OK. - At the stage wher

OK, OK.

- At the stage where you're looking phi contains only a part of the interpolated velocity dotted with the face area vector. Have a look at phi -= pEqn.flux(); - this is where the rest of the flux comes from and it's after the pressure solution. All is well

- aP is the central coefficient of the momentum matrix, packed up as a field: one value for each cell. That would be a scalar if your discretisation is easy; once you get more clever (than the current implementation), aP will be a 3x3 tensor and division will be a multiplication with the matrix inverse. Advanced, though - for easy stuff have a look at my Thesis

- FIRST SIMPLE iteration: well, you guess U as well, right? That is called the initial guess.

- to confirm: no. You also have it in UEqn.H() - read the forum again, I've explained this.

Hrv

-

 All times are GMT -4. The time now is 20:48.