[sharonyue]

Hi,
This is icoFoam's code,:

fvVectorMatrix UEqn
        (
            fvm::ddt(U)
          + fvm::div(phi, U)
          - fvm::laplacian(nu, U)
        );

        solve(UEqn == -fvc::grad(p));//this term is only executed when the momentumPredictor flag is set to true

        // --- PISO loop

        for (int corr=0; corr<nCorr; corr++)
        {
            volScalarField rAU(1.0/UEqn.A());

            volVectorField HbyA("HbyA", U);
            HbyA = rAU*UEqn.H();
            surfaceScalarField phiHbyA
            (
                "phiHbyA",
                (fvc::interpolate(HbyA) & mesh.Sf())
              + fvc::ddtPhiCorr(rAU, U, phi)
            );

            adjustPhi(phiHbyA, U, p);

            for (int nonOrth=0; nonOrth<=nNonOrthCorr; nonOrth++)
            {
                fvScalarMatrix pEqn
                (
                    fvm::laplacian(rAU, p) == fvc::div(phiHbyA)
                );

My confusion is if I set the momentum predictor is off, which line is to solve U?

[santiagomarquezd]

Hi, in the PISO method, U is predicted and then corrected in the pressure-velocity coupling loop, so that, if you don't predict the velocity, the velocity used to start the coupling loop is a derivative of the velocity in the last time step as is done in interFoam and other solvers.

In the code you've cited the matrix for the momentum equation is assembled so that you have the A and H methods available for the use in the PISO loop, the line:

HbyA = rAU*UEqn.H();

calculates a first approximation for U using the H operator applied to the velocity of the last-timestep or which was predicted when you solved the UEqn (if you did it) and the A operator. If you need it, please check the definition of these operators in chapter 2.7 of my Ph.D. thesis

The link to my PhD thesis is provided in my user page at openfoamwiki.net: http://openfoamwiki.net/index.php/User:Santiagomarquezd

and Jasak's Ph.D. and other's thesis.

Regards.