[nikhil108]

Hallo Foamers,

I have a small doudt. In UEqn.H, in solvers where momentum predictor setting exists in pimple loop. If momentum predictor is set to no then the Right-hand side of the momentum equation is not solved (skipped because of if loop). But, I dont understand, in any case, we need to solve whole momentum equation to get U right? . Why we skip gravity and dp term ? what does this mean

     tmp<fvVectorMatrix> tUEqn
     (
         fvm::ddt(rho, U) + fvm::div(phi, U)
       + MRF.DDt(rho, U)
       + turbulence->divDevTau(U)
      ==
         fvOptions(rho, U)
     );
     fvVectorMatrix& UEqn = tUEqn.ref();
 
     UEqn.relax();
 
     fvOptions.constrain(UEqn);
 
// In here

     if (pimple.momentumPredictor())
     {
         solve
         (
             UEqn
          ==
             fvc::reconstruct
             (
                 (
                   - ghf*fvc::snGrad(rho)
                   - fvc::snGrad(p_rgh)
                 )*mesh.magSf()
             )
         );
 
         fvOptions.correct(U);
         K = 0.5*magSqr(U);
     }

[Cyp]

Dear Nikhil,



When you turn off momentumPrediction, you do no solve the momentum equation.



If you look at the predictor-corrector solution algorithms used in CFD (PISO, SIMPLE, PIMPLE...), basically you do the following sequence:


- Discretize the momentum equation and form a matrix for U (optionnaly inverse this matrix to have a first guess of U)

- Form a matrix for p. This matrix depends on the coefficients of the matrix for U. Inverse the matrix for p.

- Correct the value of U point-wised based on the updated pressure field.
- Iterate until you converge.
- March in time



So, when you turn off momentumPrediction, you do update the coefficients in the matrix for U, but you do not inverse this matrix.


Cheers,
Cyprien

[nikhil108]

Hallo Cyprien,

Thanks for the explanation. Yah! now it's crystal clear.

cheers,
nick.