[Tscar]

Hello,

While looking at rhoPimpleFoam solver, I could not find where the pressure equation is solved.

The solve() command is only executed in the non-orthogonal corrector loop, yet the pressure equation is off course solved even when there are no non-orthogonal correctors.

Can you please help me understand how and where the pressure equation is solved?

Thank you!

[Tobi]

The NonOtho value is still 1, even if you set zero.

[Tscar]

Thank you for your response.

I found out later yesterday and I meant to update my question .. but you were quicker.

What I do not understand now is the pimple.finalInnerIter() in pEqn.solve(mesh.solver(p.select(pimple.finalInnerIter()))) in the correctNonOrthogonal loop.

The pimpleControl.H says that pimple.finalInnerIter() returns true for final inner iteration. Does this mean that the pEqn gets solved only on final inner iteration?

[Tobi]

That is simple. Maybe you realized that you have the dictionaries:

p
{
     solver    PCG;
     .
     .
}

pFinal
{
     solver    PCG;
     .
     .
}

in the fvSolutions file. You always solve the pressure equation:

pEqn.solve(mesh.solver(p.select(pimple.finalInnerIter())));

However, you can see that the solve() function has an argument. The argument is:

mesh.solver(p.select(pimple.finalInnerIter()))

This will return the p or pFinal dictionary and is related to the dictionaries in the fvSolutions file. You can find that using doxygen and check out the fvMatrix class:

SolverPerformance< Type > solve (const dictionary &) 
Solve segregated or coupled returning the solution statistics.  More...

For what purpose? In the final PISO loop we have to make sure that the continuity equation is fulfilled. Therefore we are using more strict settings given in pFinal (common is to set the relTol to zero). This is very important if we have scalar transport equations which really need a fulfilled continuity. If we only solve U and p, one can also choose the final PISO iteration to be less strict based on the coupling of p and U.

[Tscar]

Thank you!

By the way, I am a huge fan of your work. You are doing a great job.

[Tscar]

Can I have one more question?

In chtMultiRegionFoam, the pimple loop is introduced in the form:

for (int oCorr=0; oCorr<nOuterCorr; oCorr++)

Why is it not the following statement instead?

while (pimple.loop())

I believe it has something to do with the multiple "PIMPLE" dictionaries existing in ./system and ./system/regionNames.

How can I make the code understand stuff like pimple.loop() while making it search in the right "PIMPLE" dictionary?