[simone1981]

Hi,
I'm working with compressibleTwoPhaseEulerFoam, I have a few questions about the continuity equation in alphaEqn.H. Consider the continuity equation as written, for example, in the thesis of Ghione 'Development and validation of a two phase CFD model using OpenFoam', eq. 2.11.

- In alphaEqn.H the are two source terms Sp and Su. Considering the equation in 2.11 I expect someting like :

Sp + Su = -alpha1 div(U) + alpha2 alpha1[1/rho1 Drho1/dt + 1/rho2 Drho2/dt]

In alphaEqn.H Sp and Su are written as a function of dgdt, where in CreateFields.H

volScalarField dgdt
(
pos(alpha2)*fvc::div(phi)/max(alpha2, scalar(0.0001))
);

while in PEqn.H

dgdt =
(
pos(alpha2)*(pEqnComp2 & p)/rho2
- pos(alpha1)*(pEqnComp1 & p)/rho1
);

If I combine Sp+Su, in the first case I obtaine (this should be the case at the first call inside the pimple loop!):

alpha1 div(U) + div(U)

1) What does it mean? In the continuity equation I do not see this term.

in the second case, if I consider the formulation commented in alphaEqn. H ('Legacy semi-implicit and potentially unbounded...')

// Legacy semi-implicit and potentially unbounded form
fvScalarMatrix alpha1Eqn
(
fvm::ddt(alpha1)
+ fvm::div(phic, alpha1, alphaScheme)
+ fvm::div
(
-fvc::flux(-phir, alpha2, alpharScheme),
alpha1,
alpharScheme
)
==
fvm::Sp(Sp, alpha1) + Su
);


and I do not consider & p in dgdt, I obtain the term

-alpha1 div(U) + alpha2 alpha1[1/rho1 Drho1/dt + 1/rho2 Drho2/dt]

that fits with the equation 2.11.

2) Why & p in dgdt? Is it the inner product, right? Considering 2.11 I do not need this product.

3) How does the uncommented formulation work? I do not see something like Sp(Sp, alpha1) + Su, but only Sp + Su.

Thank you in advance

Simone

[sharonyue]

Ciao simone!

Point 2 is because that pEqnComp1 is defined as tmp<fvScalarMatrix>, u can use pEqnComp1 to call matrix, also u can use pEqnComp1 & p to call some real value.

I mean we define U matrix like this:

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

If we use UEqn & U, we will have the value of U instead of matrix.

This is my thought I'm not sure. lol

Best,