Problems about setReference() and solve PPE

chaochaoyoyo · April 24, 2016, 10:23

Hi everyone,

I am now solving a Rayleigh-Bernard problem by my own solver (SIMPLER). The boundary condition for dynamic pressure is zeroGradient. But the difficulty in solving Pressure Poisson Eqaution(PPE) hindered me for a long time.

1. I have tried to use setReference(). The solution is converged quickly. But it is obvious the reference point has a bad influence on the solution. The solution changes drastically as the reference point changes. I can't figure out the reason. Could any one tell me the skills in set reference pressure to make the solution accurate?

2. The other method I want to implement is to change continuity equation as:

$\frac{1}{\beta }\frac{{\partial {p^k} - {p^{k - 1}}}}{{\partial t}} + \frac{{\partial \rho }}{{\partial t}} + \nabla \cdot (\rho \vec u) = 0$

By doing this, the PPE is more diagonally dominant. And when the solution of PPE converged, $p_k=p_{k-1}$ , the continuity equation become the right one. This can be called an equivalent artificial compressibility method.

But I don't know how this method can be implemented in Openfoam. Becasue it requires to change the RHS of PPE during the solve() process (the fvc::ddt(1/beta,p) should be changed after each iteration). Could any one tells me how can I realize this method in OpenFOAM? Do I have to rewrite solve() function?

Any suggestions will be appreciated.

Zhanchao

Tobi · April 24, 2016, 13:30

Hello,

I am not so familiar with your method (reference would be nice). If you want to change the pressure equation, just do it:

Code:

            fvScalarMatrix pEqn
            (
                fvm::div(phid, p)
              - fvm::laplacian(rho*rAU, p)
              ==
                fvOptions(psi, p, rho.name())
            );

You should also be able to do this using the fvOptions feature of openFoam. But with the pressure / velocity coupling I am not familiar - sorry.

chaochaoyoyo · April 24, 2016, 21:55

Quote:

Originally Posted by Tobi

Hello,

I am not so familiar with your method (reference would be nice). If you want to change the pressure equation, just do it:

Code:

            fvScalarMatrix pEqn
            (
                fvm::div(phid, p)
              - fvm::laplacian(rho*rAU, p)
              ==
                fvOptions(psi, p, rho.name())
            );

You should also be able to do this using the fvOptions feature of openFoam. But with the pressure / velocity coupling I am not familiar - sorry.

Hi Tobi,

What you suggest is very helpful. Thank you.

I am reading some materials about fvOptions(). Do you know whether or not the fvOptions() function modify the source term during pEqn.solve() process? If it does, this is what I am looking for!

Best regards,
Zhanchao

Tobi · April 25, 2016, 02:47

Hello,

the function solve() only solves your matrix that you build before. The matrix you want to solve is that one

Code:

fvScalarMatrix pEqn
(

);

If you modify this equation, you will also solve a modified pressure equation. So if you set a source in the fvOptions for the pEqn, it will be replaced in the term on the RHS named fvOptions(psi, p, rho.name()).

April 24, 2016, 10:23	Problems about setReference() and solve PPE	#1
chaochaoyoyo New Member Zhanchao Hu Join Date: Apr 2016 Posts: 8 Rep Power: 10	Hi everyone, I am now solving a Rayleigh-Bernard problem by my own solver (SIMPLER). The boundary condition for dynamic pressure is zeroGradient. But the difficulty in solving Pressure Poisson Eqaution(PPE) hindered me for a long time. 1. I have tried to use setReference(). The solution is converged quickly. But it is obvious the reference point has a bad influence on the solution. The solution changes drastically as the reference point changes. I can't figure out the reason. Could any one tell me the skills in set reference pressure to make the solution accurate? 2. The other method I want to implement is to change continuity equation as: $\frac{1}{\beta }\frac{{\partial {p^k} - {p^{k - 1}}}}{{\partial t}} + \frac{{\partial \rho }}{{\partial t}} + \nabla \cdot (\rho \vec u) = 0$ By doing this, the PPE is more diagonally dominant. And when the solution of PPE converged, $p_k=p_{k-1}$ , the continuity equation become the right one. This can be called an equivalent artificial compressibility method. But I don't know how this method can be implemented in Openfoam. Becasue it requires to change the RHS of PPE during the solve() process (the fvc::ddt(1/beta,p) should be changed after each iteration). Could any one tells me how can I realize this method in OpenFOAM? Do I have to rewrite solve() function? Any suggestions will be appreciated. Zhanchao Last edited by chaochaoyoyo; April 25, 2016 at 01:21.

April 24, 2016, 13:30		#2
Tobi Super Moderator Tobias Holzmann Join Date: Oct 2010 Location: Tussenhausen Posts: 2,708 Blog Entries: 6 Rep Power: 51	Hello, I am not so familiar with your method (reference would be nice). If you want to change the pressure equation, just do it: Code: fvScalarMatrix pEqn ( fvm::div(phid, p) - fvm::laplacian(rho*rAU, p) == fvOptions(psi, p, rho.name()) ); You should also be able to do this using the fvOptions feature of openFoam. But with the pressure / velocity coupling I am not familiar - sorry. __________________ Keep foaming, Tobias Holzmann

April 25, 2016, 02:47		#4
Tobi Super Moderator Tobias Holzmann Join Date: Oct 2010 Location: Tussenhausen Posts: 2,708 Blog Entries: 6 Rep Power: 51	Hello, the function solve() only solves your matrix that you build before. The matrix you want to solve is that one Code: fvScalarMatrix pEqn ( ); If you modify this equation, you will also solve a modified pressure equation. So if you set a source in the fvOptions for the pEqn, it will be replaced in the term on the RHS named fvOptions(psi, p, rho.name()). __________________ Keep foaming, Tobias Holzmann