[gaza]

Hi
I found that my solution depends on writeInterval entry in the controlDict.
How is it possible?

[alexeym]

Hi,

Since question lacks lots of details, answer is in fact question: does your time step value depend on writeInterval?

[gaza]

Hi alexym,
I am using modified interPhaseChangeFoam from OpenFOAM extend 3.0
I set adjustTimeStep to "no" and the results also depend on the writeInterval entry.
Namely, the interface position varies significantly when I decrease writeInterval,
for example from 0.1 to 0.01. Is it bug?

[gaza]

what can be the reason? I do not understand.
In the code I have temperature equation with term fvm::ddt(rho, T)
and I used rho.oldTimes() method in createFields.H for rho. Does rho.oldTimes() method
store the previous iteration or previous time step values?

[alexeym]

Hi,

1. Does you solution depend on time step?
2. What modifications did you make?
3. What convergence criterion do you utilize (if any)?

Yes, it could be bug in:
1. your solver modification
2. your case setup
3. original solver

In general oldTime stores noting: http://cpp.openfoam.org/v4/a00931.ht...09ca405ce907a6 (though in foam-extend this could be different). There are storePrevIter (http://cpp.openfoam.org/v4/a00931.ht...7c72a819c66706) and storeOldTimes (http://cpp.openfoam.org/v4/a00931.ht...9ee55d9138fd4d) if you want to store things.

[gaza]

Hi alexeym,
Thank you for your help. I found that it is bug in my code but for now I do not know where

[gaza]

I used in the code this line to relax TEqn
TEqn.relax(0.9);

is it the same if I have in the code
TEqn.relax();
and in the fvSolution dictionary:
relaxationFactors
{
equations
{
T 0.9;
}
}

??

[alexeym]

Since you relax unconditionally, it corresponds to

relaxationFactors
{
    equations
    {
        "T.*" 0.9;
    }
}

[gaza]

Hi alexeym,
I commented the line with TEqn.relax() and now the results almost does not depend on writeInterval entry. For instance I got liquid volume fraction after 10 s as:

VOF            writeInterval
0.513542    1.0 
0.513706    0.1
0.517829    0.01
0.522513    0.001
0.519461    0.0001

What can be the reason that relax() influences writeInterval and changes the solution?

[alexeym]

Hi,

Surely it is very fascinating game called "guess what I have done wrong", and lots of people on the forum like to play it.

As I have written earlier: if your solution depends on time step, check your convergence. Did you?

Are your time steps equal and smaller than writeInterval in all cases?

[gaza]

I checked convergence and set timestep to be fixed