[Gerry Kan]

Howdy Foamers:

I have always been wondering about this and wanted to ask for your opinions.

I am setting up a transient run using a variant of rhoPimpleFoam, involving multiple species. To speed up initialization I would like to apply a steady-state initial flow field, possibly by using potentialFoam.

My questions are as follows:

1) Is it possible or advisable to do this in a single solver, i.e., the solver calling potentialFoam or solves for the steady potentialFoam through some code change?

2) Should a separate call to potentialFoam is required, how should I prescribe the results as initial solution?

Thank you very much in advance, Gerry.

[wildfire230]

Hi Gerry,


I don't think it is advisable to do this with a single solver, it's best to keep things modular. First you should run potentialFoam, and then run your other solver with the output of potentialFoam as the initial condition for the new solver. If I remember correctly, potentialFoam actually directly overwrites the 0 directory, so you probably don't need to change anything and you can just directly launch the second solver. I might not be remembering correctly. If it doesn't overwrite 0, just change your startTime in your controlDict file to whatever time directory contains the output of potentialFoam.

[LuckyTran]

Run potentialFoam and then your solver. You don't need to hack any solver. Use changeDictionary and scripting to make your life easier.


Actually you can just put this into your fvSolution dict to initialize using potentialFOAM.

potentialFlow
{
    nNonOrthogonalCorrectors 10;
}

[Gerry Kan]

Folks:

That's kind of what I thought. It might be easier to send a potentialFoam job and then rename the results to the initial conditions of the transient run.

Another question, though. Is there a way to introduce a weighting factor (between 0 and 1) so that I can set this initial condition as an interpolated average between the potentialFoam solution and the quiescent initial conditions?

Thanks in advance, Gerry.

[wildfire230]

I haven't personally heard of an option to introduce such a weighting factor. If I were desperate to do that I would probably end up writing a small utility that reads the two fields and does the weighting. There is probably a more elegant way that I am not familiar with.