[boffin5]

Followers of this forum are aware of my long running travails in creating a chtMultiRegion case to simulate the installation of a coolant radiator on an airplane fuselage. My goal is to validate a real installation done on a light aircraft in Canada, specifically with regards to claims that the associated Meredith effect will result in a drag reduction due to the heated radiator exhaust.

Over the last many months, I have labored on this simulation, constantly running into problems, and constantly fixing them, only to run into more. I have tried to methodically troubleshoot by creating simple diagnostic cases, after which I apply lessons learned to the real simulation. And I have made substantial progress, but today, again I ran into the "max iterations exceeded" message. At this point I am flummoxed.

Please note that as a check, I took the geometry files, minus the radiator porous zone, and put them into a simpleFoam simulation, where it ran just fine.

At this point, I have tried everything I can think of, and have run out of options. This is a bitter feeling, as I desperately need to get the case to run. My goal is to continue communicating with the builder of the real airplane installation, and get to the bottom of the validity of the Meredith effect.

I have simply run out of ideas, and I am fervently hoping for help from the experts in this forum, who have generously helped me before. I am attaching some files pertaining to the case, and can gladly supply more if requested. Note that the runlog has words about "sym-fairing", but this is an artifact of a version where I had the fuselage and radiator fairing as separate stl files; since then I combined them. Also find below dropbox links to zip files of both the simpleFoam case and the chtmultiRegion case.

If I am so lucky as to get help, I will be thankful, beyond words.

Alan w

https://www.dropbox.com/scl/fi/grx0e...3s5ra1kwvhpli7

https://www.dropbox.com/scl/fi/uvb4k...g290f2j67wapay

[boffin5]

Since I ran into the 5 file limit ....

[Yann]

Hello Alan,

One idea, if you didn't already try it: you could keep the porous zone in your simpleFoam simulation, so you can make sure the porous zone behaves properly before switching to chtMultiRegionFoam.

Another thing: have you tried using some limiter in fvOptions, such as limitTemperature? It can sometimes help to avoid things blowing during the first iterations.

Hope this helps,
Yann

[37269]

Why do you have your timestep set so high? I see in the controlDict that you have tried a lower value as well. Current setting is 1s, commented out are 0.001 for deltaT. Did that work better? Maybe try going even lower with that, at least in the beginning.

Your max. CFL is currently larger than 130,000. Not sure if that will give you a sensible solution at all.

[boffin5]

Thank you 37269,
Even when I tried a timestep of 0.0001, it failed at the second timestep.
No matter what I do, it fails at the second timestep.
This is so perplexing, given this paradox:

Yann helped me set up a similar cht case, with everything the same except for the use of a simplified radiator fairing. It ran okay, therefore the setup file settings are okay.

However, for the cht case with the complicated geometry files, it fails. So given that the setup files are the same, this implies that the geometry files and mesh are at fault.

But, with the complicated geometry inserted into a simpleFoam case, it runs fine. This implies that the geometry files and mesh are okay.

My head is spinning on this one, and I am at a full stop.

[boffin5]

It turns out that my boundary conditions had problems. I made changes as shown in the attached files, and at long last, I have a chtMultiRegion simulation, as illustrated by the flow in the image!

But now, I would like to make sure that data is not just garbage; I have questions about my assumptions in the boundary conditions. Mainly, as to my boundary conditions for the radiator itself. As seen in the attached image, in addition to BC's for the domain inlet and outlet, I have a radiator inlet and outlet, for which I applied similar BC's. All the boundary conditions are attached, but here is a snippet:

object      U;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

dimensions      [0 1 -1 0 0 0 0];

internalField   uniform (0.01 0 0);

boundaryField
{
    #includeEtc "caseDicts/setConstraintTypes"
    
    "frontier|ground"
    {
        type            slip;
    }
        
    inlet
    {
        type            fixedValue;
        value           uniform (0.01 0 0);
    }
    
    outlet
    {
        type            inletOutlet;
        inletValue      uniform (0 0 0);
        value           uniform (0 0 0);
    }
    
    sym-fuselage
    {
        type            noSlip;
    }
        
    rad_radinlet
    {
        type            fixedValue;
        value           uniform (0.01 0 0);
    }
    
    rad_radoutlet
    {
        type            inletOutlet;
        inletValue      uniform (0 0 0);
        value           uniform (0 0 0);
    }

    rad_radfrontier
    {
        type            noSlip;
    }
       
    solid_to_fluid
    {
        type            fixedValue;
        value           uniform (0 0 0);
    }
    
    "proc.*"
    {
        type processor;
    }

}

Is this reasonable?
Also, I have BC's for the domain, termed as 'frontier' and 'ground'. In this case 'ground' is the same as 'frontier'; I only have it this way in case I do a simulation of a car. It is shown as 'slip' above, and for most of the other BC's. Again, I don't know if this is correct; I couldn't find similar examples in the tutorials.

I want to give sincere thanks to peterhess, who gave me invaluable advice, and to many others in the forum, who have beared with me through the last months and years.

Alan w

[boffin5]

Pleased with getting my cht case to run, I was ready to iterate the flow values!

So I changed the fluid p from 84559 to 84307, and the T from 551.15 to 280.
For the solid radiator, I changed p from 84559 to 84307, and the T from 623.15 to 280.

And then it failed, with the "max iterations exceeded" message

So I looked at this thread: FOAM FATAL ERROR Maximum number of iterations exceeded

Following advice there, in fvSolution I changed ddtSchemes from steadyState to Euler. I found that this necessitated this addition in the fvSolution solvers:
"rho.*" {solver diagonal};

Still, it failed.

Another entry in thread says to change the solver from newton to brent, without explaining how. I looked into it, but it looks complicated. Perhaps someone can provide a detailed guide as to how to do it?

For now, I'm again at a full stop, wondering why my simulation works with one set of environment values, but fails when I make minor changes.

[boffin5]

Here is the latest failure report. Actually, this one was done with a timestep of 0.01.