[seregaxvm]

Hello!
I'm trying to simulate cooling of the pulse heater. Heating happens at high rate (819 [kW] in heater region) for a very short period of time (6 [µs]).
Unfortunately, simulation stops somewhere around 3 [µs] with either exceeded maxIter or negative temperature errors.
I've tried limiting minimum temperature in regions, subdividing the mesh, lowering deltaT value ... nothing seem to work.
What can I do to obtain a solution to this problem?


PS. Project may be run by issuing the command

make solve

PPS. I'm using scalarCodedSource defined in constant/heater/fvOptions to simulate a pulsed heat source.
PPPS. I it just me or scalarCodedSource does not work with MPI?

[piu58]

At first I would try to simulate just one pulse and understand how the physics need to be implemented.
At such a very short time, is it really necessary to have a transient solution here? Or may it be sufficient simply to add the amount of het in form of an increased temperature?

[seregaxvm]

I did a simulation of the model in static case with 1000 [W] of heater power (run command make setup-stationary to setup stationary solution). Everything worked fine.

I need a transient solution because I need to investigate maximum temperature in the model in time and optimize the geometry.

[seregaxvm]

I seem to overcome this problem. Previously I used a separate heating region, which was uniformly heated in time. I removed this region and added Gaussian heat source:

where A may be found from



There may be some mistake there because I have to adjust A. Or may be it's due to discretization. I don't know.

Here's my codedSource for the heater:

transientHeatSource
{
    type        scalarCodedSource;
    name        sourceTime;
    active        true;

    scalarCodedSourceCoeffs
    {
        selectionMode    all;

        fields        (h);

        codeInclude
            #{
            #};

        codeCorrect
            #{
            #};

        codeAddSup
            #{
                const scalar time = mesh().time().value();
                const scalarField& V = mesh_.V();
                const vectorField& C = mesh_.C();
                scalar xMean = 0, yMean = 0.00806, zMean = 0;
                scalarField& heSource = eqn.source();
                const scalar sigmaX = 1e-3;
                const scalar sigmaY = 1e-4;
                const scalar sigmaZ = 1e-3;
                const scalar freq = 200;
                const scalar amp = 6.3e6 * 0.130;
                const scalar impulse = 6e-6;
                const scalar T = 1/(2*Foam::constant::mathematical::pi*freq);
                const scalar func = fmod(time, T) / T;
                const scalar duty = impulse / T;
                const scalar Tot = func <= duty ? amp : 0;
                const scalar A = 0.8 * Tot / (std::pow(2*Foam::constant::mathematical::pi, 3.0/2.0) * sigmaX * sigmaY * sigmaZ);
                scalar heSourceTot = 0;
                forAll (heSource, i) {
                    scalar x = -std::pow((C[i].x() - xMean) / sigmaX, 2) / 2.0;
                    scalar y = -std::pow((C[i].y() - yMean) / sigmaY, 2) / 2.0;
                    scalar z = -std::pow((C[i].z() - zMean) / sigmaZ, 2) / 2.0;
                    heSource[i] -= A * V[i] * std::exp(x + y + z);
                    heSourceTot -= heSource[i];
                }
                Pout<< "Adding heat source: " << heSourceTot << " [W] " \
                    " " << heSource.size() << " with center at (" << xMean << ";" << yMean << ";" << zMean << ")" << endl;
            #};

        codeConstrain
            #{
            #};
    }
}

[ErenC]

AdjustableTimeStep seems active but you didn't spesify maxCo and maxDT. First, I would disable adjustableTimeStep and run, if it doesn't work lower the dt again etc.
Also, your first timestep is 5e-7 but your timePresicion is 1e-6. You might wanna lower that too.

[seregaxvm]

Originally Posted by ErenC

AdjustableTimeStep seems active but you didn't spesify maxCo and maxDT. First, I would disable adjustableTimeStep and run, if it doesn't work lower the dt again etc.
Also, your first timestep is 5e-7 but your timePresicion is 1e-6. You might wanna lower that too.

I ended up disabling it completely and doing a timestep. Otherwise simulation may skip my heat source pulse.

[ErenC]

Well, by using 1e-6 you'll get a trapezoidal'ish wave, if you need sinusodal wave you might wanna lower it.

[seregaxvm]

Originally Posted by ErenC

Well, by using 1e-6 you'll get a trapezoidal'ish wave, if you need sinusodal wave you might wanna lower it.

No, I use a square wave, so it's ok.