[.bastian]

Hi.

I am trying to set up a compressible MRF case using steadyUniversalMRFFoam or steadyCompressibleMRFFoam as a part of foam-extend 4.1.

In the boundary conditions I have specified a mass flow for the velocity at the inlet. During the first runs I noticed that the boundary condition seems to be not properly applied. As can be seen on the plot, the inlet mass flow is anything but constant.

My boundaries are defined as following:

0/U

inlet
    {
        type            flowRateInletVelocity;
        flowRate        0.177777777778;
        value           uniform (0 0 -20);
    }
outlet
    {
        type            zeroGradient;
    }

0/p

inlet
    {
        type            totalPressure;
        rho             rho;
        psi             none;
        gamma           1.693;
        p0              uniform 28500000;
        value           uniform 28500000;
    }
outlet
    {
        type            fixedValue;
        value           uniform 29706129.152;
    }

Any idea what i'm doing wrong?

[jherb]

Are you sure that you have specified the mass flow rate and not the volmetric flow rate? See https://github.com/Unofficial-Extend...hVectorField.H


What is the dimension of the phi-field in the outputs?

[.bastian]

Originally Posted by jherb

[...]
What is the dimension of the phi-field in the outputs?

That was one of the first things i've checked:

phi:

dimensions      [1 0 -1 0 0 0 0]; // which corresponds to [kg/s]

And even if this would have been mixt up the value should have been kept constant during the simulation. - Or am i wrong?

[jherb]

How have you calculated the massflows at the in- and outlet? Using function objects? Are the values on the patch faces used or the cell center values next to the patch?

Cell center values might/probably are different from the patch values. Your boundary condition sets the values on the faces.


What are the intial residuals at the end of each time step? Values for U and p should be below something link 1e-4.

[.bastian]

I've calculated the massflows with swak4foam using the following patchExpression:

patchMassFlow
     {
         type patchExpression;
         autowrite true;
         outputInterval 1;
        writeStartTime no;
         accumulations (
             sum
         );
         patches (
             inlet
             outlet
         );
         expression "phi";
         verbose true;
      }

So in my understanding it should be the flux which is leaving/entering the domain through the patch faces.
No, sadly the initial residuals are quite high.
Attached the plots of the last run.

[jherb]

Your residuals look really big.


Is this a steady state case or a transient simulation? If steady state, it is just not converged yet. If transient, you probably either have to decrease the time step significantly, or do (much) more outer corrector loops.