[bmb333]

I am modifying simpleFoam such that I can operate with two fluids, which do not mix but shares the same mesh.

I have prepared new fields which are U1, U2, p1, p2, phi1, and phi2. So, I no longer have the original U,p,phi, and such.

Now, when I try to run the following piece of code (which is itself a part of original UEqn.H in simpleFoam):

fvVectorMatrix tU1Eqn
(
fvm::div(phi1, U1)
);

I get an error saying

failed lookup of phi (objectRegistry region0)
available objects of type surfaceScalarField:
2(phi2 phi1).

Obviously, there is no phi since I intend to work with phi1 and phi2.

But, it is quite surprising that even if the input arguments to fvm::div were phi1 and U1, it still tries to find phi.


When compiling, I did not have any error.

Do you have any suggestion on how to work this out?

[dlahaye]

Not sure. Is your pressure solve (pEqn) looking for the original phi? Should you try modifying scalarTrsansportFoam first?

[bmb333]

Originally Posted by dlahaye

Not sure. Is your pressure solve (pEqn) looking for the original phi? Should you try modifying scalarTrsansportFoam first?

Hi, I appreciate your suggestion.

I tried modifying the scalarTransportFoam so that all the variables like T, U, phi are converted to T1, U1, phi1.

I ran the pitzDaily in the tutorials/basic/scalarTransportFoam, and there is no complaint. What a surprise.

I am suspecting whether this has more to do with the case directory settings than the solver itself.

Answering the other part of your comment, if my solver (which tries to adapt the simpleFoam) did use phi somewhere, then I would be getting an error during compilation of the solver because I do not declare phi at all. However, during compilation, there is no error. It only spits out an error during the execution of the case with the solver.

I will keep digging and try to experiment with the modified scalarTransportFoam with different initial/boundary conditions...

[bmb333]

I found that it might have to do with whether you are working with fvScalarMatrix or fvVectorMatrix.

For example,

fvScalarMatrix TEqn
(
fvm::div(phi1, T1)
==
fvOptions(T1)
);

This works without errors.

But,

fvVectorMatrix tUEqn
(
fvm::div(phi1, U1)
==
fvOptions(U1)
);

This will try to search for phi during execution and throw error if phi is not there as a registered object.

I am not sure whether there is some dynamic referencing that is happening to actively search for phi when you run fvVectorMatrix..?