[talop]

Good day,

I am trying to simulate natural convection from a hot cavity in air using incompressible ideal gas model. Unfortunately, simulations crash or give unrealistic flow fields such as thin jets of air flowing in from the outlet patch. I think the initial pressure field "p" is the suspect and managed to initialize the internal field with hydrostatic effects (p = p_ref + rho*delta_h). The gravity acts along the -y direction.

For the boundary patches, I have tried the following BC:

outlet_1
{
    type            calculated;	
    value           #codeStream
    {
         codeInclude
         #{ #include "fvCFD.H" #};

         codeOptions
         #{
               -I$(LIB_SRC)/finiteVolume/lnInclude \
               -I$(LIB_SRC)/meshTools/lnInclude
         #};

         //libs needed to visualize BC in paraview
         codeLibs
         #{
               -lmeshTools \
               -lfiniteVolume
          #};

         code
         #{     
                const IOdictionary& d = static_cast<const IOdictionary&>
		(
                    dict.parent().parent()                  
                );
                const fvMesh& mesh = refCast<const fvMesh>(d.db());
                // const surfaceVectorField& Cf = mesh.Cf(); 				
			
		const label id_o1 = mesh.boundary().findPatchID("outlet_1");
                const fvPatch& patch_o1 = mesh.boundary()[id_o1];
                const vectorField& Cf= patch_o1.Cf();				
                
		scalarField p_Outlet_1(patch_o1.size(), 100000.0);

                scalar p_Atm   = 100000.0;
	        scalar rho_Atm = 1.3772; // = 100000/(287*253)
	        scalar delta_h = 0.8;	
                
                forAll(p_Outlet_1, j)
		{
         	    const scalar y = Cf()[j][1];
                    p_Outlet_1[j] = p_Atm + (delta_h - y)*rho_Atm*9.81;
                }
				
                writeEntry(os,"", p_Outlet_1);
        #};
    };	
}

Then running decomposePar or buoyantFoam gives "error: no match for call to ..." from the line

const scalar y = Cf()[j][1];

Another minor issue is that I am unable to loop over the patch names so I wrote similar codedStream BCs for all 12 patches seperately. The 0/p file became large ~650 lines.

Running the code for one patch and initializing the others as type calculated and value uniform 100000.0 did not change the problem.

I am using OpenFOAM v10 (from the foundation)