Learning Forum

[aitor.amatriain]

I am considering the following engineering problem:

[Rob]

Connect the two fluid domains and label the common face as "wall" in the geometry or meshing tool. That'll give you the wall and wall shadow pair. In the Fluent bc panel, either just give the wall a thickness (no lengthways conduction) or tick the shell conduction option (allows lengthways conduction) and fill in the pop up.
The cells that are referred to are virtual so the solver has a memory storage location, you can't see or interact with them.

[aitor.amatriain]

Hi Rob, thank you for your answer. I am also planning to simulate the following geometry:
Does it matter if I define the shells in the "original" wall or in the shadow wall? For example, considering that the walls are added in the normal direction, I can put some of the layers in the "original" wall and the rest of them in the shadow wall?
And one last thing. Maybe it is obvious, but Fluent "connects" the last shell conduction layer with the fluid that is on the other side of the wall, right?
Thank you

[Rob]

You need to connect the faces in geometry. Fluent knows that the wall & wall:shadow are paired so unless you mess with the "coupled" setting it will work everything out for you. What I can't say (check the manual) is how layer 1 and layer 5 are arranged ie I think layer 1 will be next to the fluid whos surface you pick (wall or shadow).

[aitor.amatriain]

Sorry, I did not check that. If I define the shells in the "original" wall, then I have noticed that the same shells appear defined in the shadow wall. And based on the manual, yes, it seems that layer 1 is the one that is in contact with the "original" wall.
Apart from the aspects related to the implementation, is there any reference to know what is Fluent doing? It just creates a mesh of certain number of points and solves the 3D heat equation in them?
Thank you

[Rob]

That sounds about right - the wall and wall:shadow are the opposite "sides" of the wall, so in this case face different fluids.
Fluent creates virtual cells and then solves the conduction through and along the walls. This should be covered in the heat transfer section of the Theory Guide. Check for links in the Boundary Condition section of the User's Guide too.

[aitor.amatriain]

Thank you, Rob. I understand that in terms of equations this is the same as meshing separately all of the shells and then defining them as solids. Of course, for the user Shell Conduction is much easier to implement that the alternative approach.

[Rob]

You're welcome.
More-or-less. A fully resolved solid will be a bit more accurate, but the cell count cost will also be significantly higher.