Interface kernel advice needed

[joe61vette]

Hello:

In one block I have a temperature variable that is "constant monomial". In an adjacent block, a different temperature variable is "first Lagrange". Both are FEM, so I am not mixing FEM and FVM.

I would like to place something like an interface kernel between the two blocks to allow conduction heat transfer across the boundary. Is there an existing interface kernel for such a case? Is there an example somewhere for something like this?

Thanks,
Joe Kelly

[jiangwen84]

Hello:

In one block I have a temperature variable that is "constant monomial". In an adjacent block, a different temperature variable is "first Lagrange". Both are FEM, so I am not mixing FEM and FVM.

I would like to place something like an interface kernel between the two blocks to allow conduction heat transfer across the boundary. Is there an existing interface kernel for such a case? Is there an example somewhere for something like this?

Thanks,

Joe Kelly

Depends on your problems, you can use either thermal contact https://mooseframework.inl.gov/modules/combined/tutorials/introduction/step01.html or thin layer heat transfer https://mooseframework.inl.gov/moose/source/interfacekernels/ThinLayerHeatTransfer.html

[joe61vette]

Thanks Wen. Do you know if either of these work when the variables are of different types? I basically have DG on one side and CFEM on the other side. I am about to quit for today and will give this a try tomorrow.

[jiangwen84]

Thanks Wen. Do you know if either of these work when the variables are of different types? I basically have DG on one side and CFEM on the other side. I am about to quit for today and will give this a try tomorrow.

Theroetically, I do not see any issues with coupling different types of elements. However, I believe this has not been tested before, so you might have some troubles there. Let us know what you find out.

[joe61vette]

I have tried three approaches without success. The input files are attached.
interface.zip

InterfaceTester.i : uses an internal sideset between the 2 blocks. Mesh is not split. SideSetHeatTransferKernel is used. Result is non-convergence due to a ZEROPIVOT.

InterfaceTester_2.i : Mesh is split. ThinLayerHeatTransfer is used. Result is a segfault.

InterfaceTester_3.i : uses an internal sideset between the 2 blocks. Mesh is not split. InterfaceDiffusion is used. Result is non-convergence due to a ZEROPIVOT.

Any ideas?

Thanks.

[GiudGiud]

Tying this to the other thread: #23846

Seg fault in test case 2 is definitely a problem. We'll fix it
For 1 and 3, please see the other thread.