Shell Conduction
 

Greetings to all!
The problem I am solving is: forced convection in a closed air volume of complex shape with heat transfer through walls that are very different in thermal conductivity (a "pie" of different layers of material in different zones).

Since the volume is very complex in shape and heat transfer is possible along its surface, I suggest using a shell conduction model. To begin with, I solved a test problem and came across a different result for a three-dimensional and shell wall.
Description of the test problem in the attached miniature: two cubes of 1 m3 of air, in one 1 kW of energy is generated and "blown" on the lower aluminum wall, the remaining walls are adiabatic, including the wall between the two cubes. Under the second cube, there is also an aluminum plate in front, and a "pie" made of polystyrene and aluminum plate in the back. The upper face of the second cube is made of aluminum, the only interface in the model that has a boundary condition "convection" with a heat transfer coefficient of 15 W/m2K, free flow temperature of 233K.

As a result, we have a different average temperature by volume (the difference is about 15%) and different wall temperatures. The results are in the lower plane in the attached miniatures: if I see a "physical" temperature distribution for a three-dimensional wall, then in the shell model the temperature on the aluminum plate of the first cube falls to the boundaries - this is very strange and may indicate that part of the energy is lost at the "ends" of the shell aluminum wall. About the ends of the shell wall, the help contains information only that "the flows at the ends of the conductive shell wall are not included in the heat balance reports, these flows are correctly taken into account in the solution." I also could not find information about how the interaction of two boundary shell walls with one or more sets of layers is described. If we consider the contact point of the aluminum wall and the polystyrene-aluminum wall in the three-dimensional case, the contact is between aluminum and polystyrene, but is this the case with the shell wall?
I will be grateful if I get an answer to a question or a link to the literature where you can read more about the interaction of the ends of the shell wall, as well as interesting examples of solving problems with walls of complex shape using the shell wall, thank you.