I would have used a surface source not an outer wall in the second one. I have no definitive idea why it would not be the same thing though but something does not seem quite correct about your second approach. It strikes me as there is no need for an outer wall as it is a conduction only problem. The other thing is that temperature BC's are usually not very realistic - they tend to be too stiff. In your first case (and the second one) complete temperature uniformity seems difficult to pull off in an experimental set up. The other issue is the conductivities of the materials involved play a role as well. I would check the powers induced by the constant temperatures. I think you need to apply constant powers and ensure that the energy has only one path to follow. The thermal inertia's will play a roll as well and I am not sure how that affects things when you are using a prescribed temperatures for the source. The cases may not be as identical as you assume. What are you trying to figure out? How to model or how the program works? Or trying to mimic some physical situation?
Thank you for your reply. A surface source was also my first attempt but there is no way to use it with defined temperatures (just heat transfer) (at least in solidworks 2013). That is why I switched to volume sources (not very handy if you have more complex surfaces) and boundary conditions.
I am aware of the fact that fixed temperatures are not ideal in most cases but since my environment is not something I cannot model in SW my only information so far is the surface temperature (which I measured with a simple test setup). For my first rough test simulation I will just assume that the surface temperature will not change that much for small changes I do inside my setup.
But at the moment and before I continue my simulation I am more interested to understand why there is a difference between my two approaches. If there is a difference in transferred power as you mentioned I don't know why. Since the heat flow / power has to be proportional to my thermal conductivity (which is equal in both cases) and the temperature difference (which is also equal at the beginning).
The way I model my scenario should not affect my results, so there has to be an issue either with my simulation setup or with the program.
Does anyone have an idea what's wrong?