Discussion Forum

I am working on a conjugate heat transfer problem, which is a laminar air flow through a long narrow channel with a fin on each side. Fin thickness is about 1/2.5 of the flow channel width.
I have modeled the fins as solid domains (case 1) and separately as highly conductive layers (case 2). Both models were meshed using "Normal" physics controlled mesh and successfully solved with parametric sweep for Re, and in P1+P1. (same mesh for all Re)
The max and average solid surface temperatures for these two cases match closely. However, the "Total net energy power" for the 2 cases do not match. In case 1, the error (1 - Total net energy power/ Input Power) increases (almost linearly) with increasing Reynolds number, from near -0.1% for Re=500 to 7% for Re=2000.
In case 2, the error is near zero for all Re. (0.002% for all Re)

So I have 2 questions.

Q1: Why is the error increasing linearly for case 1 as the Re is increased and shows Re dependency? (Further Mesh refinement in this case has lowered the errors slightly but the trend still exists.)

Q2: Why is the error nearly zero and why is it independent of the Re for case 2?