[pete20r2]

Hi all,
I wanted to ask here before I even try something like this to:
A: see if anyone has tried this before.
B: see if this is viable.

The case that brought me this issue is the following;
I have an internal pipe flow with heat being supplied at the wall from an adjoining solid body. I am particularly interested in where the hot regions travel in a turbulent flow.
The issue I have is that I can either:
1) Use cyclic boundaries with a fVoptions imposed pressure gradient to drive flow and get correct p and U fields. Using this method I have no way to reject heat!

2) Mapped inlet and outlet boundaries, the U field can be repeated to the inlet but the p field must be zeroGradient on one side and fixedMean on the other. Using this I method I can repeat the T field from outlet to inlet while resetting the average to say 300K (setAverage).
But now the p field is wrong and does not agree from inlet to outlet.

In a straight pipe this might not be an issue but my pipe is a sinusoidal rectangular channel. (p is no consistent across boundary)

What I would like to do is combine the best of both worlds: cyclic patches between inlet and outlet and also use a modified fixedJump condition (with averaging) on the T field to prevent the temperature from running away!

Does this sound reasonable, has it been done before or is it something worth pursuing?

Regards
-Peter