Hi everyone,

I am modeling a compact heat exchanger in 3-D, as of now the model consists of 3 flow channels stacked vertically surrounded by an alloy plate. The working fluid is Helium with an operating pressure of 7MPa. The middle channel is a hot channel with a temperature of 1173K, the top and bottom channels operate at a temperature of 973K.

The modeling consists of 4 Volumes, one for each channel and one for the solid plate. When I import the mesh into Fluent the shadows of the solid faces which interface with the fluid are created properly. Neither heat flux through the channels or their wall temperature are constant or known.

The fluid density is varied via the non-compressible ideal gas model (ideal gas model leads to divergence in the solver possibly caused by the lack of a mass flow inlet). The inlet is a velocity inlet (3m/s for all fluids), a laminar model is used, and the fluid outlets are set as pressure outlets.

Once the solution converges (to 1.5e-5)I plot the default interior temperatures corresponding to the fluids. These temperature curves show the cold and hot channels coming to complete equilibrium, which makes no sense physically. I lowered the thermal conductivity of the plates by a factor of 10 and this did not change this behavior.

Periodically (roughly once every couple of hundred iterations)the residuals for continuity jump up slightly and give a curve resembling signal noise in an analog electronic system before settling down and decreasing again.

Any words of advice or suggestions would be greatly appreciated. Thanks, Justin

I have redone this model using a much more coarse mesh, allowing it to compute faster (at this point precision doesn't matter because my results are garbage) and am still getting nowhere slowly.

I have been reading the manual, and my interpretation of what it says about 2 sided walls is that fluent handles the heat transfer here based on adjacent cells. I take this to mean that I need not specify heat flux or temperature. Am I way off base for interpreting it this way?

Nevermind.

I switched to a coupled solver and it is all good now.