[Richard_R]

Hello,

I am a new user and have been modelling a barn with heating pipes, to see if the heating can be improved using forced convection. It looks promising. In the attached pictures the model is cut in half by a mirror, and the heating pipes, with flanges, are bottom left. The slot just above them is a fan blowing downwards, simulated by using fixedJump in p_rgh. The wall and roof, and air initially, are at 5C and the pipes are at 55C.

The temperature view shows air blown downwards past the pipes, and the warm air flows across the floor, then up and anticlockwise round the room. I have added some velocity glyphs. The strange thing is that although the circulation is anticlockwise, in the layer of cells along the roof it goes the other way. To investigate I added a thin extra layer of air at the roof. When it is just 10mm thick, this backwards flow is nearly all inside it. Air in this layer accelerates as it travels up the roof, reaching an impossibly high 19m/s at the top. The other picture is a closer view showing U near the top. When I simulate natural convection, by turning off the pressure jump, the circulation is then clockwise but the very fast flow at the top is still there. I think the downward flow next to the vertical wall is faster then it should be too. The zipped model is also attached.

I tried setting all the temperatures to 300K to check that it does stop. It settles down with a very slight temperature gradient (about .03K warmer at the bottom than at the top), but air circulates slowly anticlockwise, at around 0.1m/s (is this perpetual motion?) However the cells in the top layer did not have a high velocity but were almost static, as expected for a boundary layer.

The original model was based mainly on the room with thick ceiling tutorial, but this one has to be transient to see how fast it warms. Both GAMG and PCG solvers show the same effect. Also the momentum predictor made no difference whether on or off. The BC for U is noSlip. The average Courant number was about 0.8 but the maximum was around 20, presumably where the flow is rapid.

Does anyone know what is happening or how to cure it? I want to add a solid wall and roof and do a multi region analysis, but will the results be reliable?

Incidentally I found a similar report of a region of unnatural high velocity here: www.cfd-online.com/Forums/openfoam-solving/238196-chtmultiregionsimplefoam-forced-convection.html

Thanks,
Richard