[chrisf90]

Hi everybody,

Im conducting some simple natural convection simulations: I am simulating a vertical channel 3,2m in height and 0,4m in width. Both sides have a fixed temperature of 313K and the air inlet temperature is 293K. density variation of air with respect to the temperature was modelled with the boussinesq approximation.

Hence the Ra-Number is about 5e10 which indicates turbulent flow. Therfore I implemented the k-e RNG turbulence model and refined the grid in the near wall region (highest y+ value is about 1,3). But the solution seems not reasoable for me. I expected the velocity to be zero in adequate distance to the wall, also the thermal boundary layer is pretty thick (see attachment). Calculated Nu-Numbers (Q*L)/(lambda*DetaT) are similar to analyticly calculated ones. So I am quite unsure what to think about this solution.

If I conduct the same problme without Turbulence model, the temperature and velocity field seem more relieable to me.

Any input would be highly appriciated.
Regards
Chris