[marzoa]

Hi everybody,

I am working now in a problem consisting in a closed volume formed by two twin, parallel pipes connected by a transversal pipe in each end. The central part of the pipes is at very low temperature (< 2K) while the walls near the ends and the ends themselves are at higher temperatures (between 4 and 17 K). The fluid inside is Helium (as Real Gas), at low pressure (80 mbar); subsequently, density is very low, around 1.9 kg/m^3, but it is still gas. The simulation has been run in steady-state.

So, natural convection is the main effect that drives the fluid inside the volume. I applied k-epsilon turbulence model + Xu correction (for bouyancy-driven flows) refining the prism mesh near walls to capture velocities and heat transfer properly. Then I switched to k-w SST model, enabling low-Re damping factors, because Re numbers are fairly low and the gas in the central part of pipes is still., so flow seems to be transitional.

Monitoring several variables along the volume, they appear to be constant, except the global boundary heat flux: a sum report of this variable has been done, taking into account every boundary in the problem. Then, if the solution acheieved corresponded to a steady-state, this value should be near zero, while it is not: it hits the value of -2E8 W/m^2.

It seems that solution is not yet steady, or it is oscillating. Coupled implicit solver was used, reaching a Courant number of 200. Natural convection is a non-steady physic phenomena, but, as far as I know, this solver makes time-marching, so it should not be that difficult to approach a steady solution.

Anyone with experience modelling natural convection could help me? I wanted to know if the models selected are adequate for a problem of this nature and why the energy imbalance is not zero.

Thanks in advance.

Best,

M.