[fly829]

Hi.
I have some problem to compute high Rayleigh-number(e^10) natural convection,but the solution does not converge.

Problem geometry is a wind turbine nacelle(closed region), one of wall consider constant heat flux, the other wall consider constant temperature, no inlet and outlet.

the setup i use is

steady
consider gravity
model: energy and k-epsilon
density consider incompressible ideal gas
Presto!

but the continuity and energy are not converged

how do i setup the high Rayleigh-number natural convection?
can someone give me some comment?

Thank~

[skullCFD]

Natural convection problems are difficult to converge. Lowering the relaxation factors might help. I find SIMPLE to be better. The body force can be high for high Ra which might also cause convergence problems. A transient simulation with very small time steps might help in the beginning. You can use that solution and run a steady state simulation later. I am not sure if all nacelles have the BCs that you have but one important thing to check for natural convection problems is to see if there exists convection in the first place (i.e., which wall in your case is the hottest one, top, bottom, left or right?). If not, you might face convergence problems.

[fly829]

Thanks for replying~

i already decrease momentum relaxation factor to 0.1

however the continuity didn't convergence

the B.c i set is

Wall 1: constant heat flux
Other wall: constant temperature(assumption)

i will try your suggestion, transient first, and steady later.

Thanks
Regards

[skullCFD]

Transient simulation with very small time-steps helps a lot of times. Also, increasing the value for pressure under-relaxation to 0.6 or 0.7 with momentum less than 0.4 (you already have 0.1) will help bring the continuity residual down (may not work all the time but worth giving a try).