Natural convection convergence problem
 

Hi,

I'm trying to solve a square cavity (1x1m) with natural convection heat transfer. Both top and bottom are insulated, left surface is T=330K and right surface at T=300K. The Boussinesq approximation is use for density.

Thermophysical properties:
rho=1.127
Cp=1007
mu=1.9114e-5
k=0.0271
Beta=0.006092
gravity= -9.81

I already tried to start the simulation with default solution methods and after a few iterations chance momentum and velocity values in order to achieve convergence of the result, but it doesn't worked.

Any other helpful suggestion?
Thanks in advanced ;)

Hello!
Please share your case file to be checked.

Regrads

Hi Bruno,
I suggest setting up a transient simulation (best way for natural convection in closed domain). Otherwise you have to control under relaxation factors.

Regards

Hello Bianco

Could you expand your thought about why you need to do transient calculation for convective internal flow? It is because of iterations graph going up?

I am doing calculation also with natural convection in closed domain and would like to know more

Hi zombiaska,
according to ANSYS Guidelines (and my experience with air flow in rooms) it's difficult to reach convergence, for continuity equation specially: this is due to a problem of mass balance.
ANSYS suggests 2 ways to solve this problem:

1) Set up a transient simulation: the domain starts with a certain mass (calculated from initial conditions, of course) and mantains this value during the entire simulation.

2) Boussinesq approximation: it helps the convergence by using a constant density in some equations.

Furthermore I suggest using RNG-k-epsilon as turbulence model in case of heat transfer in rooms. Anyway it's very difficult to find an ideal model for this kind of case: it takes a lot of time and effort! :rolleyes:

Best Regards

Hi Bruno
your problem my be is turbulent.check rayleigh number.

Hi
For natural convection use body force weighted for the pressure and power law scheme for the momentum and the energy equations.

try to set a initial velocity for the domain in the "Solution initialization", try this to be close to the average velocity you expect or less

Hi bruno,
Reduce the relaxation factors of momentum to 0.5 and reduce energy to 0.95 and check your solution..if its not converged, go till 0.1 for momentum and 0.9 for energy..
sudhir

Great advice for natural convection problems!
 

This helped get my simulation working. Thanks!

Could you please explain a little why Body-Force-Weighted and Power Law are better for convergence?

Many people recommend PRESTO! for pressure for buoyancy driven flows.