[bachga]

hello everyone,
i'm having a problem with the convergence of the result of the simulation of the natural convection. Are there anyone have experience about it?

[virendra_p]

i have some experience in multiphase buoyancy flows but not in natural convection.
Anyways i'll share few things:
1) My problem was flow through a pump sump ( open channel with pump at end) with surface air suction. This particular setup was very sensitive to initial solution so i first converged the solution with single phase setup and then used fluent's 'interpolate' feature to initialize the multiphase setup. Convergence issues were solved by this approach.
If this process makes any sense in your problem, please try it out.
2) I found that steady state simulations produced absurd results so i had to stick to transient simulations.
3) At the end of this html file you'll find some guidelines for high Ra flows, see if that helps. html --> http://aerojet.engr.ucdavis.edu/flue...ug/node572.htm

[windykawis]

I also have an experience about natural convection but not sure that model is correct.
I used boussinesq model and specified operating density in operating condition.

[bachga]

Originally Posted by windykawis

I also have an experience about natural convection but not sure that model is correct.
I used boussinesq model and specified operating density in operating condition.

i also use boussinesq model for density but i think the operating density was defined in the properties of the material for fluid which i use here is oil. My problem is heat transfer by natural convection in the transformers. Do you have any idea about it?

[aquac]

Originally Posted by bachga

i also use boussinesq model for density but i think the operating density was defined in the properties of the material for fluid which i use here is oil. My problem is heat transfer by natural convection in the transformers. Do you have any idea about it?

Reduce the under-relaxation factors under Solution Controls -> Under-Relaxation Factors. Cut the Pressure and Momentum factors by half and take down the Energy factor to 0.8 or lower if necessary. Re-initialize if the solution doesn't seem to be converging.

You can also try playing with gravity; set it at -1 and see if it seems to converge; increment it steadily till you converge at -9.8.

Another thing to try is Adapting the mesh...go to Adapt->Region and manually adapt some part of the mesh.

Make sure you're using PRESTO! discritization for pressure

If all else fails, disable solving for the Flow equations under solution controls and try just solving for Energy first.


Natural convection (and buyoncy driven flows in general) are some of the hardest simulations, so you have to be resilient. Expect having to run thousands of iterations to get a solution.

Hope that helps.

[bachga]

Thank you . I'll try it now