Hi I would like to ask about natural convection problems at high Rayleigh number in closed cavities using Fluent. I have been testing the natural convection with Fluent with the article Xu W. and Chen Q.1998. "Numerical Simulation of Airflow in a Room with Differentially Heated Vertical Walls", ASHRAE Transaction 104(1):168-175. The author of the article simulates a real dimension room with a Rayleigh number around 10e10. I repeated the case and after some tests I discovered that Fluent have difficulties in natural convection in closed cavities running with a high Rayleigh number (even using a fine mesh, presto!, and hybrid). I have been using the real physical properties of the air except for the gravity because I use it to adjust the Rayleigh number. The convergence for all residual is okay until Ra=10e6. The convergence is very hard with Ra=10e7 and it is impossible with Ra=10e8, even when it is supposed to be laminar. I will really appreciate some comments about this situation and/or suggestions to solve the problem. Thank you so much

I was working on a high Ra natural convection. However the fluid was water, there were some tips and trics. I don't renember everything but i'll try.

-segregated solver, sec. order except the pressure: body force weighted,

-important thing, operating density strongly coupled with corresponding temperature, I'm not shure, but try lowest value of density for your conditions

-I didnt use a bousinesq app. It is good practise to write own properties UDF with exact dependence of density in the term of temperature (only)

-try to manipulate under-relaxation factors. Renember about strong coupling between energy and momentum eq. Take a special care about this.

I hope it will help. Good luck.

Michael

Michel I am very glad with your feedback. I will tray your suggestion and then I will comment what happened.