Problem with buoyancy driven in enclosure
Hi all;
I'm simulating of a cubic closed water Tank (enclosure) in a buoyancy driven (Natural Convection) manner. This tank is heated by an electrical element located on beneath(below) plane as Temperature of this below plane is always 350 (kelvin) whereas upper plane temperature is in the ambition condition (300 kelvin) and cooled by ambient. Also, side walls are insulated. So, natural circulation occures into the water enclosure.How can I resolve this problem? I used a piecewiselinear for water density definition to be considered buoyancy force, but the results that I took, seem not to be logical and even those diverged in some cases!:( Thanks a lot for any suggestion and advise to be treat this problem. 
Hi Asghari_M,
I'm only a newbie, but maybe this can help you: http://my.fit.edu/itresources/manual...f/tg/tut05.pdf cheers 
Hi alessandro;
That's good. I read it beforehand. However the fluid in that tutorial is air and air density follows boussinesq approximation in that problem. But, My working fluid is water and I use piecewiselinear approximation for the calculation of water density. Should I use boussinesq approximation for the calculation of the air density? I would be appreciated if you give me some the guidelines to be resolved these problems. Thanks a lot for any suggestion, too. 
I'm so sorry, but I use CFD only like a support for my job: Architecture. I do not have the necessary knowledge to give you a more specific help :(

Thanks again Mr.Zomparelli;
However, I 'll welcome for anybody coming to our discussion and giving further suggestion for presenting a basic solution, because the problem has not been still resolved.:( Asghari_M 
This should work.
Try with the following settings:
Presto! for Pressure; Second O. Upwind for Momentum and Energy. 
You can use Boussinesq approximation for water. Even otherwise, some of the ways to deal with free convection in water are:
1. to first solve a pure conduction problem and then use it as initial guess and apply free convection. 2. to solve a transient problem first and use it as initial guess for the steady simulation. 
Hi Dears;
I examined Boussinesq approximation, S.O discretization for energy solving and a transient solution for my problem, dT=0.01. Nevertheless, Spite of these settings , the solution isn't logical, yet:(. Really, this problem seems to be simple to solve. However when solving residuals for continuity and vector velocities aren't logical, yet. What should I do? Can air as a material help for checking the settings ? If necessary, I can attach my figures in next message. Also, for more precise results, you can discuss with me more via this email address: mehdiasgh@gmail.com . Thanks a lot for any suggestion, again. 
Also, I think that this problem is in the turbulent regime since Grashof number is about 4.54E+11 (turbulent regime starts in grashof number approximately 56000). Is there a method for solving this problem by turbulent modelings?;)

When u r simulating Natural convection, water density is given by boussinesq, and also first start running simulation with small value of gravity "g" and then go towards 9.8, start with standard pressure solver and after first convergence change pressure solver to body force weighted. I hope it will work.

All times are GMT 4. The time now is 14:40. 