CFD Online Discussion Forums

CFD Online Discussion Forums (
-   FLUENT (
-   -   Natural Convection problem in Fluent - urgent (

NSV July 24, 2007 01:02

Natural Convection problem in Fluent - urgent
Hi all, Thank you for your time. I have a canister filled with oil. Few solid components are immersed in that oil and they are dissipating heat(temperature is given to few surfaces and heat flux values to the rest). If I want to model the natural convection in oil using Fluent, what is the procedure or how should I go about it? This is my first project, so kindly help me please. Thanks. NSV

Joe July 24, 2007 05:50

Re: Natural Convection problem in Fluent - urgent
tutorial 5

Raj July 24, 2007 08:22

Re: Natural Convection problem in Fluent - urgent
Hi Joe, Thank you for your interest. In the tutorial he mentions the Rayleigh number and then says that the 'g' value is adjusted. But g is same everywhere. The surface temperature of one component is 343k and the other component is disspiating 400w/m2. I gave g value as -9.8m/s2. The continuity residuals are increasing rather than fall as it iterates!!! I am confused here. Please help me. thank you. regards Nsv

AJ July 24, 2007 08:42

Re: Natural Convection problem in Fluent - urgent
Sometimes natural convection problems require special treatment to get convergence.... Try with g=1 m^2/sec and get a converged solution and then increase g to 9.81 m^2/sec to get the final convergence....

Sometimes convergence could be related to your mesh density etc..

You could also switch the relaxation parameters for pressure and continuity ( 0.3 and 0.7 to 0.7 and 0.3)

Hope it works...


Nsv July 25, 2007 00:04

Re: Natural Convection problem in Fluent - urgent
I have changed the g value and the relaxation parameters and the continuity residuals are just going up and there is no change in the pattern. What also bothers me is even as the residuals are increasing, Fluent is not showing any error message. velocity and energy residuals have more or less remained constant, but the mass residuals keep increasing with each iteration. any ideas or thoughts? Thanx in advance.

MW July 25, 2007 16:07

Re: Natural Convection problem in Fluent - urgent
I know people says this alot but.. what does your mesh look like? Have you got a picture? Also, are looking for steady state or a time dependent solution? From my recent own experience, natural convection problems can be quite tricky.

Olfa April 23, 2014 08:57

Natural convection

I'm working with Fluent for the first time. I began studying the example of the natural convection from a heated horizontal cylinder. But I have a problem with the convergence of the solution. It fluctuates even after 200.000 iterations and doesn't converge.
The upper surface of the test cabin is defined as pressure inlet type boundary condition. The laterla and bottom surfaces of the cabin and the surface of the cylinder are defined as wall type bondary condition. Pressure-based solver is used and second upwind scheme is selected for the Momentum and energy equations. PRESTO! is used for pressure. SIMPLE is used for the pressure-velocity coupling method. I used Hybrid initialization. The problem is the same for tri and quad meshing.
I have to note that I use decoupled methode SIMPLE.

Can you please help me .


CFD-fellow April 24, 2014 07:19

Try pressure velocity coupling with SIMPLEC.
Also some academic study about natural convection is needed to find out how some fluent tutorials converge natural convection problems.

Best regards

enayath May 5, 2014 13:35


Could you fix it? I have a similar problem.

Olfa May 6, 2014 04:21

Hello Enayath, i tried to fix it. The best solution now is to reduce the under-relaxation factor of momentum, it's like to force convergence. I used the the standard initialization with velocity (y=0.001). It's recommended in this link
But the problem that i don't understand how can I set the lenth, the depth and the area in the reference values.

thank you

Olfa May 6, 2014 04:25

I forget to tell you that I work with the compressible fluide not any more with Boussinesq approximation without specified density.

All times are GMT -4. The time now is 18:35.