CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   STAR-CCM+ (https://www.cfd-online.com/Forums/star-ccm/)
-   -   Natural convection in an infinite vertical cavity (https://www.cfd-online.com/Forums/star-ccm/91957-natural-convection-infinite-vertical-cavity.html)

 Zeppo August 28, 2011 06:39

Natural convection in an infinite vertical cavity

3 Attachment(s)
I need to simulate natural convection flow between two differentially heated, infinite vertical walls. The problem is two-dimentional, Rayleigh number is 1.01E+5.
Attachment 8989
The simulation file (created in CCM+ 4.04) can be downloaded from http://www.sendspace.com/file/bwkssb
Analitical solution is stated in Benjamin Gebhart - Buoyancy-Induced Flows And Transport. Look at temperature and velocity plots
Attachment 8988
Reference density is located on the left side in the middle between top and bottom sides. It's calculated according to ideal gas law.
Attachment 8991
Physics
http://img3.imageshack.us/img3/9256/pic04i.png
I wonder if it's correct to model infinite walls applying outlet pressure boundary conditions on the top and bottom sides. Is in OK? The point is that I can't make in converged (the residuals are:
Attachment 8987

 ping August 29, 2011 02:21

buoyant flow is naturally quite unstable and you might never achieve a steady solution. You need to set some monitoring points and dump some scenes to see what is changing every few iterations. Then try the unsteady solver with a reasonable dt - start with .01s maybe.
Also depending on you density range, you probably don't need ideal gas - try gravity + Boussinesq

 Zeppo September 2, 2011 12:35

4 Attachment(s)
Quote:
 Originally Posted by ping (Post 322015) buoyant flow is naturally quite unstable and you might never achieve a steady solution.
I tried increasing dynamic viscosity tenfold (resulting in decreasing Rayleigh Number (Ra) to 1E+04) and solution was achieved.
Velocity vector scene
Attachment 9055
Velocity plot
Attachment 9056
Temperature plot
Attachment 9058
Values are distributed relative to the center axis symmetrically so that they correspond with analytical solution. Bringing Ra back to 1E+5 results in solution to diverge. :confused:

Quote:
 Originally Posted by ping (Post 322015) Also depending on you density range, you probably don't need ideal gas - try gravity + Boussinesq