Dear all,

I have reduced the under-relaxation factor, used 1st order scheme, follow every steps suggested by buoyancy simulation from the Fluent user guide, yet failed to obtained a converged solution.

The simulated building block size is 20m x 20m x50m, the closing domain of outdoor air is 200m x 200m x 250m.

Boundary conditions: 1 pressure inlet (top face of the domain) and 4 pressure outlets(4 side faces)

For the mesh, I have produced a mesh with boundary layer growth factor (b/a) of 1.3, started at the first row (a)of 0.03 and the number of row of 4.The interval size of the mesh volume is 10. Thus, a fine mesh near the building wall surface and coarse mesh at the surface of the closing air domain is produced. (For Gambit meshing, I used meshing type T-Grid, and the elements option of Tet/Hybrid)

As the solution did not converge, and so the velocity vectors produced looked unexpected. From your experience, can you suggest me about some important points I may ignore in making a good mesh or boundary conditions? And also, how many iteration steps you usually see when the solution starts to converge from your experience?

Thank you very much

Regards,

Cadrian

which multiphase are you using? are you using any turbulence model? if yes, check your y_plus/ aspect ratio of mesh.. Try with reduced URF..

Dear Sujith,

I am using the Boussineq model of the air. And also I am using k-e turbulence model. And I have tried a under-relaxation factor as low as 0.7 (is that enough?)

Anyway, do you have any suggested value of y_plus/aspect ratio of mesh?

Thank you so much =)

Cadrian

I had similar issues with buoyancy flows... perhaps there is no stationary solution, so you should use the unsteady solver and see what happens!

Hi Cadrain,

try running it transient with say 5 iterations per time step. You can calculte a velocity scale given

Bak_Flow