[enayath]

Hello,

I am trying to model natural convection in a closed domain. I use K-e/realizable and with enhanced wall treatment and with fully buoyancy feature on. 2 problems exist:
1- I cant not set turbulent BC for my BCs...the turbulent feature is not activated...Do I need to change my BC type from wall to other thing?
(now all of them are wall-axis of symmetry and interior)

2- When I start to calculate with current settings, fluent runs but the continuity residual is in order of 10e4!
energy 10e-10,K,epsilon,x and y velocity 10e-6,

Any ideas how to handle these two problems?

Thank you

[Bionico]

Hello enayath,
can you post a picture of your problem?

1) If you have a closed domain (a room, for example) you shouldn't define turbulence BC;

2) Natural convection problem usually are difficult to converge, check your mesh first of all

Regards

[enayath]

Hello Bionico,

Thanks for the reply. As can be seen, the domain is closed...so for your first note, what should I do for the BC? How Fluent can detect the type(laminar or turbulence) if I do not specify the BCs? just by setting laminar or turbulent in models I am done?

About the second point, I am pretty sure the quality of mesh is not bad.

Minimum Orthogonal Quality = 5.83808e-01

Maximum Aspect Ratio = 4.65313e+00

Max Skewness:0.64

The values of Y plus is less than 5 which means the mesh is pretty good for heat transnfer calculation.right?

Any other idea how to limit the residual for the continuity?

Thank you,
Hooman

[Bionico]

Hello enayath,
when you define the turbulence model, Fluent solves the equations of Fluid dinamics with turbulence parameters. So if you have an open domain you can define turbulence intensity (for example) in inlet; in closed domain instead you can define BC as Wall Temperature, Wall flux etc.
It's difficult to converge natural convection in closed domain: I suggest using boussinesq approximation for density in Material Menu and RNG k-epsilon as turbulence model.
Moreover you should set up a transient simulation because it converges better in such a case.

Regards

[enayath]

Bionico,

I do not want to use boussinesq approximation as the temperature difference is not small(the minimum is 30) and based on what I searched for maximum deltaT=20, we can use boussinesq approximation. I want to give the unsteady answers as the initial for transient calculation. I tried also for transient.it is still high for the continuity residuals.

Hooman