Separation
 

I use CFX 5.6 to solve one problem with bouyancy effect. The circulation is expected, then I use 'Mesh Control' for fine grid in the expected zone. After I got the converged solution. I used vector plot and found the circulation in that zone. Then I refine and expand the radius of that 'Mesh Control.' This time I got the overflow error. I try to use finer and expand the radius and then get the overflow error again and again and again. I suspect that there might be the seperation there. Can CFX deal with seperation automatically or do I have to set anything? and how do we check in CFX if there is the separation? Thank you very much.

Re: Separation
 

Hi Atit,

CFX can handle seperations just fine. They are a fundamental part of just about any fluid flow.

By overflow error I assume you mean the linear solver fails rather than running out of memory.

If your flow is steady state have you tried mesh adaption? That can be very helpful in refining the mesh where it is needed.

Regards, Glenn

Re: Separation
 

Dear Mr.Horrocks, Thank you for your comment. Yes, you are right. It is the fail of linear solver. In addition, my case is steady state, and I use 10 step of mesh adaption. The adaption variables are velocity, pressure, and density. The error always occurs after a few step of mesh adaption.

Atit

Re: Separation
 

Hi Atit,

Do you mean 10 remeshes in mesh adaption, or 10 levels of mesh adaption? 10 levels of mesh adaption steps is a lot and will probably cause problems due to the mix of very small and very large elements.

Also, why are you refining based on velocity, pressure and density. Generally it is best to choose only one variable to refine on. If you are doing a compressible simulation then density would be a good choice. If an incompressible simulation then pressure or temperature.

Regards, Glenn

Re: Separation
 

I mean 10 levels of mesh adaption. My problem is about natural convection of air. The air density is changed in some parts of the domain. In one part of domain, air flow through elbow. In this part, while density is only slightly changed, the velocity is changed siginficantly, the circulation also occur and the separation may be occured. Should I use the density as the adaptive variable? I afraid that the adaptive mesh, with density as the adaptive variable, cannot capture the phenomena at the elbow region. I think velocity should be chose as the adaptive variable, shouldn't it? Thank you very much.

Re: Separation
 

Hi Atit,

In that case, yes, velocity sounds like the best variable to use. Also, I would reduce the adaption levels. Ten is too much. Between 3 and 5 is more reasonable.

Glenn

Re: Separation
 

Thank you so much. I will try 5 steps.

Atit