Asymmetric Flow
 

Hi,

I am simulating the compressible internal flow within a curved duct using a RANS-based flow solver. The duct is symmetric wrt the center plane. The solution that I get is rather surprising as the flow after the mid longitudinal station loses its symmetric nature and becomes totally asymmetric downstream the duct.

If you have any idea what really cuases this, please help me out here. I really do not have any answer for that. It cannot be correct I think, so there muct be something wrong in my flow calculations.

Thank you.

Rose

Re: Asymmetric Flow
 

Just to mention that when I said "there muct be something wrong in my flow calculations", I should have said that the code I am using is very robust and has been used in the lab for many years, it works very well for many applications, and my case is actually very close to those applications, I just do not understand why the solution looks unreasonable.

Rose

Re: Asymmetric Flow
 

Are your calculations 2D or 3D? How do your profiles look? Like symmetric profiles that are biased to one side? What kind of RANS Model do you use? Does it have some handling for rotational effects? Asymmetries can occur because of system rotation

Re: Asymmetric Flow
 

My calculations are 3D. The symmetric pattern starts being disappeared from the middle of the duct. If you look at the cross-sections at tdifferent longitudinal locations, the bottom separated region circumferentially moves towards the right side and by the very downstream, it even reaches to the top of the duct; ie, a full 180-degree rotation. The core of the flow however is not affected that much.

The RANS model of the code is: 0-, 1- and 2-equation, although for my first cases, 0-equation model seemed to produce more accurate solutions, so I am now more stick to this model than 1- or 2-eqs, it is also much faster.

System rotation should not be a problem here because if it was, this asymmetric pattern should have started from the duct entry, not from the mid-way.

Any advice?

Rose

Re: Asymmetric Flow
 

Hi,

Such asymmetry can be physical if there is a symmetry-breaking bifurcation of the flow resulting in multiple stable solutions. The only constraint is that the ensemble-average of the realizable flows has to be symmetrical. Do your computations always result in the same asymmetric pattern or sometime in a mirror-reversed image (you could try to change the initial solution, possibly to an asymmetric one to check it)? Moreover, from my experience, such bifurcation are generally related with separation: is your duct curved enough to result in some reverse flow?

Hope this helps,

Lionel