Turbulence models become invalid at Re=2000~5000?
 

Recently, i use Fluent to simulate jets and two opposed jets and s strange resut occurs: when Re=>10000 and the inlet velocity is more than 10m/s,the simulation results are acceptable and in agreements with the experimental results, but at Re=2000~5000 and the inlet velocity is about 2~5m/s, all turbulence modles (including KW,SKE, RKE and RSM) give unsatisfactory results and the simulation will need more time to converge. At about Re=4000, the turbulence intensity at the nozzle exit is high to 5% when i use hot-wire to measure in the experiment. This question puzzled me for a long time and i post it on this forum to seek helps. My questions are: (1) If and why turbulence models of Fluent become invalid at Re=2000~5000? (2) How to simulate the turbulent flow at Re=2000~5000? which turbulence model is more valid? Warmly welcome your answers, thanks!

liweif 2007-8-24

Re: Turbulence models become invalid at Re=2000~50
 

http://www.google.com/search?hl=en&q...ent+transition

des/les my offer better performance, but tt is a sticky field.

Re: Turbulence models become invalid at Re=2000~50
 

Thank you, Joe! But I still have a question: commonly,at Re=2000~5000, the flow is in the region of turbulent transition, but in fact jet of Re>300 is a turbulent jet(Hinze O,Turbulent),and the turbulence intensity is high to 5%. So, for jet of Re=2000~5000,the flow should be fully-developed turbulence,why Flunt can not predict validly this flow?

Re: Turbulence models become invalid at Re=2000~50
 

Hi

I'm currently working on negatively buoyant (cooled) jets flowing down a wall with Reynolds numbers of the same order of magnitude as yours (4000). The problem seems that in the transition flow from laminar to turbulent regimes, the turbulent models are less efficient. If you are in the border line, you might encounter problems with the turbulence models. I finally used the DES model with the standard wall function, which seems to give satisfactory results. Is your case a free jet or a wall bounded one?

Re: Turbulence models become invalid at Re=2000~50
 

Thanks, AAA! My case is free jet or two opposed jets and 2D axisymmetric grids and pressure outlet boundarys were used. Can you tell me more about DES? Is it used in 3D grid and is unsteady resolution?

Re: Turbulence models become invalid at Re=2000~50
 

Hi

Simply put, DES (Detached Eddy Simulation) is a new hybrid turbulence model within Fluent 6.3 that uses k-e (or k-w) with wall functions near walls and LES for the rest of the domain. The advantage of it is that it is less computationally extensive (but less accurate) than LES and more accurate (but more computationally expensive) than the two-equation models. Most importantly for your case, it uses the "realizable" k-e model, which is best suited for axi-symmetric jet flows (according to the manual)! And actually, it can only be used in a 3D domain and the simulations must be unsteady (i.e. you can not use it for steady simulations). Give it a try, you cant go wrong!

I have a question for you. Does your study about jets involve finding the "entrainment rate" into the jet?

Kind Regards

AAA

Re: Turbulence models become invalid at Re=2000~50
 

Thanks AAA for your helps! I have computed entrainment rate from the simulated results and experimental measurements. Any question?

Re: Turbulence models become invalid at Re=2000~50
 

Could you breifly explain how you managed to do that in both methods (numerically and experimentally)? Is there an informative book/article in this regard?

My jet represents a cold air curtain of an open display case, which is a negatively buoyant flow. The way I'm finding entrainment now is by creating a surface near (parallel to) the jet and finding the perpendicular velocity vectors crossing it coming from the surrounding air. This, along with the area of entrainment and the average density allows me to find the mass flow (m=rho*A*V). But this is not accurate enough. I have to deduct the spillage, which can not be precisely found.

Thanks in advance.

Re: Turbulence models become invalid at Re=2000~50
 

My method seems similar to you and i calculate the entrainment rate like this: select a plane parallel to the jet exit and a line from axis to the outside and find the point at axial velocity ~=0, then integrate the flux in the plane in which axial velocity is >0 (the velocity has the same derection as the exit velocity is defined >0), and this flux is denoted by Qx, So (Qx-Q0)/Q0 is the entrainment rate, where Q0 is the flux at the jet exit. This method is used to engineering field and i don't know if it is accurate enough to you?

Re: Turbulence models become invalid at Re=2000~50
 

Thanks for the reply,

The equation you are using doesn't take into acount the Turbulence Intensity or the shape of the initial velocity profile of the jet, which I'm trying to establish. I'd be grateful if you point me to any engineering book. I'm thinking about using the vorticity concept to specify the line of entrainment. Are you aware of any book that talks about this in detail?

Kind regards

AAA