Hi,

I use one high Reynolds K-e turbulence model to simulate a free jet behavior with outlet velocities between 2m/s and 20m/s. the results is very strange. the jet behavior in the centerline velocity decay is almost the same, but according to the experimental results, when the outlet velocity is lower than 6 m/s, the centerline velocity will decay faster and faster. can somebody tell me is that because I use the wrong model?

BR

No, because the your conclusion was based on the statement " according to the experimental results,...". So, it is a good idea to identify "the experimental results" first and look into it. I don't think there is such thing as a " wrong k-epsilon turbulence model".

Hi,

I thought of the following things you can try:

(1) Check the Reynolds number of the simulation to that in the experiment. Is the flow laminar (steady/unsteady), transitional or turbulent??? (2) Try to do a laminar flow calculation. (3) Check the inlet turbulence intensity and length scale. (4) Also I would think that a change in flow behavior (below 6m/s) should be due to some transition in flow charachteristics. Can k-e model handle that?? (5) Try other source of experimental data. (6) In jet simulations, free boundary (their extent and boundary conditions) play an important role.

In short, are you computing what you have experimented???

Good Luck !!!!!!! Anil

Hi,

As a new CFD user, I was very very appreciate everybody's comments.

I have try the laminar simulation, but the results is worse--the centerline velocity decay will become very slow. It could because in laminar flow, the energy diffusion speed is not so fast.

One guess for why the centerline velocity decay become faster when the outlet velocity less than 6 m/s is because the disturbulence of room air. but how can I simulate the disturbulence of room air?

BR

It is hard to answer your question unless we have some ideas about the conditions you are using, such as the Reynolds number based on the jet exit velocity, the diameter, and the total number of mesh points in the radial direction, the initial potential core region and the farfield. And also, before you try to address the problems, it is a good idea to obtain the grid independent solution first. These items may be relatively new to you, but I think they are essential to your problem. The jet speed alone is not enough to know exactly what you are doing.

Hi,

Thanks a lot for your imformation.

in my case, the diameter of outlet is from 4 cm to 16cm and outlet velocity is form 2m/s to 20 m/s. can you recommend me some paper about such kind of free air jet simulation?

BR

There is a good reference on the subject:"Injection and Mixing in Turbulent Flow" by Joseph A. Schetz, Professor and Chairman of Aero and Ocean Engineering Dept., Virginia Tech. Blacksburg, Virginia,Published as Volume 68, Progress in Astronautics and Aeronautics by AIAA in 1980, ISBN 0-915928-35-3. It covers various types of jet and wakes, theory, data and a complete list of references. It is very important to fully understand the flow behavior before attempting the CFD solution. Try to plot the actual velocity field in real scale first and then develop a mesh arangement to provide a good representation of the velocity profile. You need highly stretched mesh in the radial direction as well as in the axial direction. This is very important when using Navier-Stokes solver (vs boundary layer solver).

Hello

I have experienced the same kind of behavior in the study of jet flows. It showed that the potential core wanished faster when usin unstructured mesh then when using structured mesh. This could be a result of the greater false diffusion occuring when using unstructured mesh on a flow with a very dominating velocity component.

Unfortunately I havent had the time to investigate it further, but I think that false diffusion could be the reason for the "missing" potential core.

There is some pretty good experimental data in 'Boundary Layer Theory' by H. Schlichting (McGraw Hill, 1979), that you can use to compare your data.