CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Main CFD Forum (
-   -   inflow condition (

Feitoudaer March 21, 2007 14:26

inflow condition
Can anyone provide me a good (preferably simple as well) method to generate inflow conditions for spatial jet simulations? I am currently doing LES of turbulent planar jets. I am trying to validate my code with the experimental data of Gutmark & Wygnanski (1976) on planar jets. The inflow condition seems a big problem for me. Things like jet growth rate, self-similar location are influenced a lot by the inflow condition. Thanks a lot for your comments.

Andrew March 21, 2007 17:02

Re: inflow condition
Can you tell me what the state of the boundary layers are as they depart from the exit of the nozzle?

Feitoudaer March 21, 2007 17:45

Re: inflow condition
I am using a hyperbolic tangent profile,

u = (U1+U2)/2 + (U1-U2)/2*tanh(y/theta/2)

for the mean streamwise velocity, and a simple sine wave for the streamwise disturbance,

u' = Sum[An*sin(2*pi*fn*t)] (Sum on n=0,1,2,...)

where f0*theta/Uc=0.033 (the fundamental mode; Uc=(U1+U2)/2), f1=f0/2, f2=f0/4, ...

The lateral disturbance is random numbers, and the spanwise (periodic) disturbance is also sine waves with random phases.

Currently I am imposing the disturbances at 5% on all the three components of velocity.

Andrew March 22, 2007 09:15

Re: inflow condition
Hmm, using inflow profiles of a hyperbolic tangent nature does tend to lead to poor results in terms of things like fluid entrainment and mixing in shear layers - if you have access to an Athens account, look at McMullan, Gao, and Coats (2007) in Int. J. Num. Meth. Fluids for a comprehensive dataset on inflow conditions for mixing layers - most of the data in that paper will apply to jets too.

Forcing the flow with fluctuations that are derived from Linear Stability Theory tends to impose a certain pattern of growth on the flow. Using randomised disturbances by themselves allows the shear layer to develop in a much more natural fshion.

Feitoudaer March 22, 2007 11:21

Re: inflow condition
Thanks a lot for your comments.

I read carefully on the inflow condition part of the paper you recommended. They concluded that the boundary layer profiles should be used instead of hyperbolic tangent profiles to get good results. In their case, they run a precursor boundary layer simulation to get the inflow data for the mixing layer simulations. This is out of my capability due to both of the time and deficiency of the experimental data.

However, I believe many people are still using various hyperbolic tangent profiles as the inflow streamwise velocity profile for spatial jet simulations. I think the point is that mean profiles may be less important than the fluctuation parts, which provide the mechanism for trasition and breakdown.

The final paragraph is quite interesting to me. Are you suggesting I should also use a white noise profile for the streamwise perturbation and let the flow itself to pick up the important components? Thanks.

Andrew March 22, 2007 11:34

Re: inflow condition
As the author of that paper I referenced, I can tell you that analytical Blasius profiles make no difference compared to time-dependent precursor simulation inflow data, as I have tested this myself.

The most important parts of the inflow condition are:

1) Resolving the dominant lengthscale of the flow - in this case, the momentum thickness of the boundary layers that exit the jet nozzle.

2) Providing a disturbance environment that allows a natural selection of the dominant instability modes - frequencies imposed from Linear Stability Theory effectively force the layer into a particular pattern of development, whereas random noise allows the most excited frequency to be selected by the flow itself - see the thesis be de Bruin for more information (

I'm going to dig out that Wygnanski paper from the library and see what the relevant parameters are. I should be able to tell you more about it after I do so.

Feitoudaer March 22, 2007 12:47

Re: inflow condition

It's a good and simple indication for me that "resolving the dominant lengthscale of the flow" is important. In this sense, if I set the momentum thickness to be 0.05, as in my current case, do you think that the grid spacing around the jet shear layer should be 0.05 or smaller? This seems a quite strict requisite for the resolution of a LES.

Andrew March 25, 2007 16:09

Re: inflow condition
I've had a read through of that 1976 paper, and I did not find any mention of the momentum thickness of the boundary layer as it left the jet orifice. How did you arrive at a value of 0.05?

All times are GMT -4. The time now is 05:02.