CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (http://www.cfd-online.com/Forums/openfoam-solving/)
-   -   How to define the turbulence intensity and mixing length at an outlet (for k-eps)? (http://www.cfd-online.com/Forums/openfoam-solving/72171-how-define-turbulence-intensity-mixing-length-outlet-k-eps.html)

 david39 January 27, 2010 14:14

How to define the turbulence intensity and mixing length at an outlet (for k-eps)?

Hello,

I am new in the OpenFOAM world so please apology if my question is simple.

I have a model with an inlet defined as total pressure inlet boundary condition and an outlet defined as pressure outlet boundary condition.
The aim of the simulation is to determine the massflow in a pipe.

For the inlet boundary, I have used the type "turbulentIntensityKineticEnergyInlet" for 0/k and "turbulentMixingLengthDissipationRateInlet" for 0/epsilon in order to define the turbulence quantities at the boundary.
Thanks to this boundary condition type, the turbulence quantities depend on the velocity at the inlet (massflow).

At the outlet, I have backflow. I already have added an extrusion to limit the backflow but there are still recirculation at the oulet boundary.

For the moment, I use a "inletOutlet" to define the turbulence for the backflow but k and epsilon are constant.

So, for the outlet, I would like to define 0/k and 0/epsilon in a similar way as at the inlet (by using the turbulent intensity and the mixing length) to make the backflow turbulence depending on the backflow-massflow but for the moment, I have not found how to define the turbulence at the outlet.

How should I define 0/k and 0/epsilon for the outlet in order to specify a turbulent intensity and a mixing length for the k and epsilon calculation ? Is it possible in OpenFOAM?

It would be nice if someone could help me.

Thanks,
David

 olesen January 28, 2010 04:25

Quote:
 Originally Posted by david39 (Post 243946) How should I define 0/k and 0/epsilon for the outlet in order to specify a turbulent intensity and a mixing length for the k and epsilon calculation ? Is it possible in OpenFOAM? It would be nice if someone could help me.

This does not work directly out of the box. You'd need to roll your own boundary conditions that would look like "inletOutlet", but calculate k/epsilon along the same lines as your inlet condition.

But the fundamental question: if you have backflow, how can you know a fixed value for the turbulent intensity and mixing length a priori?

 david39 January 28, 2010 13:49

I agree with your remark : we cannot know the turbulent intensity and the mixing length a priori. So the question is to know what is the "best" pratice :
1) To use constant values for k and epsilon ?
or
2) To use an turbulent intensity and a mixing length to calculate k and espilon depending on the velocity of the backflow ?

I will make some tests to investigate the influence of constant values for k and epsilon on the solution results and on the convergence of the simulation.

Thanks.
David

 bb_ September 10, 2010 02:01

Hello David,

have you made any progress on this problem? I guess I'm facing a similar issue. I'm having convergence problems and I suspect that they are due to wron/inappropriate boundary conditions. As U and p should be fine, I'm guessing it has something to do with the turbulent boundary conditions.

Thanks,

bb

 nileshjrane September 10, 2010 05:10

Why do you want to be so specific in giving outlet k and epsilon BCs??? A crude estimate is all what is needed, as the amount of reversed flow is not so significant. A good guess would be: specify k and epsilon as "\$internalField" value. This is generally good enough. If you have large reversed flow which affecting your solution then you better rethink on your domain size.

 bb_ October 13, 2010 07:01

Thanks for your reply. You're right, the workarounds you posted are probably the better ways of dealing with this kind of problem.
For my case, I think the problem was something different, not the turbulent boundary conditions ... now I got everything smoothly running, without changing the domains boundaries.