[Dan1788]

Hi,

My project is based on LES of jet flames. To generate appropriate turbulent inflow conditions for the jet, I did a precursor LES simulation of a pipe with periodic boundary conditions in the streamwise direction and achieved fully developed turbulence.

My question now is this: After achieving fully developed turbulence in the pipe, is it conceptually correct to take a slice of the pipe and map it to the jet as the turbulent inflow boundary condition for the jet ?? Also do I need to do this mapping at every time step or just for time = 0 is enough ?

I am really confused with this. Please let me know if anyone can suggest something. Thanks!

[FMDenaro]

Originally Posted by Dan1788

Hi,

My project is based on LES of jet flames. To generate appropriate turbulent inflow conditions for the jet, I did a precursor LES simulation of a pipe with periodic boundary conditions in the streamwise direction and achieved fully developed turbulence.

My question now is this: After achieving fully developed turbulence in the pipe, is it conceptually correct to take a slice of the pipe and map it to the jet as the turbulent inflow boundary condition for the jet ?? Also do I need to do this mapping at every time step or just for time = 0 is enough ?

I am really confused with this. Please let me know if anyone can suggest something. Thanks!

Hello,
have you read this issue in the literature (journals, books)?
The answer to your question is not so simple....
when ensured you get a fully developed turbulence you need to store the velocity field in a plane but for many time-steps to cover a full period (given for example by the lenght of the pipe and the average bulk velocity).
Due to the fact that you have a jet, I think you should better prescribe that as inflow in a small pipe that then generates the jets in the free environment

[LuckyTran]

Originally Posted by Dan1788

After achieving fully developed turbulence in the pipe, is it conceptually correct to take a slice of the pipe and map it to the jet as the turbulent inflow boundary condition for the jet ?? Also do I need to do this mapping at every time step or just for time = 0 is enough ?

Taking a slice and mapping the instantaneous velocity is exactly what you need to do in order to get the right inflow conditions. This means mapping at every time-step.

In general I think your approach is sound, but I like to point out a concern. Fully developed turbulence is not so straightforward to achieve in the instantaneous sense (LES/DNS). To get the correct fully developed turbulence using periodic boundary conditions, the length of the periodic domain must be sufficiently long that the eddies (the fluctuations) are not recycled too soon. The minimum periodic length to achieve converged velocities is roughly 2*pi*delta or 2*pi*r and 4*pi*delta for rms velocities, longer for higher order statistics.

Reference provided: " The influence of pipe length on thermal statistics computed from DNS of turbulent heat transfer" doi:10.1016/j.ijheatfluidflow.2011.09.003

The minimum length can quickly become longer than the length of your jet carrier. Obviously there is no motivation to do LES of a pipe longer than your jet carrier, at the same time it is not technically correct to say that your jet has a fully developed inflow if length of your periodic domain is shorter than this length. Depending on how realistic you want your inflow, it may be better to model the influence using a synthetic eddy approach or to simulate the upstream jet carrier directly using LES without using periodic boundary conditions.