Learning Forum

[Alex Lee]

I am using ANSYS fluent 2021 r1 to simulate a smooth duct with a turbulence model of LES. But the results (heat flux, pressure drop, wall shear stress and so) are far lower than the values in the journal. Can anyone help my problem?

Here are my simulation steps:

1) Run RANS simulation until convergence (the simulation results of RANS are in good agreement with the journal)
2) Use the command 'solve/initialize/init-turbo-vel-fluctuations'
3) Switch to LES and set the dynamic smagorinsky-lilly Model (dynamic stress and dynamic energy flux) SGS model
4) Change the fluctuation algorithm for the velocity-inlet to vortex method and set the number of vortices to 1000 (I saw from a guide that my number should be set to N/4, but it seems that 1000 is the maximum value for fluent)
5) Running calculations.Besides, I have not changed any boundary conditions or parameters.

[Federico]

How fine is your mesh? LES requires really fine mesh to capture the smaller turbulence structures. 

[Alex Lee]

 

Thank you for reply.

My model size is 2m × 0.5m × 0.125m and the number of grids is 835 × 110 × 70. The x-direction is the flow direction and the y-direction is the spanwise direction. The spanwise direction uses periodic boundary conditions and the y+ in the z-direction is about 1.0. The grids in the other two directions are evenly distributed.

 

[Federico]

What is your time step size and Courant number? Remember you should aim at CFL ~ 1.

You can review this document here for Best Practice: Scale-Resolving Simulations in Ansys CFD | Ansys

[Alex Lee]

 

 

I’m sorry for replying so late. My flow direction grid has a uniform length of 6e-3m, an average flow velocity of 1.771m/s, and a time step of 1e-3s. Of course, for boundary layer grids, this time step may not be sufficient.

I forgot to mention it before that I used the profile file of the outlet parameters of the smooth rectangular channel  simulated by RANS as the inlet boundary condition for LES simulation. Recently, I found that the pressure drop simulated by RANS did not match the experiment. I think this may be the key to my current problem, and I may create a new post to describe this problem in detail.

This is my new post: RANS calculation result does not match the theoretical value (ansys.com). If you have any suggestions for my current (and previous) problems, please feel free to discuss them in this post:

Anyway, thank you for your help. Have a nice day !