[pacoelpocero]

Hello everyone,


I am doing simulations on a closed domain. Reynolds number is in the transition region (around 2300) so I am using a laminar model with a fine mesh. The problem is that the stationary is unstable, and the values are oscilating around an average value. Due to the mesh, it is imposible to me to make a transient simulation, so I am asking if there is any way in FLUENT of getting the average value of the monitoring parameters.



Also, I would like to know if it is possible to make a selective frequency damping in FLUENT, as I would like to add an extra term to Navier Stokes where the diference between to the calculated the velocity and the average velocity of the latest iterations is dumped by a factor.


Thank you for your attention.

[Starcatcher]

Hi,
The oscilation can have various reasons.
1. Your Mesh around inhection is too coarse.
2. What iteration scheme with what order are you using? Is there a pseudo time step, or cfl numer? Then it might be too small.
3. Also the fact that you are around transition-Re makes me suspicious too. I'm not sure if a laminar model can handle this operational point well. But I do not see your geometry.

I would try to vary the cell size around the inlet boundaries.


For this extra Termin, I would like to put your attention on the under relaxation numbers in the gui. Maybe they are enough for your purpose.

[pacoelpocero]

Originally Posted by Starcatcher

Hi,
The oscilation can have various reasons.
1. Your Mesh around inhection is too coarse.
2. What iteration scheme with what order are you using? Is there a pseudo time step, or cfl numer? Then it might be too small.
3. Also the fact that you are around transition-Re makes me suspicious too. I'm not sure if a laminar model can handle this operational point well. But I do not see your geometry.

I would try to vary the cell size around the inlet boundaries.


For this extra Termin, I would like to put your attention on the under relaxation numbers in the gui. Maybe they are enough for your purpose.

1. My actual mesh is of 15 Million elements, I supposed it is fine enough, nevertheless, I have to other finer meshes in case of necessity.
2. I am using a Coupled scheme with CFL of 20. For pressure I´m ussing PRESTO! (as it is a mixed convection problem) and for the other variables I am using a second order discretization. Under relaxation numbers are 0.5 for velocity, pressure, density and temperature.
3. It is a rectangular duct with a bent in the middle, U-shaped.