[soggy316]

I am modeling flow past a circular cylinder in 2d

I have split the fluid domain into two parts - seperated by the box that I have drawn in this image -

Now, I am just validating the mesh for Re=100 before I get down to moving the cylinder.

Now for dynamic meshing i have to use a tri mesh around the cylinder. Now, when I run the case, I observe the following

1. I get a strouhal number exactly 1/2 of what I should be getting if my surrounding domain is tri (regardless of the presence of a quad boundary layer), refining the grid does not help.

2. If I use a quad grid I still get a strouhal frequency 1/2 that of the expected one.

the above two points used 1st order upwind scheme for momentum discretization

3. If i change the quad grid to QUICK discretization, then my strouhal frequency matches perfectly with the expected one.

4. The discretization has no effect on the tri mesh strouhal frequency

5. If in my most coarse tri mesh case, i reduce the time step by 1/10th then my strouhal frequency increases but is still some way off the expected value


Has anyone else experienced similar problems while using FLUENT? or am i missing something really simple?

My parameters are

inlet vel: 1m/s
density: 1 kg/m3
viscosity: 0.01 kg/ms2
diameter:1m
Domain - 32mx16m
Subdomain - 2mx2m