SimScale CAE Forum

Von Karman vortex street - LES

I am trying to study the conditions (geometry, dimensions, velocity, ETC) required to produce the vortex shedding known as the Von Karman Street and have used this as a template but it takes too long to compute.

I have tried to mesh a smaller working space but have only gotten laminar flow so far. Currently I am under the impression that a curved inlet, like Ali’s template, is required to produce the shedding I desire.

My mesh has been named mesh 2 and simulation parameters are under “my attempt” found here: https://www.simscale.com/projects/kirin/large_eddy_simulation_-_flow_over_cylinder_validation_1/

thank you for any feedback.

Hi, @kirin, the inlet shape shouldn’t matter. I looked at you ‘my attempt’ setup, I presume you are just checking setup here? for vortex shedding, you need a transient simulation, which ‘my attempt’ isn’t. The first mesh is very good but for a transient is very expensive, have you tried running your second mesh as a transient simulation? This will be almost 10 times quicker than the first mesh per iteration and also allow you to run on fewer cores for efficiency. Since the cells are larger you might even get longer timesteps if you are using auto time stepping. Also on the second mesh, you may have to refine the wake to allow a detailed capture of the vortices.

Kind regards,
Darren

@1318980 thanks for the reply. I ran a short test changing the simulation to transient and from what i could tell the flow was still very much laminar.

I found another similar simulation by vardhan17 and i noticed that he used a laminar turbulence model, had the boundary layer turned on and was using icofoam as the solver. im pretty sure the boundary layer will affect the results but would there be any change with the turbulence model and the solver?

@kirin, you should definitely consider inflating boundary cells yes. I think we need to consider these questions:

  1. What oscillation frequency do you expect
  2. Are you giving enough time for flow to develop (oscillations build up over time reaching a kind of false steady state first before starting to oscillate then shed)
  3. is the mesh actually fine enough to capture the vortices?

if the mesh isn’t fine enough the turbulence model may be averaging it out.

Kind regards,
Darren