Hi, I'm wanting to do an LES simulation of flow around a circular cylinder using CFX5.6. Due to the small time steps required for the simulation I was wondering if the following would be a valid option. Start my simulation with a much larger time step - say 2 orders of magnitude, then once I have something representing a steady oscilation of my forces on the cylinder, then drop the timestep back down to the lower desirable level (1e-4sec). The other option I thought of was to use a restart from a transient ke model which I ran and then just use the small time step right from the start. Are there any issues I should be wary about if I adopt either approach ? Cheers for any help Bob

Hi. I am also working in LES with CFX, but using the TASCflow for the same. There are two option available in TASCflow for LES models. viz. Samgorinsky and dynamic model. where as in CFX5.6 there is only one model implementedi.e. Smagorinsky. regarding the part of the initialisation, I also still doubtful over this issue. But firstly starting with a traditional k-e model and then shifting to LES model is a good choice, I feel. So once the flow has traversed through the domain, you can switch to the LES model and use an appropriate time step.

Cheers Sudar I guess its the only reasonable approach, else excessive solution time is waisted. Bob