LES simulation of bluff body at Re(d)=1200
 

Hello Evereyone,
I have been trying to run an LES simulation of flow around a bluff body with rectangular trailing edge at Re(d)=1000.
I have run the same LES simulation for Re(d)=20000 without any problem. Now I am using the same settings for Re(d)=1000, and the resuduals begin to oscillate after only a few time steps. The solution actually diverges after some 150 time steps.
Mesh size is close to 5x10^6. I have tried various time steps from 1/100 to 1/10000 of the biggest time scale of the problem (vortex shedding period).
Smallest mesh size (closest to surface) is d/1000.
My specific questions:
What should I use for initial condition? (right now, I use zero)
Which algorithm should I use? (SIMPLE, PISO, ...)
Any other specific suggestions for low Re LES are highly appreciated.