hi

I am working at the moment in modeling flow around cylinder using Ansys Flotran package FE, the flow of Re 200, the fluid is water. I face a problem of the wave length as i cant get a close measurement comparing to experimental, the vortices more of double street shedding, the wall of the pipe and the cylinder are no-slip boundaries, the uniform velocity of 0.04 m/s and the cylinder is shifted 1 mm from the centre, the pipe diam is 32mm, i have tried flow with un-symmetrical to give it "kick start" but again getting the same results of nearly double the length of the wave. also longer time running, smaller time step of 0.1sec or some times smaller but no results, please sir could you advice me on this matter. thank you very much regards mahmoud

(1). If I were you, I would repackage the program and send it back to the vendor for a refund. (2). For such problem, you shouldn't have to use a commercial code at all. They should have used the problem as a benchmark test case before they ship out the code. (3). It shouldn't be very difficult to write a 2-D code at this Reynolds number. Well, that's my personal opinion. I hope that you are doing everything right.(w.r.t. the code setup for your problem)

A wrong frequency in the wake of a cylinder can be obtained in the simulations when the boudaries are not treated properly. What happens is a reasonance phenomenon. This might be the case, but there could be other problems. Changing grid resolution and the Reynolds number might help to track the problem. Sometime ago a similar question was raised in the cfd-forum. This problem might therefore be 'common'.

If you think that the bounday conditions might affect your results, then check:

Abarbanel et al. 1991, Journal of Fluid Mechanics, vol. 225, page 557.

PG