||December 4, 2010 17:17
I wanted to ask a couple of stuff related to unsteady modelling: I'm doing the exact same thing you discussed; I first get a converged steady-state solution, then I run it unsteady. Here's what happens though:
- I'm analyzing the wake of a 3D cylinder in cross flow. Since for most Re numbers (including mine) the wake is unsteady, so the contour plots of variables oscillate inside the wake.
The funny thing is, for the first 90 - 100 time steps, the wake oscillates, but after that point, and even when the residuals go down and flatten out at low levels, the wake oscillation disappears. The wake looks symmetric, like the average, mean version of all the oscillations.
Any idea why this is happening? Could it be related to the size of the time step and/or the number of pseudo-time iterations at each time step? Or could it be related to the boundary conditions?