Hello All,

I am trying to get a 3-D case to converge of a laminar flow around a vertical cylinder protruding from a ground plane. From other cases that I have seen the only way to get the horseshoe vortex to appear in the corner of the ground plane and the cylinder you need to have second order accuracy. When I run the case with first order accuracy the residuals become very small (10^-6) and the cl and cd converge to a constant value. If I then switch to second order accuracy the residuals will not get lower than 3*10^14 and the cl and cd oscillate and never converge. Is there something else that I need to do to get reliable second order results. I thought that if I let the first order solution to converge then switch to second order that it should help with convergance. Any help would be appreciated.

Thank you in advance...Rob

Hi Rob,

It does not sound as if you are running a transient simulation? This would be important to capture vortices and other dynamic features. Make sure you choose a time-step small enough to resolve the transients in the flow as well.

good luck...

That is right, I was running a steady state solution. I have yet to venture on to the unsteady time-dependant solutions. Will changing to the unsteady solution make the difference you think? Thank you for your response Mystic.