Does anyone know any tricks to get my solution to converge for a long period (~15 secs) sloshing flow? The excitation of the domain is just some A * sin (2*Pi/T * time). If my period is long then the solution (moment time history) doesn't look smooth. If I shorten it then things look great and match theory. Any suggestions?

By period do you mean the flow time? In this case the moment time history couldn't look smooth because of non-linear effects due to the interaction between the frequency 1/T of your excitation and the natural frequency of the fluid in the tank. As long as your solution converges well (preferably within 10 or less iterations) every time step, I never found unrealistic results...

Hope this helps

Edi.