I have asked a similar question before, but now I need a touch more detail. I am modelling a ships hull in Fluent, both 2D & 3D. I can setup the steady state multiphase problem with region adaption and patching and I have seperate inlets & outlets for the initial water volume (a box with an inlet face with an upper & lower half, the upper is air in and the bottom is water in, the bottom inlet gets a velocity and a fraction of one, gravity is on), but the solution never seems to converge & if I get results when the error gets low, I get nonsensical results (such as all the water behind the vessel seeming to drain out). About the only time I get good results is within the first 20 iterations (then it looks realistic), but I don't know if I can trust those results to far. Is there any way I can improve the accuracy of the model and solution.

Depending on the multiphase model you are using, the problem needs to a transient set-up. I am working on a problem that appears steady; however, after some work the transient solution works much better.

Then how do you keep a continuous flow moving past the body. Whenever I tried a transient solution, the volume of water rushed past & I got a half second of data with the full effects noticable, then the water was gone (as you can tell, I don't have a lot of experience with this type if simulation).