This issue has to do with the initial waterplane height for the free-float iteration. When you specify a weight (and/or cg) for the hydrostatics, the solution becomes iterative in which the solver must balance hydrostatic forces. The iteration has to start somewhere and by default we choose the mid-height of the bounding box of the selected surfaces. Without appendages (e.g. a keel) this is usually a reasonable guess, but with appendages it can cause the solver to fail. You can override the initial plane height by checking the box “Override initial plane height for free float iteration” and entering a value that is reasonably close to the resultant waterline.