If you sketch in the _far_ side of the worm teeth in the D-L drawing, you'll see they're pretty much horizontal, so they can mesh with the spur gear.
I'm not convinced that FEA will help much in analyzing such an assembly, because the contact area is concentrated at the mesh, so the gears will Brinnell and/or wear against each other rapidly at first, effectively changing the contact geometry until the area becomes large enough to support the rated load.
I don't know if an FEA package can help you figure out what that eventual contact shape will be, and it will probably give odd/misleading results regardless of what you guess/assume about the contact geometry.
The D-L product looks like the sort of thing you'd develop by building a few prototypes and trying to map the effective lifetime against the applied load, hoping to find a load at which the wear plateus for a while, and calling that the rated load for a rated lifetime.
Or maybe D-L actually used FEA or some other toolset to help them arrive at a satisfactory design. Call them and ask.
Mike Halloran
Pembroke Pines, FL, USA