Unfortunately I don't have useful advice for you, but I just wanted to chime in and say that I also have the most luck doing pressure-pressure boundary conditions for this type of analysis.
BTW, have you tried creating a custom gas that mimics saturated steam and running that? That way you don't have to worry about the phase change.
Thanks for the replies. I have just now run the model using water (liquid). The system does indeed converge - although at crazy levels of performance - 120,000rpm may be just a bit high for liquid water :-) torque is a few orders of magnitude above what can be provided, BUT it seems to prove that the physical model etc. is valid.
It does therefore suggest that the modelling of Steam (real gas) has some discontinuities.
The plot thickens .. see this extract from the help file:-
This is how the system responded to liquid water:-
Because your model seems to run somewhat OK on air, I would have to conclude that there may be an error in your steam table in your database. You can also try running with water as a real gas (the Flow database lsits water as a real gas, not sure how this should be different than steam), or R-123 as a real gas as R-123 is commonly used in centrifugal compressors. If your program is still not converging, you need to contact your reseller to obtain SW opinion of issue as the solver does not seem to like steam.