As a conclusion, we would put the emphasis on the fact that StarCD is quite hard to learn. The help is very poor and does not really help the reader. Luckily, there is a guide that is well done and that help through the different stages. It becomes hard when you don't know what is the aim of an option.
As for the results, it seems that StarCD provides good results and that it has a large amount of possibilities. We haven't had the time to check everything in detail, but the options seems to be very complete.
Once you get the basis, then you can easily make your simulations and the calculation time is not  longer than the one in Fluent.

Eventually, StarCD seems to be a good choice, even if it is hard to drive at the beginning.
In the future, it could be interesting to make the same simulations using a droplet model. As a matter of fact, we have been surprised to see the amount of phenomenon taken into account for a droplet simulation (heat exchange, vaporization...).