compressor loss model with other gases

[livecreature]

I am about to design a compressor using co2. but most literatures show that design methodology is based upon working fluid air, in particular, some empirical correlations used in loss model. If I adopt the same loss model developed from air to use for CO2, how confident are I to assure the loss model is valid for this particular working fluid?

Btw, is there anyone who has experience on designing a compressor with working fluid above (near)the critical point? Any comments are welcome?

Thanks,