I remember many years ago doing a blending test with 2 different refrigerants in a tank (unfortunately it has been so long ago that I forget which two were used). Anyway both refrigerants were charged into the tank as a liquid, and one was specified as having a lower boiling point then the other.

After both refrigerants had been charged into the tank, we saw a static pressure that seemed fairly close to what was expected for the lowest boiling point refrigerant (based on the tank temperature). Then we shook the tank. This is the amazing part... as we shook it, the static pressure actually dropped lower then what it was to begin with, and was now at a point somewhere between the two refrigerants as compared to a temperature/pressure chart.

We knew we had charged in sufficient liquid of both refrigerants to still have a saturated liquid present. And we also knew that by shaking the tank that the 2 refrigerants would be encouraged to blend with each other. But to see the pressure drop below the pressure/temperature properties of the lowest boiling component was quite a surprise to us. It was as if we had created an entirely unique refrigerant.

Perhaps we had!

Anyway I tell this story to highlight the unusual effects that can occur when blending refrigerants in an autocascade. This is why it can be so tricky to design the proper blend to fullfill the application. There are programs and tables that will describe the properties of two blended refrigerants. But when you get into 3 or more, I only know of one person that has ever created a program to accurately model such a thing, and he isn't sharing it (and unfortunately it's not me).

The best initial approach I can suggest (and could perhaps be considered a rule of thumb), is to try to pick refrigerants with a difference in boiling points of approximately 50C (+/- 10C). Of course an autocascade will work even if these conditions are not adhered to, but the idea is that this should give you the best initial results. The rest comes down to intuition based upon a lot of experimentation (or should we say a "gut" feeling). Some people also call it black magic.