My rationale is as follows. Of the 3 components, T-9 and LPS 3 provide mechanical protection. T-9 penetrates, but covers with a thin layer. LPS 3 does not penetrate, but provides a thick coating layer. The two together provide a thick protective coat that from the inner side penetrated any cracks in the surface. On a chain, LPS 3 causes the dried-out mixture to behave as a paste. Now ACF-50 penetrates, but its most important role is to provide a chemical protection. For me it works just as well on steel as alu. Without ACF-50, the chain will remain rust-free in a rainy weather for several weeks. With ACF-50, I can pull it to the level of months. Of course it depends on mechanical impacts. Say in winter I may be rubbing the chain against snow and then the rust may appear after 6 weeks rather than 3-6 months. Well, 6 weeks is very good. With any commercial chain lube I never went beyond 2 weeks in persistent wet conditions. One negative of ACF-50 is that the mixture takes a longer time to dry out than without ACF-50.

The reasoning above may sound naive, but the combination of components works like a dream and I have reasons to believe the above is a roughly correct explanation. The latter is from observing what happened if I used just one or a combination of two components. I use the mixture on anything on the bike, including inside of the frame, and I apply it to cars that can experience even harder beating than a bike and that mixture holds there too, preventing rust for months+ on quite exposed elements.

As I mentioned before, I clean no nothing before application of the mixture. I put it right on top of any salt, mud, etc. on the bike, with what seems a complete impunity.

Since I arrived at the mixture through experimentation, combined with some reading, I cannot exclude that one could arrive at a still better mixture of this type, e.g. replacing ACF-50 by Eezox.