I should also say that phase transitions are used to describe the abrupt change in physical parameters as some external conditions are changed. Cellular automata to my knowledge don't really have continuously variable parameters.
> "In contrast to a well-mixed chemical reaction system, reaction–diffusion (RD) networks can produce spatial patterns, where some species are more abundant in some parts of the substrate and less abundant in others."
"Emulating cellular automata in chemical reaction–diffusion networks"
Long range order is sometimes exhibited by cellular automata, but cellular automata is too general a framework much of the time to be a useful calculational tool.
In other words, they do know of it, and the formalism doesn't make these kinds of predications to my knowledge
https://en.wikipedia.org/wiki/Cellular_automaton
EDIT: Appreciate the feedback from you two. :)