Firefighter phase II (in cab operations) can only be performed after phase I recall has been initiated (ASME 17.1 section 2.27.3.3). Phase I (elevator recall) can be initiated several ways--designated lobby detector, remaining lobby detectors, and shaft, pit, and control room smoke detector activation. Fireman can also manually initiate phase I recall by use of their key at the designated lobby outside the cab. Keep in mind shaft, pit, and control room smoke detectors are required to initiate a "flashing" fireman's hat as opposed to lobby detectors which are "steady"--this is to warn firemen that there is more potential danger of using the elevator as opposed to a "steady" fireman's hat, It also warns them that shunting of elevator power is now more probable. They can still use the elevator via phase II in cab controls--but at their own risk.
As mentioned previously, the theory is that smoke detectors being used for phase I recall will detect smoke and recall the elevator and occupants to the appropriate level prior to disconnecting elevator motor power via a heat detector in the shaft, pit, or control room. The heat detectors, in turn, are to shunt power prior to sprinkler head release for a couple of reasons--water on incab elevator controls can cause unpredictable and unwanted operations and also to prevent the operation of those elevators that could be affected by wet brakes caused by sprinkler water release. Electrocution of occupants is not a major concern as elevator motor power is only required to be disconnected and not incab lights, fans, etc..--proper grounding should prevent this.
With this theory, I can't think why shunting would want to be delayed. For the most part, if there is a heat detector there should be a smoke detector in close proximity to provide safe recall. I believe waterflow switches are allowed to be used in lieu of heat detectors, but they absolutely cannot have any type of retard mechanism by code. Very uncommon since waterflows without retard mechanisms are just asking for trouble--I prefer heat detectors with a lower activation temp. than the adjacent sprinkler head as required.
Monitoring of the power to the shunt trip breaker coil to initiate a supervisory signal came about in NFPA72 1999. This is for obvious reasons--if you don't monitor the power that controls the shunt coil you would never know that the shunt function had been disabled except during annual testing.
NFPA 101 has the ASME 17.1 excerpts regarding elevator operation and interface.
NFPA 72 also has some good information in Supplement 2 at the back of the book.
