I know nothing about codes for elevators but would think that any codes would want the elevator to stop in a position where passengers can easily get out if at all possible upon any kind of failure. I could easily see that meaning a backup system with limited ability at the very least. Limited ability meaning it will be able to move the car to the nearest normal access level for evacuation and not necessarily allowing full normal operating cababilities.
If you read that article, it reasons that killing the power just before (heat detector tripped) or concurrently with waterflow is desired. It is an article worth reading if you intergrate fire alarms and elevators.
From page 15 of the link I posted:
The annex text suggesting a time delay before power removal was added in 2002 and is not a requirement. The text was proposed as a requirement, but moved by the technical committee to the annex so that it would only be a suggestion. However, the technical committee statement in the Report on Proposals14 indicated that their intent was to require immediate power shutdown. Hence, the suggested text is in direct conflict with the intent of the body of the code. The submitter’s substantiation was that the delay would minimize the chance of an occupant being trapped in the elevator. The elevator safety task group that drafted the original framework and the coordinated requirements in the various codes and standards debated the relative risks involved with trying to ensure safe parking of the elevator car before removing power. The elevator experts convinced the group that the application of water before removing power can be a higher risk than the possibility of entrapment – assuming compliance with the requirements for initiating Phase I recall by smoke detectors. The group was also concluded that operation of a sprinkler in a machine space or hoistway (particularly at the top) would most likely mean that there was a fire, which could cause unpredictable operation of electronic controllers with results that might be worst than the risk of entrapment. The idea of delaying the water discharge until it could be confirmed that the elevator was parked has also been debated and rejected in part due to the increased complexity adding additional failure mechanisms as well as the unpredictable effects of the unsuppressed fire on the elevator equipment.
The use of time delays and pre-action sprinkler systems would require considerable engineering and engineering coordination of four systems: 1) elevator controllers, 2) sprinklers, 3) fire alarm and
4) electrical. The problem with these proposed methods is that there are no data on the actual statistical need, the failure and success rates and the various hazards that might be created. To some extent, this is also true of the more simple approach currently required by the coordinated set of codes and standards. Nevertheless, the elevator safety experts, fire protection engineers, fire alarm and sprinkler experts, fire fighters, code experts and others who have worked on the coordinated codes and standards have settled on the simpler approach.