This has come up time and time again in my area. The NEC allows normal and emergency power to be in the same box so long as there is a divider that separates them. When this divider is installed it is considered 2 separate boxes and independent of each other as required by code.

The question is, when removing the cover on this box that has both, lets say receptacles, do you have to turn off both normal and emergency power to work on just one of them? In other words, if you only had to do work on the normal receptacle, lets say replace it, do you have to turn off both the emergency and normal to do it even though you will not be removing the receptacle of the emergency power in accordance with NFPA 70E? Remember they share the same cover.

What happens if you are working on the Normal Receptacle and the Emergency Power comes on and there is a fault in the emergency receptacle on the other side of the divider? Pretty could chance you will be caught in the flash. Unless both sources of power are LOTO and been tested de-energized, you will have to wear the proper PPE for the other receptacle than the one you are working on. In other words:
Work on Normal Receptacle - PPE for Emergency Receptacle
Work on Emergency Receptacle - PPE for Normal Receptacle

The e power is already on. I tend to agree with you over this issue , its just the way Nfpa 70E was written that some are having a debate about it.
This is one of the main reason why I'm not to keen on using dividers.

I fail to see how a receptacle could trigger an arc flash....what is the interaction with it?

There has been a case where the divider in the box moved when the other receptacle was being worked on and when it moved it caused a short on the other side. The arc flash wasn't really the issue because of course it is too small, it is the potential for electric shock and causing an unplanned outage on that other circuit that could have been prevented if both were turned off.

The reason some of us do not like putting both normal and emergency in the same box with dividers is that the practice is you want to always have a supply of power on. If you have to do work on the normal power, then you left the emergency on. If you had to do work on the emergency power, then you left the normal on especially if you are working in a health care facility like hospitals etc.. With both the normal and emergency in the same box it appears you have to turn both off to do the work.

Health care facilities are notorious for violating OSHA when it comes to power. You have to demonstrate greater hazard to a fatality. Look up the energized work permit in subchapter S and the letters of interpretation with regards to "infeasible" and "continuous process". The fact that someone might need an emergency procedure vs. redirecting to another facility or using various backup procedures is not an excuse for putting someone's life at risk.

This is one of the main reasons we want to start forcing them to stop using the same box for both the normal and emergency. There is pressure to not turn things off. They try use the exceptions in osha to justify working on the receptacles hot.

Maybe a code proposal to NFPA 99 can help in this area. What do you think?

Here is the offending OSHA requirement (1910.333(a)(1)). It is quoting what is now 130.2 in 70E from an earlier edition of NFPA 70E:
Live parts to which an employee may be exposed shall be deenergized before the employee works on or near them, unless the employer can demonstrate that deenergizing introduces additional or increased hazards or is infeasible due to equipment design or operational limitations. Live parts that operate at less than 50 volts to ground need not be deenergized if there will be no increased exposure to electrical burns or to explosion due to electric arcs.

Note 1: Examples of increased or additional hazards include interruption of life support equipment, deactivation of emergency alarm systems, shutdown of hazardous location ventilation equipment, or removal of illumination for an area.

Note 2: Examples of work that may be performed on or near energized circuit parts because of infeasibility due to equipment design or operational limitations include testing of electric circuits that can only be performed with the circuit energized and work on circuits that form an integral part of a continuous industrial process in a chemical plant that would otherwise need to be completely shut down in order to permit work on one circuit or piece of equipment.

The issue here is that note 1 in particular within the text seems to almost imply that you would never consider deenergizing anything in a healthcare facility. Never mind the fact that killing power to a lighting panel usually puts the lighting panel itself in darkness. Within an industrial plant, the term "continuous industrial process" usually gets called out instead as a crutch for all kinds of abuses of energized work. OSHA has described the key test in a letter of interpretation dated 12/19/2006. There used to be another much better letter as well but that seems to have been removed. The key test is this:

Would an orderly shutdown introduce greater or additional hazards, or would it simply alter or interrupt production? If the answer is that it merely alters or interrupts "production", then the exception does not apply. For example, although the panel may be supplying power to an operating room, can you reschedule to another time where the room is not in use? Similarly can the electrician use a flashlight if the lights go out? Can you post attendants or do rounds in lieu of loss of a fire alarm system? If the answer to any of these is yes, then the issue at hand is not a matter of greater or additional hazards.

Good examples would be for instance where a panel would need to be worked on immediately due to an emergency but which would disconnect power to an operating room while a patient is currently undergoing open heart surgery. Another example is in glass plants that make float glass (windows) where the process runs for 10-20 years between shutdowns and in which an orderly shutdown would effectively cause the furnace to collapse on itself, entailing a huge construction effort to rebuild the glass furnace lasting several months and requiring several times the number of personnel compared to the regular staff, all exposed to all the hazards of an ongoing construction project involving countless cleanup and construction hazards, causing a huge amount of additional hazards and exposures compared to normal operation.

Note that except for diagnostic electrical work, you need an EEWP which must document the specific reason why energized work is justified under 70E. In most industrial facilities there is usually a safety manager or a production manager, or both, signing off on the permit in addition to an electrical person. I highly suggest you put the following text just above the signature blanks directly on the EEWP form:

According to an OSHA Letter of Interpretation dated 12/19/2006, "...to qualify for the exception, the employer must, on a case-by-case basis, determine if the orderly shutdown of the related equipment (including the panel) and processes would introduce additional or increased hazards. If so, then the employer may perform the work using the electrical safe work practices found in §§1910.331-1910.335, including, but not limited to, insulated tools, shields, barrier, and personal protective equipment. If the orderly shutdown of the related equipment and processes would not introduce additional or increased hazards, but merely alter or interrupt production, then the de-energization of the equipment would be considered feasible, and the exception would not apply."

This puts the approver on notice that justification must be given on a case-by-case basis, and that the test is whether or not an orderly shutdown creates additional or increased hazards in and of itself. Personally, I would never sign off and agree to do electrical work when the justification is not specific because if someone on my watch gets injured or killed. I want to be in a defendable position when OSHA investigates.