Emergency and egress lighting: backup power, the 90-minute rule, and testing

What emergency and egress lighting is, and why it exists

Emergency lighting is the lighting that comes on automatically when normal power fails and keeps the path of egress lit long enough for people to get out. Egress lighting is the broader job: the illumination of the means of egress, the corridors, stairs, ramps, changes of direction, and the exit discharge to the public way. The two overlap. The same fixtures often serve both, but the test is different. Normal egress illumination has to be there while the building is occupied. Emergency illumination has to be there when the power is not.

This is a life-safety system, not an amenity. The whole point is the few minutes after a fault, a fire, or a utility outage when the building goes dark and a crowd has to find the door. People do not move toward a black hallway. They freeze, they crowd the one lit exit, and that is how a bad night turns into a worse one.

Day-to-day light levels, fixture layout, and the energy code that governs the normal lighting are a separate design problem. See the commercial lighting design guide for the footcandle targets and the controls. This guide is about what happens when that normal lighting drops out and the backup has to carry the egress path.

How much light does an egress path need?

Emergency illumination of the means of egress has to provide an average of 1 footcandle and, at any single point, a minimum of 0.1 footcandle, measured at the floor along the path. That is the figure in NFPA 101, the Life Safety Code, and the IBC mirrors it for the means of egress. The reading is taken at floor level because that is where feet and obstacles are, not at the work plane where you measure office lighting.

The average matters less than the dark spot. A path can average well over 1 footcandle and still have a black patch at a turn, a landing, or under a soffit where a fixture does not reach. NFPA 101 caps the maximum-to-minimum ratio at 40 to 1, which is the code's way of saying no single dead zone. Uniformity is the part crews miss, because the photometric average looks fine on the sheet while a stair turn sits in shadow.

The level is allowed to fade over the discharge. NFPA 101 lets emergency illumination decline to an average of 0.6 footcandle and a minimum of 0.06 footcandle at the end of the 90-minute period, since batteries sag as they drain. Design and test to the initial numbers, but know the end-of-duration floor is lower by design. Confirm the values against the adopted edition, because the AHJ controls what gets enforced.

How long must emergency lights stay on?

Emergency illumination has to last a minimum of 90 minutes, or one and a half hours, after normal power fails. That number runs all the way through the life-safety codes: NFPA 101 for the egress illumination, UL 924 for the listed equipment, and the battery-capacity rule in NEC Article 700. Ninety minutes is the design duration for the egress lighting, the exit signs, and the backup source feeding them.

The second number is just as hard. The emergency lighting has to come on within 10 seconds of losing normal power. Ten seconds is the gap a person can stand in a dark stair. Longer than that and the crowd is already moving blind. On a battery unit the transfer is effectively instant. On a generator the 10 seconds is the clock the engine and the transfer switch have to beat, which is why generator-only egress lighting needs a fast-starting set and an automatic transfer scheme built around that window.

NEC 700 backs the duration with a capacity rule. A storage battery serving an emergency system has to carry the full load for the 90 minutes without the voltage at the load falling below 87.5 percent of nominal, so the lights are still bright enough at minute 89, not just lit. A battery that holds voltage for 20 minutes and then collapses passes a quick flick test and fails the people in the building.

The three ways to power emergency lighting

There are three sources for emergency lighting, and the code recognizes all three. Unit equipment, the self-contained battery fixture, puts the battery in or next to each light. A central battery or inverter system puts one big battery bank in a room and feeds the emergency lighting circuits from there. A generator backs the emergency lighting as part of an emergency power supply system, the EPSS, through a transfer switch. NEC Article 700 lays out the permitted supply sources, commonly at 700.12.

The choice is about scale and where you want the maintenance to live. A small tenant build-out runs on unit equipment because it is cheap and there is nothing central to maintain. A hospital, a high-rise, or a large institutional building leans toward a central system or a generator because chasing batteries in 200 separate fixtures is its own failure mode. Many real buildings mix all three: generator-backed exit signs and stair lighting, a central inverter for a critical floor, and battery packs filling the gaps.

The table below is the quick decision frame. The right answer is whatever the design, the building size, and the AHJ support, and it is worth settling before the rough-in, because the wiring and circuiting are different for each.

Unit equipment: the battery pack and the bug eye

Unit equipment is the self-contained emergency light: a battery, a charger, a transfer circuit, and one or two lamp heads in a single listed assembly. The twin-head bug eye on the wall is the classic. So is the recessed downlight or the troffer with an integral battery pack that drives a few hundred lumens when the power drops. The battery exit sign is the same idea in a sign body. Each unit is its own little emergency system, independent of every other one.

The appeal is simple. There is no central room, no inverter, no dedicated feeder. You tap the unit to the normal lighting circuit in the same area so it senses that area's power loss, and the listing handles the rest. Every unit has a push-to-test button and a pilot light, which is how the monthly test gets done and how a dead battery announces itself if anyone looks.

The weakness is the same as the strength: the maintenance is everywhere. Two hundred fixtures means two hundred batteries on their own clocks, and the failures are scattered and quiet. The cheap units fail first, usually the battery at three to five years, and a building that bought the bottom of the catalog inherits a rolling replacement program nobody budgeted. Unit equipment is right for the small job and a maintenance trap at scale.

What is a central inverter for emergency lighting?

A central inverter is one UL 924 listed unit, an uninterruptible power supply built for life safety, that holds a battery bank in a room and feeds the emergency lighting circuits throughout the building from that one place. When normal power fails, the inverter carries the emergency lights from its batteries for the 90 minutes, with the same 10-second-or-better transfer. It is the central-battery answer for a building too big to run on scattered packs.

The reason large buildings go this way is the maintenance and the monitoring. All the batteries sit in one conditioned room where you can check voltage, run the test, and swap cells without a lift and a ladder at every fixture. A central system also opens up better battery chemistry and real monitoring: pure-lead or lithium banks that last far longer than the sealed cell in a bug eye, plus logging and remote alarms that tell you a problem before the fire marshal does. Lithium banks can run 12 to 15 years against the 3 to 5 years typical of the small sealed batteries.

It is not free. The inverter, the room, and the dedicated emergency circuits cost more up front, and the system is a single point you have to keep alive. The trade is fewer, better-maintained batteries in one place against many cheap ones everywhere. On a building of any size that math usually favors the central system, but the design and the AHJ make the call. A central inverter that feeds normal loads is not an emergency system, so it serves emergency lighting only.

The generator and the emergency power supply system

A generator backs emergency lighting as part of an emergency power supply system. Normal power drops, the engine starts, the automatic transfer switch moves the emergency loads to the generator, and the lights ride through. The whole sequence has to fit inside the 10-second window, which is why a generator serving life-safety loads is a fast-start set, commonly a Type 10 system under NFPA 110, the standard that governs the EPSS itself.

The wiring side is the emergency branch, sometimes called the life-safety branch on healthcare jobs. Under NEC 700 the emergency loads are kept on their own circuits and separated from normal and from other standby systems, so a fault on the normal side cannot pull down the egress lighting. The generator covers far more than lighting, but the egress lighting and exit signs are usually first in line on that branch.

Generators get their own depth elsewhere. The transfer scheme, the ATS, the load steps, and the start sequence are a topic of their own, so coordinate the lighting design with the generator and transfer design rather than treating them separately. One blunt point: a generator that fails to start takes the whole egress path with it. That single-point risk is exactly why codes often pair a generator-backed system with battery units or a central inverter on the most critical paths, so a no-start does not mean total darkness.

Exit signs: illuminated, legible, and on the backup

An exit sign marks the way out and has to stay legible when the power fails. The common type is the internally illuminated LED sign with its own battery, which keeps the face lit for the 90 minutes after normal power drops. LED replaced the old incandescent and fluorescent signs because it draws almost nothing, which means a smaller battery carries it the full duration. The signs are listed to UL 924, and that listing is what makes them code-acceptable.

Photoluminescent signs are the no-power alternative. They have no wiring and no battery. They charge off the ambient light while the building is occupied and glow on stored energy when the lights go out, which means the surrounding light has to stay on the face continuously during occupancy at the level in the sign's listing. They work where running emergency power to a sign is impractical, but only if the charging light is reliable and the AHJ accepts them.

Placement is half the job. Signs go at every required exit and at the points along the path where the way out is not obvious, the changes of direction and the decision points, so a person always has a lit sign in view heading out. UL 924 tests visibility to 100 feet, so a sign has a reach, and a path longer than that needs another sign. The failure mode inspectors find first is the dark or missing sign at the turn, where the occupant needs it most and there is nothing there.

Laying out the egress path so nothing goes dark

The design problem is covering the whole path of egress with no dark spot, on the backup, at the floor. That means the corridors, every stair and landing, each change of direction, the doors, and the exit discharge all the way out to the public way. The discharge is the piece people forget. The path does not end at the exit door. It ends where the occupant reaches a safe public space, and the code follows it there.

You hit the level the same way you do normal lighting, with a photometric layout, but the constraint is tighter because you are working with the reduced output of battery heads or emergency drivers, not the full fixture. A fixture that throws plenty in normal mode may put out a fraction in emergency mode, so the emergency photometrics are their own calculation, not an afterthought on the normal plan. See the lighting design guide for the lumen method and the layout mechanics.

Watch the geometry the average hides. A long straight corridor is easy. The trouble is the stair turn, the vestibule, the spot under a beam or a soffit, and the threshold where one space hands off to the next. Those are where the minimum 0.1 footcandle and the 40-to-1 ratio get blown, and they are exactly the points a person has to make a decision in the dark. Lay the heads for the turns, not the straightaways.

The emergency branch circuit and survivable wiring

The emergency lighting circuit is not an ordinary branch circuit, and the difference is the part that gets wired wrong. The circuit feeding a battery unit has to be the same one that lights the area normally, so the unit senses a local power loss, and that circuit cannot have a switch that can kill the emergency function. You do not put the bug eye behind a wall switch someone can flip off. Under NEC 700 the emergency loads stay on their own circuits, dedicated to emergency use, kept clear of normal and other standby wiring.

That separation is the safety logic. If the normal lighting and the emergency lighting share a fault path, the event that takes the building dark takes the emergency lighting with it. Keeping the emergency branch independent means a fault, a tripped breaker, or a switched-off circuit on the normal side leaves the egress path lit.

On larger and central systems, wiring survivability becomes part of the design: the emergency feeders are routed and protected so a fire does not sever the path before people are out. The detailed circuit-integrity and survivability rules overlap with the fire alarm world, where the same fire-rated and pathway-survivability thinking applies to notification circuits. Coordinate the two, because a fire that burns through an unprotected emergency feeder defeats both systems at once.

The codes that govern emergency and egress lighting

Three documents do most of the work, and they do different jobs. NFPA 101, the Life Safety Code, sets the egress illumination level, the 90-minute duration, the 10-second response, and the testing. The IBC carries parallel means-of-egress illumination and exit-sign requirements on the building-code side, so a project usually answers to both. NEC, NFPA 70, governs the electrical installation: Article 700 covers emergency systems, the permitted power sources, the wiring separation, and the battery capacity.

Knowing which book owns which requirement keeps you out of trouble at plan review. The performance, the footcandles, the duration, comes from NFPA 101 and the IBC. The wiring, the circuits, the source, comes from the NEC. UL 924 is the product standard that the equipment, the unit lights, the inverters, the exit signs, is listed to, so when a spec says UL 924 it means the gear, not the installation.

Article 700 sits next to two siblings worth naming. NEC 701 covers legally required standby systems and 702 covers optional standby, which are different animals with different rules, so do not treat a 702 system as if it satisfies a 700 emergency requirement. Emergency systems also carry selective-coordination requirements so an upstream device does not trip and dark the whole branch for a downstream fault, a topic that lives with the overcurrent design. Article and section numbers move between code cycles, so confirm them against the adopted edition and the local amendments before you cite them on a submittal, and let the AHJ settle the close calls.

How often do you test emergency lighting?

Two tests, on two clocks. Every 30 days you run a functional test for at least 30 seconds: drop the unit to its battery, confirm the lamps light and the transfer works, and look at it. Once a year you run the full-duration test, the 90-minute discharge, where the system has to carry the load the entire one and a half hours on battery. NFPA 101 sets both, and the annual is the one that actually proves the battery.

The monthly 30-second test catches the surface-charged battery, the one that reads full and lights for a moment, then dies. Thirty seconds is enough to see the lamp come on and tell that the transfer fired. It is not enough to prove the battery has any real capacity left, which is the whole reason the annual exists.

The annual 90-minute test is where dead batteries get caught honestly, because a tired cell that holds for 20 minutes and collapses passes every monthly flick and fails the one that counts. Both tests have to be documented. To the fire marshal and the OSHA inspector, a test that is not written down did not happen, and the record is the first thing they ask for. Skip the log and you have done the work and still failed the inspection.

Self-testing and self-diagnostic luminaires

Self-testing fixtures run the code-required tests on themselves and tell you the result. A self-diagnostic unit performs the 30-second functional check on its monthly schedule and the longer-duration check on its annual schedule automatically, then signals pass or fail with an indicator light or a network report. The point is to take the manual burden, the walk-the-building-with-a-ladder-and-a-clipboard burden, off the staff who never quite get to it.

On a building with hundreds of fixtures this is the difference between testing that happens and testing that does not. Manual monthly checks across a large facility quietly slip, and the slip does not show until the annual or until the night the power drops. A self-testing fixture that flags its own failed battery turns a hidden problem into a visible one before it matters.

Self-testing does not erase the recordkeeping, and it does not always cover everything. The fixtures handle their own checks, but you still confirm they are reporting, you still log the results, and the AHJ may still want a witnessed annual on certain systems. Treat self-testing as a tool that makes the program reliable, not a reason to stop paying attention to it.

Batteries, replacement, and the test log

The battery is the part that wears out, and it is the part the whole system depends on. The small sealed cells in unit equipment and battery exit signs typically last 3 to 5 years before they lose enough capacity to fail the 90-minute test, and they fail on their own schedules all over the building. Heat shortens that life, so a unit over a warm doorway or in a hot mechanical space dies sooner than the one in a cool corridor.

Central systems change the maintenance picture without removing it. The batteries are in one room, easier to check and swap, and better chemistry buys time: pure-lead and lithium banks can run far longer than the sealed cell in a bug eye, with lithium reaching 12 to 15 years. The inverter still needs its scheduled checks, and the bank still has a finite life that the monitoring should be tracking.

The test log is the record that ties it together. Every monthly and annual test gets dated and recorded, batteries get a replacement date, and the failures get noted and corrected. That log is what proves the system was maintained, and it is what the next owner, the facility manager, and the fire marshal all read. A pile of dead batteries with no record is the common end state of a building that bought the cheap units and never set up the log.

Commissioning the emergency lighting

Commissioning is where you prove the system does what the drawings claim, before the building is occupied and before it is somebody else's problem. The core of it is a walk of the egress path with a light meter, reading the floor along the whole route, to confirm the average lands at 1 footcandle and no point drops below 0.1, with the normal lighting killed and the system on its backup. The photometric on the sheet is a prediction. The meter on the floor is the proof.

Then you test the function and the duration. Drop the normal power for real and confirm the lights come on within the 10-second window. Run the 90-minute discharge to confirm the system carries the full duration and the level has not collapsed below the end-of-duration floor. Walk the exit signs and confirm each is lit, legible, and placed so the path is always marked. Check that every unit's circuit cannot be switched off.

Document all of it. The walk readings, the response time, the duration result, the sign check, and any deficiency and its fix become the commissioning record and the baseline the annual tests get compared against. A system commissioned on paper and never measured on the floor is the one that has a dark stair nobody finds until the night it matters.

Data centers and critical-facility egress

Critical facilities add a wrinkle, because the white space is a large, often windowless room where a person can be a long way from a door with racks blocking the sightlines. The egress path through the data hall, down the rows, to the exits and out, has to be lit on the backup like any other, but the geometry is harder. Tall racks cast shadows, the aisles are long, and a single bank of heads at the wall will not reach the middle of the room.

These buildings almost always have a generator and often a central inverter already, so the emergency lighting usually rides the same heavily backed power that protects the load, but the egress lighting is still its own life-safety circuit, separate from the IT power. The mistake is assuming the building's heavy backup covers egress by default. It does not. The egress lighting has to be designed, circuited, and tested as a life-safety system in its own right.

The path also runs through spaces people rarely walk, the electrical rooms, the gen yard, the long service corridors, and those are exactly where a missing head or a dark turn hides until the test. Walk the whole route on the meter, including the parts nobody uses, because the code follows the path of egress all the way to the public way.

The owner-side maintenance

When the job is done and the building is occupied, the emergency lighting becomes a standing maintenance obligation that runs forever, and the owner owns it. The monthly 30-second test, the annual 90-minute test, the battery replacements, and the log do not stop when the certificate of occupancy prints. They are the program the facility staff has to run for the life of the building.

The AHJ and the fire marshal are the enforcement. They show up, they ask for the test records, and they spot-check the lights, and a building that cannot produce a current log has a finding whether or not the lights actually work. The documentation is not paperwork for its own sake. It is the evidence that the life-safety system has been kept alive.

The honest handoff is to tell the owner what they are taking on and set the program up before you leave: who runs the monthly test, who runs the annual, where the log lives, and when the first batteries are due. Self-testing fixtures and a central system with monitoring make that program survivable. A building full of cheap bug eyes and no plan makes it a slow failure that nobody notices until an inspection or an outage.

What to document

The record is what proves the egress path is lit, the backup holds, and the tests are getting done. It is what the fire marshal reads, what the next contractor reads, and what answers the question a year out when a battery is found dead. Capture it by area, so a finding points to a fixture and not to the whole building.

For each area, record the fixture or system serving it, the measured floor illumination on the egress path, the result of the annual 90-minute test, and the monthly functional-test record. Add the backup source, the battery type and install date, and the date and result of commissioning. If a unit failed and was replaced, the log shows when and what.

Common mistakes

• Egress path under 1 footcandle, or a dark spot at a stair turn or change of direction that blows the 40-to-1 ratio.
• Battery or backup that cannot carry the full 90 minutes, so it passes a quick test and fails the people in the building.
• Skipping the monthly 30-second test or the annual 90-minute test, or running them and never writing them down.
• Exit sign missing or dark at a decision point, or beyond the 100-foot visibility reach with no second sign.
• Emergency lighting circuit on a switch or breaker that can turn the bug eye off, instead of a dedicated emergency circuit.
• Dead or aging batteries with no replacement schedule and no test log, the slow failure that cheap unit equipment guarantees.
• Lighting the exit door but not the exit discharge all the way to the public way.

Field checklist

Standards and references

NFPA 101, the Life Safety Code, is the document that sets the egress illumination level, the 1 footcandle average and 0.1 footcandle minimum at the floor, the 40-to-1 ratio, the 90-minute duration, the 10-second response, and the monthly and annual test schedule. The IBC carries parallel means-of-egress illumination and exit-sign requirements on the building-code side, so most projects answer to both.

NEC, NFPA 70, governs the electrical work. Article 700 covers emergency systems: the permitted power sources, the wiring separation and dedication, and the battery-capacity rule that keeps load voltage above 87.5 percent through the 90 minutes. Article 701 covers legally required standby and 702 covers optional standby, which are separate and do not substitute for a 700 emergency system. NFPA 110 governs the emergency power supply system where a generator backs the load. UL 924 is the product standard the emergency lighting equipment, inverters, and exit signs are listed to.

The exact article and section numbers shift between code cycles, so confirm them against the edition the jurisdiction has actually adopted and any local amendments before citing them on a submittal. The AHJ and the fire marshal enforce the result and settle the judgment calls, and a project specification can be tighter than the code minimum, in which case the spec governs.

Units, terms, and definitions

Emergency and egress lighting carries a handful of terms that read differently across a drawing set, a product sheet, and a code section, so the same idea shows up under more than one name.

Illumination is measured in footcandles, abbreviated fc, which is lumens per square foot; the metric equivalent is lux, where 1 footcandle is about 10.8 lux. The means of egress is the whole path out, from any point in the building to the public way, including the exit access, the exit, and the exit discharge. The EPSS is the emergency power supply system, the generator and its controls. Unit equipment, the central inverter, and the generator branch are the three backup sources.

Footcandle (fc) / lux

Light at a surface; lumens per square foot. 1 fc is about 10.8 lux

Means of egress

The full path out: exit access, exit, and exit discharge to the public way

Unit equipment

Self-contained emergency light with its own battery, charger, and transfer (the bug eye)

Central inverter / central battery

One UL 924 system feeding the emergency lighting circuits from a single battery bank

EPSS

Emergency power supply system, the generator and controls backing the emergency loads

UL 924

The product standard for emergency lighting and exit-sign equipment

Max-to-min ratio

Uniformity limit, capped at 40 to 1 so no single dark spot on the path

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• How many footcandles do you need for an office?
• What is the difference between conventional and addressable fire alarm systems?
• What is the difference between emergency and standby power?
• What is the difference between open and closed transition transfer?
• What is selective coordination?