Emergency and standby power systems: NEC 700, 701, and 702 explained

Emergency and standby power, and why the class is the first decision

Emergency and standby power systems are the backup sources that pick up a building's loads when the utility fails. The piece that confuses people is that the NEC does not regulate them by the hardware. A generator is a generator whether it backs a hospital or a server room. What the code regulates is how critical the load is, and it sorts every backup system into one of three buckets: emergency, legally required standby, and optional standby. The bucket, not the machine, sets the rules.

That sorting is the whole game. The class decides how fast the power has to come back, whether the wiring has to be kept separate from everything else, whether the overcurrent devices have to selectively coordinate, and how the system has to be tested for the life of the building. Two buildings can have the identical generator and transfer switch, and the install, the inspection, and the maintenance are different because the load behind them is classified differently.

So the first decision on any backup job is not what to buy. It is what the load does. Pin the classification down before sizing, before the one-line, before the gear goes on order, because if you build an optional standby system and the load turns out to be life safety, you do not have a small fix. You have the wrong system.

The three NEC classifications

The NEC splits backup power across three articles, and they run from most stringent to least. Article 700 covers emergency systems, the loads that are required by law for human life and safety: egress and exit lighting, exit signs, fire alarm, and the loads a building code or fire code says must stay up for people to get out safely. Article 701 covers legally required standby, the loads a code or the AHJ requires but where loss does not immediately threaten life: smoke control, certain elevators, building systems that aid firefighting and rescue. Article 702 covers optional standby, the loads the owner chooses to back up because losing them costs money, not lives.

The line between them is not the equipment and not the size. It is who said the load has to be backed up, and what happens in the seconds and minutes after it goes dark. Emergency loads are mandated by code and a dark egress path can kill someone. Legally required loads are also mandated, but the harm is slower. Optional loads are nobody's mandate but the owner's.

Get the classification from the building code, the fire code, and the AHJ, not from the electrical drawings alone. The electrical engineer wires what the load is, but what the load is gets decided by the life-safety analysis upstream of the electrical design. That is the part that surprises crews who think of this as an electrical question.

What is an NEC 700 emergency system?

An NEC Article 700 emergency system is the backup that supplies loads legally required for life safety when normal power fails. These are the loads a building or fire code names for safe egress and life protection: the illumination of the means of egress, exit signs, fire alarm and detection, and in many occupancies the loads needed to evacuate or shelter people. If the code says a load must be powered for people to survive the event, it is an emergency load and Article 700 governs it.

Article 700 is the most demanding of the three, on every axis. The power has to be restored fast, commonly within 10 seconds of losing normal. The emergency wiring has to be kept independent of all other wiring. The overcurrent devices have to selectively coordinate so a downstream fault cannot dark the whole branch. The equipment has to be listed for emergency use, and the system has to be tested on a schedule for as long as the building stands. None of that is optional, because the load behind it is life itself.

The egress lighting and exit signs are the classic 700 loads, and they have their own depth: the footcandle levels, the 90-minute duration, the unit-equipment versus central-inverter choice, and the testing all live in the emergency and egress lighting guide. This guide is about the system class that those loads sit inside, and the rules that flow from the class.

What is legally required standby power?

Legally required standby is the NEC Article 701 class: loads a code or the AHJ requires to be backed up, but where loss of the load does not immediately threaten life the way a dark stair does. The textbook examples are smoke control and stairwell pressurization, certain elevators, some ventilation, and systems that aid the people responding to the emergency rather than the people escaping it. Someone with authority decided these have to stay up, so they get a backup source, but the clock and the harm are slower than emergency.

The response time relaxes to match. Legally required standby is commonly required to transfer to the alternate source within 60 seconds of losing normal, against the 10 seconds an emergency system gets. The transfer equipment still has to be automatic and listed and marked for the use, and Article 701 still calls for selective coordination of the overcurrent devices. What it does not carry is the full wiring-independence rule that 700 imposes, which is the practical difference an installer feels.

The trap here is assuming legally required is just emergency with a longer clock. It is its own article with its own loads, and the loads come from the building code and the fire code and the AHJ, not from a generic list. Smoke control on one project is legally required standby; on another it is part of the emergency system. Confirm the class from the codes that apply to the actual occupancy.

Optional standby: the loads the owner chooses

Optional standby is the NEC Article 702 class, and it is defined by what it is not: no code requires it, and no life depends on it. The owner backs up these loads because losing them costs money or comfort. A data center's IT load, a manufacturing process that ruins a batch if it stops, the heat in a building full of pipes that freeze, a grocery store's refrigeration, an office that wants the lights and the elevators to ride through an outage. All of it is optional standby, because the choice to back it up belongs to the owner, not to a code official.

Because no code mandates the load, no code mandates a transfer time for it. Optional standby has no required restoration clock. The owner and the load set the requirement: an IT room might want a no-break transfer through a UPS, while a comfort load can take a minute or two without anyone caring. Article 702 still requires the installation to be done to code and the transfer equipment to be suitable, but the performance target is the owner's to set.

Where crews get this wrong is in both directions. They build an optional system to the full weight of Article 700, spending money the load never justified, or worse, they treat a life-safety load as optional and quietly strip the separation, the coordination, and the testing off a load that legally needs them. The class is not a budget knob. It is a finding about what the load is.

Why the classification drives everything else

Every requirement that follows on a backup system hangs off the class, which is why the classification is worth fighting over early. The transfer time comes from the class. The wiring separation comes from the class. The selective-coordination requirement comes from the class. Whether the equipment has to be listed for emergency use comes from the class. The testing program the owner inherits comes from the class. Change the classification and you change all of it at once.

This is also why misclassifying is expensive instead of just wrong. An emergency system that got built like an optional one is missing the 10-second transfer, the independent raceways, the coordinated overcurrent devices, and the listed gear, and you cannot bolt those on at the end. The separation has to be designed into the raceway layout. The coordination has to be designed into the breaker and fuse selection. By the time an inspector catches a 700 load on a 702 system, the fix is a rebuild.

Run the classification before anything else and write it on the one-line. The first thing a reviewer should see is which article governs each backup load, because that single label tells them what the rest of the design has to satisfy. A backup system with no stated class is a system nobody can check.

How fast must emergency power transfer?

An emergency system under NEC Article 700 commonly has to restore power to the life-safety loads within 10 seconds of losing normal power. Legally required standby under Article 701 commonly has to transfer within 60 seconds. Optional standby under Article 702 has no code-mandated transfer time at all, because no code requires the load. Those are the response times that fall straight out of the class, and they are the single clearest reason the class matters to the hardware.

The 10 seconds is the figure NFPA 110 carries as a Type rating, where a Type 10 system means the load terminals cannot be without acceptable power for more than 10 seconds. That clock has to swallow the whole sequence: the controls sensing the loss, the engine cranking and coming up to speed and voltage, and the transfer switch moving the load. It is why a generator serving life-safety loads is a fast-start set, not a machine that takes a minute to warm up.

Treat the numbers as widely held expectations, not as something to quote from memory on a submittal. The exact wording and the exact seconds live in the adopted NEC edition and the referenced NFPA 110, and they can move between cycles. Confirm the transfer time against the code the jurisdiction has actually adopted and the AHJ before you commit a number to the design.

The sources an emergency system is allowed to use

An emergency system can be fed from more than a generator, and the NEC names the permitted sources, commonly at 700.12. The list runs to several: an on-site engine-generator set, a storage battery, an uninterruptible power supply, a separate utility service where the AHJ allows it, and fuel cell systems. Unit equipment, the self-contained battery light, is its own permitted source for the lighting it serves. The choice depends on the load, the duration it has to ride, and what the class allows.

On-site fuel is the reason most life-safety power is a diesel generator. A generator that carries its own fuel does not depend on anything outside the building, where a second utility service can fail in the same regional event that took the first one, and a battery only lasts as long as its charge. For a load that has to ride a long outage, the on-site generator is the source that does not have a string attached to it. For a load that only has to bridge the seconds to a 90-minute duration, a battery or central inverter is often the right and cheaper answer.

Getting the source machine right, the generator sizing, the pad and clearances, the fuel and exhaust, the neutral and grounding, is the subject of the standby generator and transfer switch guide. Here the point is narrower: the class limits which sources qualify and how independent the source has to be, and a separate utility tap that the AHJ will not count as independent does not satisfy an emergency system no matter how convenient it looks on the one-line.

Keeping the emergency wiring independent

The wiring rule is the one that most sharply separates an Article 700 emergency system from a 702 optional one. Emergency circuits generally have to be kept independent of all other wiring, in their own raceways, boxes, and cabinets, so that a fault on the normal system or on another standby system cannot reach over and take the emergency system down with it. You do not run the emergency conductors in the same conduit as the normal feeders, and you do not share a junction box between them except where the code specifically permits it.

The logic is the failure you are designing against. The whole point of an emergency system is that it works on the worst night, when a fault or a fire is already loose in the building. If the emergency wiring shares a path with the normal wiring, the same event that creates the emergency can sever the system meant to carry people through it. Independence keeps the two failures from being the same failure.

This separation is also a marking and identification job, not just a routing one. The emergency raceways and boxes have to be identifiable so the next electrician does not unknowingly tap a normal load onto the emergency branch or pull the emergency conductors into shared pipe during a renovation. Legally required standby under 701 carries its own separation expectations, generally lighter than 700, and optional standby under 702 does not carry the independence rule at all. The separation you owe is the separation the class demands.

What is selective coordination, and which classes require it?

Selective coordination means the overcurrent devices are selected so that a fault opens only the device nearest the fault, leaving everything upstream closed and the rest of the system energized. On an emergency system that is the difference between a downstream fault clearing one circuit and the same fault tripping a main that darks the entire emergency branch. The NEC requires selective coordination on both emergency systems and legally required standby systems, commonly cited at 700.32 for emergency and 701.32 for legally required in recent editions.

Optional standby under 702 does not carry the selective-coordination requirement, which is one more way the class drives the design. On a 700 or 701 system the coordination has to be engineered, not assumed: the breaker time-current curves and the fuse selectivity ratios are studied across the full range of fault currents so that the nearest device always wins the race. The code commonly requires that this be selected and documented by a licensed professional engineer or another qualified person, and that the documentation be available to whoever inspects, maintains, and operates the system.

This is the requirement crews underestimate because it is invisible until a fault finds it. A system can pass every functional test and still lack coordination, because coordination is about which device opens when two are in series, and that only shows up on a real fault. Selective coordination has its own depth as an overcurrent-design topic. Confirm the exact section numbers against the adopted edition, because they have moved between code cycles, and the older editions placed these requirements at different numbers.

The automatic transfer switch in the system

The automatic transfer switch is the device that senses the loss of normal power, signals the source to start, and moves the load to it, then moves it back when normal returns. On emergency and legally required systems the transfer has to be automatic, and the switch has to be listed and marked for that class of use, which is a different listing than a switch sold for an optional standby installation.

How the switch is wired, the open-versus-closed transition, the service rating, and the neutral switching are install decisions covered in the standby generator and transfer switch guide. How the switch is proven, the loss-of-normal transfer timing, the retransfer, the time delays, and the bypass operation, is the subject of the transfer-switch commissioning work referenced there. The point for the system class is narrower: the transfer time the switch has to meet, and the listing it has to carry, both come from whether the load it serves is emergency, legally required, or optional.

One blunt point belongs here. The transfer switch is where a backup system most often fails to do its job, not because the switch is bad but because the start signal was never proven end to end. A switch that looks right and a generator that runs fine still leave the load dark if the contacts that tell the engine to start were never actually traced.

The egress and life-safety loads on the emergency system

The loads that define an Article 700 emergency system are the life-safety loads, and the largest and most visible of them is the egress lighting. Illumination of the means of egress, the exit signs, and the fire alarm and detection system are the loads that have to ride the outage so that people can find the way out and the building can warn them. In many occupancies the emergency system also carries elevator recall, fire-pump-related controls, and other loads the building code ties to life safety.

These loads are why the emergency class is the strict one. A dark egress path is not an inconvenience. It is the difference between an orderly evacuation and a crowd pressing toward the one lit door, and the codes treat it that way. The 10-second transfer exists because that is about as long as a person will stand in a dark stair before moving blind, and the independence and coordination rules exist so that the event causing the outage cannot also kill the lighting that carries people through it.

The egress lighting is deep enough to be its own subject: the footcandle levels, the 90-minute duration, the choice between unit equipment, a central inverter, and a generator branch, the exit-sign placement, and the monthly and annual testing all live in the emergency and egress lighting guide. Treat that guide as the load-side detail and this one as the system class those loads sit inside.

Assigning loads to the right class

The hardest part of this work is not the wiring. It is deciding which load belongs on which system, because the building code, the fire code, and the AHJ make that call and the electrical design follows it. A load is an emergency load only if a code requires it for life safety. It is legally required standby only if a code or the AHJ requires it without it being immediate life safety. Everything else the owner wants backed up is optional standby. The same physical load can land in a different class on a different occupancy.

Build the load assignment as a deliberate document, not as an afterthought on the panel schedule. For each backed-up load, write down what it is, which code requires it, and therefore which article governs it. That document is what tells the rest of the design which separation, which coordination, and which transfer time apply to that load, and it is the first thing a reviewer checks.

The table below shows how the same kinds of loads sort. It is a frame, not a rule, because the occupancy and the adopted codes move loads between classes. A smoke-control fan can be a legally required standby load on one job and part of the emergency system on another, and an elevator can be life-safety, legally required, or optional depending on what the building code says about it.

The fire pump and its own article

The fire pump is a life-safety load, but it does not live in Article 700. Its power supply has its own NEC article, commonly 695, which sets stricter and different rules than the general emergency-system article because a fire pump has to run when everything around it is failing. The fire pump circuit is sized and protected so it keeps the pump running rather than protecting the pump, which is the opposite of how an ordinary motor circuit is built: you want the pump to run to destruction during a fire, not trip off and leave the sprinklers dry.

That is the part that catches people who treat the fire pump as just another emergency motor. The overcurrent device ahead of a fire pump is set to carry locked-rotor current indefinitely, the wiring has survivability requirements, and the supply arrangement is spelled out separately from the building's emergency system. A fire pump can be fed from the normal service, a separate service, or an on-site generator, and the arrangement has to satisfy its own article and the applicable NFPA fire-protection standards.

Coordinate the fire pump power with the emergency power design, but keep the rule sets straight. Confirm the fire pump supply against its own NEC article and the fire-protection codes that apply, and against the AHJ, because the requirements differ from the general emergency-system rules in ways that matter for sizing and protection.

NFPA 110 and the emergency power supply system

Where the NEC says which loads need backup power and how the circuits are run, NFPA 110 governs the performance of the system that supplies them, the emergency power supply system, or EPSS. The EPSS is the whole package taken together: the energy source and the prime mover, the generator, the transfer equipment, the controls, and the accessories, treated as one system that has to perform. NFPA 110 is the standard a generator-backed emergency or legally required system is commonly built and tested to.

NFPA 110 classifies an EPSS three ways, and the three are easy to mix up. Type is how fast, in seconds, the system has to restore power to the load, so a Type 10 system restores within 10 seconds. Class is how long, in time, the system has to run on its on-site energy without refueling, so a Class 2 system carries at least 2 hours of fuel and a Class 48 carries 48. Level is how critical the application is, with Level 1 the most stringent, for installations where failure could cost life, and Level 2 where failure is a less immediate hazard.

Those three labels, Type, Class, and Level, are what a spec uses to pin down the EPSS, and they map onto the NEC class without being identical to it. NFPA 110 also drives the installation acceptance test and the ongoing testing the owner inherits. The detail of how a generator EPSS is built and accepted lives in the standby generator and transfer switch guide. Confirm the Type, Class, and Level against the adopted edition and the project specification before citing them.

Signals, annunciation, and monitoring

A backup system that nobody is watching is a backup system that fails quietly. The codes and NFPA 110 require monitoring and remote annunciation on many emergency and legally required systems, so an operator away from the equipment can see the source's status and its alarms. The single most useful point on that annunciator is the not-in-auto signal, because a generator or transfer switch left in off or manual after a service call will not run on an outage, and the annunciator is the only thing that tells anyone before the lights go out.

The points that have to reach the annunciator commonly include the source running, the source supplying the load, and the trouble and not-in-auto conditions. On a generator EPSS that list grows to cover the engine alarms. The monitoring also has to watch the integrity of the system itself, the battery charger, the source availability, so a problem that develops between outages is caught between outages rather than discovered during one.

Modern systems report to the building management system as well, and that mapping is part of the install, not a nicety. A point that reads right on the local controller but never reaches the annunciator or the BMS leaves the operator blind, which is exactly the failure that only surfaces when someone forces the alarm and watches both ends. Prove every point end to end during commissioning instead of trusting that the contacts that look right actually carry the signal.

How often are emergency and standby systems tested?

Emergency and standby systems are tested on a recurring schedule for the life of the building, not just at commissioning, and the schedule comes from the codes and from NFPA 110 for a generator EPSS. The common cadence is a monthly operational test, where the system is exercised and the transfer is proven, and a periodic full-load and longer-duration test, often annual, where the source is loaded and run long enough to prove it actually carries the load. For battery-backed egress lighting the parallel is the monthly functional test and the annual 90-minute discharge.

The monthly test catches the obvious failures: a set that will not start, a transfer that will not throw, a battery that is already dead. The longer load test is the one that matters, because it proves the system carries its rated load under heat rather than just starting and idling. A generator run only on light building load wet-stacks and glazes its cylinders, so NFPA 110 ties the exercise to reaching the maker's recommended exhaust temperature, not to a clock alone.

Document every test, because to the AHJ and the fire marshal a test that is not written down did not happen. The record is the first thing they ask for, and a building that cannot produce a current log has a finding whether or not the system actually works. Set the exercise program and the log up before turnover, because the program that is not stood up at handoff is the program that quietly never starts. Confirm the required intervals against the adopted NFPA 110 edition and the AHJ.

How does the data-center case split across classes?

A data center is the cleanest example of why a single building carries more than one class of backup power at once. The IT load, the racks, the cooling that keeps them alive, is optional standby under Article 702, because no code requires the servers to stay up. The owner backs them because an outage costs money, so the IT power is engineered to the owner's uptime target, usually a no-break ride-through on a UPS with generators behind it, not to any code transfer time.

Sitting underneath that, the same building still has an emergency system under Article 700 for its own life safety: the egress lighting, the exit signs, the fire alarm. Those loads are mandated by code exactly as they are in any other occupancy, and they carry the 10-second transfer, the separation, and the coordination that the emergency class demands. The building has two layers of backup power that answer to two different masters, and they are designed, separated, and tested separately.

The mistake is assuming the heavy backup that protects the IT load covers the life-safety loads by default. It does not. The egress lighting in a data hall is its own life-safety circuit, separate from the IT power, and it has to be classified, separated, and tested as an emergency system in its own right. The deeper power-chain and redundancy design for the critical load is its own subject by topic; the point here is that the two classes coexist and stay distinct.

The healthcare essential electrical system

Healthcare facilities get their own framework, because keeping patients alive needs more than the general emergency-system articles cover. The NEC handles hospitals and many other healthcare occupancies under Article 517, which builds an essential electrical system, the EES, out of branches rather than a single emergency system. The common structure splits the EES into a life-safety branch, a critical branch, and an equipment branch, each with its own loads and its own rules.

The split exists because a hospital has loads that the general articles do not address. The life-safety branch carries the egress and life-safety loads that look like an ordinary Article 700 emergency system. The critical branch carries patient-care loads and the equipment that supports life and treatment, the loads that keep a patient alive through the outage, which 700, 701, and 702 do not specifically address. The equipment branch carries the larger loads, often delayed-automatic, that the facility needs to keep running but that can wait a few seconds longer. The life-safety and critical branches have to be kept separate from other wiring and clearly marked, much like the emergency-system separation but with the healthcare specifics layered on.

If you are wiring a hospital, the essential electrical system is its own discipline and Article 517 governs it, alongside NFPA 99 for healthcare facilities and the same NFPA 110 for the EPSS behind it. Confirm the branch structure and the requirements against the adopted editions and the AHJ, because healthcare carries the strictest and most specific rules in this whole area.

Designing the system: classify, then build to the class

The design sequence is the reverse of how crews instinctively approach it. You do not start with the generator. You start with the loads, classify each one by its criticality and the code that mandates it, and only then design the source, the transfer, and the wiring to satisfy the class you found. Classify first, build second. Every downstream decision is determined by the class, so guessing the class to save time at the front guarantees rework at the back.

Once the class is settled, the requirements fall out in order. The class sets the transfer time, which sizes the source's start-and-transfer performance. The class sets the wiring separation, which shapes the raceway and box layout. The class sets the selective-coordination requirement, which drives the overcurrent-device selection and the coordination study. The class sets the listing the gear has to carry and the testing program the owner inherits. The source itself, the generator or battery or inverter, gets sized to the load by the load calculation, a separate exercise.

On a real building with mixed loads, you do this per system, not once. The emergency loads get a 700 system, the legally required loads get a 701 system, the optional loads get a 702 system, and they are kept appropriately separate. The design is not one backup system but a set of them, each built to the class of the loads behind it, and the one-line should show which is which.

The AHJ, the adopted code, and the local amendments

The model codes set the framework, but the authority having jurisdiction is what makes any of it enforceable on a specific job. The AHJ adopts a particular edition of the NEC and the referenced standards, takes local amendments that can be stricter than the model text, and has the final say on the judgment calls: which loads are legally required, whether a separate service counts as an independent source, and what the testing and documentation have to show. The class of a backup system is often the AHJ's call as much as the designer's.

This is why every number in this area carries a hedge. The 10-second and 60-second transfer times, the section numbers for selective coordination, the permitted sources, and the testing intervals are widely held, but the exact requirement is whatever the adopted edition and the local amendments say, and editions change every cycle. Quoting a section number from memory on a submittal is how you lose an inspector who is working from a different edition than you are.

Engage the AHJ early, especially on the classification and on anything unusual: a second-service emergency source, a phased existing-building upgrade, an occupancy where the load classification is not obvious. Finding out at the final inspection that the AHJ classifies a load as emergency when you built it as optional is the kind of rework that no schedule survives.

What to document

The record is what proves a backup system was classified, built, and tested to the right rules, and it is what the next engineer, the facility manager, and the AHJ all read. The single most important thing it captures is the classification itself: which NEC article governs each backed-up load and why, because that label is what every other requirement hangs off. A backup system with no documented class is a system nobody can verify.

For each backup system, record the class and article, the loads it serves and how they were classified, the source and whether it qualifies for the class, the required and the measured transfer time, the wiring-separation approach, the selective-coordination study and who sealed it, the NFPA 110 Type, Class, and Level where it applies, and the testing program handed to the owner. The table below is the minimum a defensible record carries, and the source-side detail lives in the cross-referenced generator and egress guides.

Common mistakes

• Misclassifying the load, so a life-safety load lands on an optional system or an optional load gets the full weight of an emergency one.
• Running the emergency wiring in shared raceways and boxes with normal or other-system conductors, instead of keeping the 700 wiring independent.
• Skipping selective coordination on a 700 or 701 system, so a downstream fault trips an upstream device and darks the whole branch.
• Designing the wrong transfer time for the class, missing the 10-second emergency window or assuming an optional load needs no clock at all.
• Treating an optional standby system as emergency, or an emergency system as optional, instead of building to the class the load actually is.
• Ignoring the NFPA 110 testing and the exercise program, so a system accepted clean quietly degrades between outages.
• Quoting transfer times and section numbers from memory instead of confirming them against the adopted edition and the AHJ.
• Assuming a data center's heavy IT backup covers the building's life-safety egress loads, which are a separate emergency system.

Field checklist

Standards and references

The NEC, NFPA 70, is where the classification framework lives. Article 700 covers emergency systems, with the life-safety loads, the fast transfer commonly held at 10 seconds, the wiring-independence rule, the permitted sources commonly at 700.12, and the selective-coordination requirement commonly at 700.32 in recent editions. Article 701 covers legally required standby, with the 60-second expectation and its own selective-coordination requirement commonly at 701.32. Article 702 covers optional standby, with no mandated transfer time and no selective-coordination requirement. The section numbers have moved between cycles, so confirm them against the adopted edition.

Beyond the three general articles, the specialized systems have their own. The fire pump power supply falls under its own article, commonly 695, with rules that differ from the general emergency-system rules. Healthcare facilities fall under Article 517 and the essential electrical system, with its life-safety, critical, and equipment branches, alongside NFPA 99. Critical operations power systems fall under Article 708, the COPS article, for facilities that must run continuously for public safety and national security. Each is its own rule set, not a variation on 700.

NFPA 110, the standard for emergency and standby power systems, governs the EPSS performance, classifying it by Type, Class, and Level and setting the installation acceptance and ongoing testing. NFPA 101, the Life Safety Code, sets the egress illumination and the life-safety performance that the emergency loads have to deliver. Equipment is listed under UL, and transfer switches to UL 1008. Above all of these sit the project specification and the AHJ, which adopt the edition, take the amendments, and decide what is enforceable. When a standard and the spec disagree, the stricter controlling document wins.

Units and terms

The vocabulary in this area reads differently across the NEC, NFPA 110, and a manufacturer's submittal, and the same system can be described by an NEC class and an NFPA 110 classification at once, so the labels stack. Transfer time is in seconds, the count from loss of normal power to acceptable power at the load. The NFPA 110 Type is that same time as a rating, while the Class is a run time in hours and the Level is a criticality ranking.

The terms below are the ones that travel across the whole subject. The NEC class and the NFPA 110 classification are related but not identical, so read each on its own terms, and confirm any number against the adopted editions and the AHJ before citing it on a submittal.

Emergency system (NEC 700)

Backup for code-mandated life-safety loads; the strictest class, with the fast transfer, the wiring independence, and selective coordination

Legally required standby (NEC 701)

Backup for code-required loads that are not immediate life safety, with a longer transfer expectation and its own coordination requirement

Optional standby (NEC 702)

Owner-chosen backup for property and continuity, with no code-mandated transfer time

EPSS

Emergency power supply system: the source, prime mover, transfer equipment, and controls taken together under NFPA 110

Type / Class / Level

NFPA 110 ratings: Type is restore time in seconds, Class is on-site run time in hours, Level is criticality

ATS

Automatic transfer switch, which senses the loss of normal power and moves the load to the alternate source

Selective coordination

Overcurrent devices selected so only the device nearest a fault opens, required on emergency and legally required systems

Transfer time

The interval from loss of normal power to acceptable power at the load; the class sets the limit

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• What size standby generator do I need?
• How long must emergency lights stay on?
• How do you size a data center generator?
• What is an NFPA 110 acceptance test?
• What are the conditions to parallel two generators?