Generator paralleling and synchronization field guide

Generator paralleling, and what the common bus buys you

Generator paralleling is running two or more generator sets electrically tied to a common bus so they behave as one larger source. Instead of one big set carrying everything, you have a plant of smaller sets that start, synchronize, close onto the bus, and split the load between them. Done right, the bus does not care which set is feeding it. The load sees one stable source.

The reason to build it that way comes down to three things, and they do not weigh the same on every job. You parallel for capacity when the load is bigger than any single set you want to buy and maintain. You parallel for redundancy when the load cannot go dark if one set drops, the N+1 case. You parallel for efficiency when the load swings and you would rather run two sets near their sweet spot than one huge set lightly loaded and wet-stacking. Most real plants are built for some mix of the three.

This guide is the multi-genset plant. The single set and its automatic transfer switch are their own job, covered in the standby generator and ATS guide, and the question of which loads are emergency versus legally required versus optional, and how fast each has to come back, lives in the emergency and standby systems guide. Here the focus is what changes the moment there is a second set on the bus: the synchronizing, the load sharing, the dispatch, and the protection that none of it needs until two machines have to agree.

Why parallel generators at all?

You parallel because one set cannot give you everything the building needs at once. Capacity is the obvious driver. A data center or a large hospital can need more standby power than any single set the manufacturer builds at a price and a footprint you can live with, so you put several medium sets on a bus and add them up.

Redundancy is usually the stronger reason. With a single set, the set is the system, and the system is down whenever the set is down for service, for a failed start, or for a fault. Put N+1 sets on the bus, size the load so the plant carries it with one set out, and you can lose a machine or pull it for maintenance and the load never notices. That is the whole argument for paralleling on a critical facility, and it is why the count is N+1 or 2N, not just N.

Efficiency and modularity round it out. A diesel running at 20 percent load for years glazes its cylinders and carbons up, so load-demand control runs only the sets the load actually needs and shuts the rest down. And a paralleled plant grows. You add a set and a breaker cubicle later instead of replacing one set with a bigger one. The data center driver is all of these at once: the block load is large, it cannot drop, and it grows.

What is generator synchronization?

Synchronization is matching an incoming generator to the bus it is about to close onto, so that at the instant the breaker closes there is almost no difference between the two sides of the open contacts. A generator is a rotating voltage source. The bus, if it is already live, is another. Tie two live sources together that are not matched and the difference between them has to go somewhere, and it goes as a violent surge of current and a mechanical jolt through both machines.

So before a set's breaker closes onto a live bus, the set is brought into agreement with that bus on four counts at once. The voltage magnitudes match. The frequencies match, which means the engines are turning at the same electrical speed. The phase angle is near zero, the two waveforms crossing zero together. And the phase rotation is the same, A-B-C to A-B-C. Only when all four hold inside a tight window does the close command go out.

Stress this part to anyone learning the plant: synchronizing is not optional and it is not a formality. It is the one step that keeps a paralleling operation from becoming a destructive one. Every other feature on the switchgear assumes the sets came onto the bus in sync. The synchronizing check is what makes that assumption safe.

The four conditions to parallel

Four things have to match between the incoming set and the live bus before the breaker closes. Miss any one and the close is wrong, but they do not fail the same way, and knowing which is which tells you what broke when a sync will not complete.

Voltage magnitude is trimmed by the AVR. A mismatch here does not jolt the machine so much as it throws the reactive sharing off the moment the set is on, the higher-voltage set hogging the kVAR. Frequency is trimmed by the governor, and a frequency difference means the phase angle is sliding, not holding, so the synchroscope keeps turning instead of settling. Phase angle is the timing of the close itself: you close at or near zero degrees, the two waveforms in step. Phase rotation is the one you set once and verify, because it is fixed by how the machine and the bus are wired, not by a control trim.

Phase rotation is the brutal one. Voltage and frequency and angle are all matters of degree, and a small error gives a small surge. Rotation is binary. Get A-B-C against C-B-A and closing the breaker is a bolted fault across the machine, full stop. That is why rotation is proven at commissioning with the set running, on every source, before anyone trusts the auto-sync to ever close that breaker.

The synchroscope and the sync-check relay

The synchroscope is the instrument that shows you the phase relationship as a pointer that rotates around a dial. Twelve o'clock is in phase, zero degrees. When the incoming set is a hair slow or fast against the bus, the pointer rolls, and the speed of the roll is the frequency difference. A pointer creeping slowly clockwise toward twelve o'clock is what a good manual sync looks like. You close at the top, with the needle rising slowly into twelve, not racing past it.

The sync-check relay is the device that does not trust the operator or the auto-sync to get that right. Commonly the ANSI device number 25, it supervises the close: it independently verifies that voltage, frequency, and phase angle are all inside the permissive window before it allows the breaker to close, and it blocks the close if they are not. On an automatic plant the sync controller drives the governor and AVR to bring the set in and then commands the close, but the 25 relay sits in series with that command as the last gate.

Auto-sync is the norm on modern switchgear and manual is the backup. Either way, the sync-check is the safety, not the convenience. The exact relay model, its window settings, and whether the function lives in a discrete relay or inside the genset controller are set by the design and the switchgear manufacturer. Do not change those setpoints in the field without the design behind you.

What happens if you close a generator out of sync?

Closing a generator breaker out of sync slams two mismatched sources together, and the energy in the mismatch discharges in a fraction of a cycle. The result is a large transient current, often on the order of a fault, and a sudden torque reaction in the engine and the alternator. The machine is yanked into step with the bus whether it wants to go or not.

The damage is mechanical as much as electrical. The torque slams through the coupling, the crankshaft, and the alternator rotor. People have cracked crankshafts, sheared coupling bolts, shifted stator windings, and twisted shafts on a single bad close. At best the breaker trips on instantaneous overcurrent and you have stressed the machine you did not see fail yet. At worst you have a set on the ground and a long lead time on parts.

This is why the sync-check is not negotiable and why nobody bypasses it to make a stubborn set close. A set that will not synchronize is telling you something: the rotation is wrong, the voltage is off, the governor will not settle, or the metering is lying. Find that before the breaker closes. The check that blocks a bad close has just saved the most expensive part of the plant, and that is exactly its job.

Sharing real power (kW) with the governor

Once sets are paralleled, they have to split the real load, the kW, in proportion to their size, or one set carries the building while the other loafs. Real power sharing is a governor job, because real power follows engine torque, and torque follows the fuel the governor commands. Tell a set to make more power and you are telling its engine to push harder against the others on the bus.

The frequency is shared by everyone on the bus, so you cannot make one set take more kW by speeding it up. They all turn at the same electrical speed once tied together. What you change is the power each set produces at that common speed, by raising or lowering its governor reference. Load-sharing control coordinates those references so that if the plant is at 60 percent, every set is near 60 percent of its own rating, not one at 90 and one at 30.

When kW sharing is off, you see it on the metering first and on the engines second. The hogging set runs hot and loud while the loafing set barely works, and if the imbalance is bad enough the underloaded set can be driven backward into motoring, which the reverse-power relay should catch. Governor mismatch, a wrong droop setting, or sets running in two different control modes are the usual causes, and they get sorted on a load bank at commissioning, not during an outage.

Sharing reactive power (kVAR) with the AVR

Reactive power, the kVAR, is the other half of the load and it shares on a separate channel. Where real power follows the governor, reactive power follows excitation, so kVAR sharing is an AVR job. The set with the higher excitation, the stronger field, pushes out more reactive current, the same way the higher-voltage set tried to during synchronizing.

Left alone, two AVRs holding slightly different voltage setpoints will fight over the reactive load and one set will swing toward leading or lagging while the other takes the opposite. The fix is the same idea as governor droop, applied to voltage. Reactive droop, sometimes wired as a cross-current or droop CT scheme, lets each set's voltage sag a little as it picks up reactive load, which makes the sets settle into a stable split instead of chasing each other.

kVAR imbalance is quieter than kW imbalance because it does not show up as an engine running hot. It shows up as circulating reactive current between the sets, extra heating in the windings for no useful output, and a power factor at each machine that does not match the plant. Watch the per-set kVAR and power factor on the metering during the load test. If one set is deep lagging while another is leading at the same kW, the AVR sharing needs trimming.

What is the difference between droop and isochronous?

Droop and isochronous are the two governor control modes, and the difference is what happens to frequency as load goes up. In droop, the speed reference falls as the set takes load, so frequency sags a little from no load to full load, commonly 2 to 4 percent. That sag is the sharing mechanism: tie several droop sets together and they each settle at the load where their drooping curves meet the common frequency, so load splits without any communication between them.

Isochronous holds frequency dead flat regardless of load. A single set on isochronous gives you rock-steady 60 Hz, which is what you want, but two isochronous sets with no coordination will fight, because each one tries to hold the exact frequency and they cannot both win. So isochronous paralleling needs a load-sharing line, a communication link or load-share controller that tells the sets how to split the kW while one reference holds the frequency.

The practical lay of it: droop is simple and stable and gives you sharing for free, at the cost of frequency that moves with load. Isochronous gives you constant frequency at the cost of needing real load-sharing controls. Modern paralleling switchgear runs isochronous load sharing across all the sets, which is constant frequency plus coordinated sharing. The classic rule for a simple manual setup, one set isochronous to set frequency and the rest in droop, still applies on basic plants, but the controller does the coordinating on anything modern.

Paralleling switchgear and the control system

Paralleling switchgear is the lineup that ties the sets together and runs the show. It is more than breakers in a can. It is the generator breakers, the bus, the metering and protective relays, the synchronizing equipment, and the control layer that decides which set starts, when it closes, and how the load splits.

Architectures vary, and the design picks one. In a controller-based plant the intelligence lives in the genset controllers and they talk to each other over a network, with the switchgear mostly providing the breakers, the bus, and the protection. In a master-control plant a central PLC or master controller runs the sequencing and the load sharing and the sets follow it. Both are common, and both have to handle the same jobs: start and stop the sets, synchronize and close each breaker, share kW and kVAR, dispatch on load demand, and shed or add load.

What you check at commissioning is the same regardless of architecture. The sync function and the 25 supervision work on every breaker. The load sharing holds across the range on a load bank. The protection trips on the conditions it is supposed to. The dispatch starts and stops the right sets in the right order. The exact relays, the controller models, and the network are the manufacturer's design, and the submittal and the sequence of operations are what you commission against, not a generic idea of how it should work.

The genset controller

The modern digital genset controller is the box on each set that has absorbed most of what used to be a rack of discrete relays and a separate synchronizer. Names you will see in the field include Woodward and Deep Sea, among others, and the integrated paralleling controllers from the set manufacturers. One controller per set handles the engine, the alternator, and the paralleling.

On the paralleling side it does the synchronizing, driving the governor and AVR to bring the set into the window and commanding the close through the sync-check. It runs the load sharing for both kW and kVAR, usually by talking to the other controllers over a network so the sets coordinate without a separate master. It carries protection functions, and it handles the dispatch logic, starting and stopping its set on the load-demand signal and according to its priority.

The thing to know in the field is that the controller is configured, not just installed. The setpoints, the protection, the sharing parameters, and the sequence are loaded to match the design. A controller swapped out for a spare without its configuration is a set that will not parallel correctly, even though the hardware is identical. Back up the configuration, label it, and treat it as part of the commissioning record.

The paralleling sequence, start to load

The operation runs in an order, and the order is the safe part. On a loss of utility, the transfer scheme signals the plant to start. Either all the sets crank at once or a first set is selected, depending on the design. The first set up to speed and voltage closes onto a dead bus, because there is nothing to synchronize to yet.

With the bus live behind the first set, every other set synchronizes to that live bus and closes in turn through its sync-check. As each set comes on, the load sharing pulls it up to its share of whatever the bus is carrying, so adding a set does not bump the load that is already running. Then the load is allowed onto the bus, usually in blocks with their own priorities and time delays, so the plant adds load in steps it can swallow rather than all at once.

On a plant with load-demand control, the sequence keeps working after the plant is up. As load rises the controller starts and synchronizes another set before the running sets are maxed, and as load falls it unloads a set, opens its breaker, and shuts it down to save fuel and run hours. The whole cycle, start, dead-bus close, sync the rest, share, add load, then dispatch up and down, is what the sequence of operations document spells out and what you prove step by step at commissioning.

Dead-bus closing and first-on logic

The first set onto a dead bus does not synchronize, because there is no voltage on the bus to match. It just closes, energizes the bus, and becomes the reference everything else syncs to. That is the dead-bus close, and it is the one close in the whole sequence that skips the sync-check, by design, because syncing to nothing is meaningless.

The catch is making sure only one set ever does that. If two sets both see a dead bus and both decide to be first, they can close onto each other through the bus, which is an out-of-sync close by another name. So the control has dead-bus first-on logic, sometimes called dead-bus arbitration: the sets stagger their dead-bus close commands, or a priority decides who goes first, so exactly one set energizes the bus and the rest are forced to synchronize to it.

This is a place where a control failure is expensive, so it gets tested deliberately. At commissioning you confirm that with multiple sets racing to a dead bus, one and only one closes first and the others wait and sync. A plant that lets two sets close dead-bus together has a logic gap that will eventually find an outage to fail on.

Load-demand dispatch and runtime rotation

Load-demand control runs only the sets the load needs and parks the rest. The controller watches the total load against the online capacity, and when the load can be carried by fewer sets with margin to spare, it unloads and stops a set. When the load climbs toward the online capacity, it starts and synchronizes another set before the running ones run out of room. That keeps the running sets in their efficient band instead of idling a big plant at light load.

The margin and the time delays are tuned, not guessed. Stop a set too eagerly and the next load step trips the plant because the remaining sets cannot pick it up fast enough. Start too late and you overload before the new set is on. The dispatch carries a spinning-reserve margin so there is always enough headroom to ride a load step while the next set comes up.

Priority and rotation keep the run hours even. If the same set is always first on and last off, it wears while the others sit, so the dispatch rotates the priority, often by accumulated run time, so the fleet ages together and maintenance lands at sensible intervals. Set the rotation wrong and you find out at the next service, when one set has triple the hours of its neighbors.

Paralleling with the utility

Paralleling the plant with the utility is a different animal from paralleling sets with each other, because now you are tying your machines to the grid. Plants do it for soft load transfer, so the load never blinks when it moves between utility and generator, and for peak shaving or export, where the sets run alongside the utility to shave demand or push power back. The momentary or continuous tie to the grid is what makes it a closed-transition or a true parallel operation.

The grid does not care about your machines, and a fault or a swing on the utility side now reaches your sets, so utility paralleling carries interconnect protection the island plant does not need. The scheme commonly includes sync-check, under and over voltage, under and over frequency, and reverse or directional power, in the ANSI scheme the 25, 27, 59, 81, and 32 functions, plus anti-islanding so your sets drop the tie and do not keep energizing a utility section that has gone dead. That protection of a dead line is what keeps a lineman from being backfed.

Where the plant exports or runs in parallel with the utility, the interconnection is governed by the utility's requirements and commonly by IEEE 1547 for the interconnection of distributed resources, along with the local utility's own interconnect agreement. The exact relay list, settings, and witness testing are set by the utility, the design, and IEEE 1547 as adopted. Confirm them with the utility before you energize the tie, not after.

Open transition vs closed transition

When the load moves between the utility and the generator, it moves one of two ways, and the difference is whether the two sources are ever tied together. Open transition breaks before it makes: it opens the source the load is on, the load is dead for a moment, then it closes the other source. That blink is a few cycles to a second or so, and for many loads it is fine. It needs no synchronizing to the utility because the two sources are never connected.

Closed transition makes before it breaks: it synchronizes the generator to the utility, closes so both sources feed the load at once for a brief overlap, then opens the source it is leaving. The load never loses power. The price is that for that overlap you are paralleled with the grid, so you need the sync-check and the interconnect protection, and you need the utility's blessing for even a momentary parallel.

Pick open transition when a short outage on retransfer is acceptable and you want to avoid the utility-paralleling protection and approvals. Pick closed transition when the load cannot take the blink, a data center or a process that trips on a few cycles of interruption. Soft loading, ramping the generator load up or down across the closed window, is the extended version, and it is what keeps the utility from seeing a step when the plant takes or sheds the whole load.

Protection on a paralleling bus

A paralleled bus needs protection a single set does not, because now a fault or a misbehaving machine can be fed by every other set on the bus. The protection has two jobs: clear faults on the bus and the feeders, and protect each set from the others and from the bus.

The set-protection functions you commonly see, in the ANSI device numbering, include reverse or directional power to catch a set being motored, sync-check to supervise every paralleling close, over and under voltage and over and under frequency to drop a set that the bus is dragging out of limits, and loss of field to catch an excitation failure that would pull reactive power the wrong way. Differential and overcurrent handle the faults. The reverse-power and the sync-check are the two that exist specifically because the set is paralleled.

Treat the relay numbers as a vocabulary, not a specification. The actual protection scheme, the functions enabled, and the settings come from the coordination study, the design, and the switchgear and relay manufacturer, and they are proven by injection testing at commissioning per the project and the NETA acceptance specification. Stress two of them above the rest when you walk a plant: synchronizing supervision, so nothing closes out of phase, and reverse power, so a dead engine does not get spun by its neighbors.

Reverse power (32) protection

Reverse power protection catches a generator that has stopped making power and started absorbing it. Commonly the ANSI device number 32, it watches the direction of real power at the set's breaker, and if power flows into the generator instead of out of it, past a setpoint for a set time, it trips the set off the bus.

It matters because of what a motored set is doing. If a set's engine falters, loses fuel, or has its governor backed off too far while the breaker is still closed, the other sets on the bus drive it as a motor, spinning the engine through the alternator. A motored diesel is not just useless load on the healthy sets. It can be damaging, and on some prime movers a motored condition is its own hazard. The reverse-power relay sees the power going the wrong way and opens that set's breaker before the others waste themselves carrying it.

On a utility parallel, the same function does double duty, blocking export where export is not allowed and catching the set being back-driven by the grid. The pickup and time delay are set to ride through the normal reverse swing at synchronizing and load changes while still catching a real motoring event, and those settings belong to the design and the relay manufacturer. This is the protection people forget on a first paralleling job, and it is the one that quietly saves an engine.

Commissioning and the load-bank test

A paralleling plant is only as good as the day it was proven, and the proving is more than starting the sets. You confirm phase rotation on every set and every source first, because rotation is the close that destroys the machine. Then you sync and close each breaker, watching the sync-check do its job and confirming it blocks a bad close.

The heart of the commissioning is the load-share test on a load bank. You bring the plant up, load it across its range with resistive and where called for reactive load banks, and confirm the sets split kW and kVAR in proportion at light, medium, and full load. You watch the per-set metering, not just the bus total, because a plant can carry the right total with one set hogging and one loafing. You prove the dispatch by walking the load up and down and confirming the right sets start, sync, share, unload, and stop in the right order.

Then you prove the failures. You fail a running set and confirm the others pick up the load without dropping the bus, the N+1 case the whole plant exists for. You confirm reverse power trips a motored set and the protection clears a simulated fault. On a critical facility this folds into an integrated systems test with the transfer scheme and the loads, and the load-bank and integrated-test detail lives alongside the emergency and standby systems guide. The acceptance testing of the gear and relays follows the NETA specification and the project, witnessed and recorded. A plant that was never load-tested as a plant has not been commissioned, it has been turned on.

The data-center paralleled plant

Data centers are where paralleling shows up at its most demanding, because the load cannot drop and it arrives in big steps. The plant is built N+1 or 2N, sometimes as several independent paralleled blocks feeding redundant distribution, so a set or even a whole block can fail or be serviced with the load still up. The paralleling switchgear is the part of the design that ties the sets into those redundant blocks and decides how a lost set is covered.

Block loading is the hard part. When the utility fails, the UPS systems carry the critical load on battery for the seconds it takes the plant to start and close, and then the whole block of UPS rectifiers and the mechanical cooling load drops onto the generators at once. The sets have to accept that step without the frequency or voltage diving far enough to trip the bus. Large motor loads on the cooling plant add inrush on top of the step.

So the data-center plant is sized and commissioned for the block, not the steady load. You confirm at commissioning that the plant takes the design block load in the steps the sequence calls for and holds frequency and voltage inside the window through each step. Get the block loading wrong and the plant starts fine, carries a load bank fine, and then trips the first time a real outage dumps the building onto it at once.

Runtime balancing and maintenance

A paralleled plant needs its run hours kept even, or maintenance bunches up on one set and the redundancy quietly erodes. Load-demand dispatch with priority rotation does most of this automatically by promoting the least-run set, but it only works if the rotation is actually enabled and the priorities are set, which is worth confirming rather than assuming.

Exercise the plant the way it will run, paralleled and under load, not just one set at a time at no load. A monthly run that only proves each engine cranks does not prove the sets still synchronize and share, which is what they have to do in an outage. Where the standby system has an exercise requirement, the multi-set version means proving the paralleling and the dispatch, not just the engines.

When you pull a set for service, the plant is running on its N margin, so the redundancy is spent until that set is back. Plan the service window for when the load is low and another failure is least likely, and put the set back through a sync and load check before you call it returned to service. A set that was off the bus for a controller firmware update or a governor adjustment has to prove it still parallels before it counts toward N+1 again.

What to document

A paralleling plant has a long memory problem: the settings that make it share and protect correctly live in controllers and relays, and they are invisible until something is wrong. The record is what lets the next engineer confirm the plant still matches its design after a part swap or a firmware update.

Capture the system one-line and the sequence of operations, the synchronizing setpoints and which device supervises each close, the load-sharing mode and parameters for kW and kVAR, the protective relay functions and settings with the coordination study behind them, the dispatch priorities and rotation scheme, the dead-bus first-on logic, and the genset controller configuration backups. Record the commissioning results, the load-share numbers per set across the range, and the failure and block-load test outcomes, so a future test has a baseline to compare against.

Common mistakes

• Closing a breaker out of sync, or bypassing the sync-check to force a stubborn set to close.
• Skipping the phase-rotation check on a source, and finding out at the first close that A-B-C met C-B-A.
• Poor kW or kVAR sharing left uncorrected, from governor or AVR mismatch or sets in different control modes.
• Running sets in mixed droop and isochronous without the load-sharing controls that mode actually needs.
• No reverse-power protection, so a motored set gets driven by its neighbors until something breaks.
• No dead-bus first-on logic, letting two sets close onto a dead bus at once.
• Treating utility paralleling like island paralleling, with no interconnect protection, anti-islanding, or utility approval.
• Commissioning the engines but never load-testing the plant as a plant, so sharing and dispatch are unproven.

Field checklist

Standards and references

The framework for the backup system lives in the NEC, NFPA 70, which sorts the system into emergency, legally required, or optional standby under Articles 700, 701, and 702, and that classification drives how a multi-set plant has to be built and tested. The classification detail belongs to the emergency and standby systems guide. NFPA 110 covers emergency and standby power systems as a system, including the testing and maintenance the owner inherits, and a paralleled plant has to be exercised and maintained against it.

The paralleling and synchronizing themselves are governed less by a single code article and more by the design, the coordination study, and the switchgear and genset manufacturer. The protective relay functions are commonly identified by ANSI device numbers, sync-check 25, reverse power 32, under and over voltage 27 and 59, and under and over frequency 81 among them, but the scheme, the settings, and the testing come from the design and the relay manufacturer, proven by acceptance testing under the NETA specification.

Where the plant parallels with the utility, the interconnection is governed by the utility's interconnect requirements and commonly by IEEE 1547 for distributed resources, as adopted by the utility and the jurisdiction. Confirm every relay number, setpoint, and witness requirement against the adopted code edition, the project specification, the manufacturer's documentation, and the utility before you cite them on a submittal or rely on them in the field. The two things to never let slide regardless of which standard governs are synchronizing supervision and reverse-power protection.

Units, terms, and conversions

Paralleling spans the engine side and the electrical side, so the same plant gets described in mechanical and electrical units across the engine sheet, the alternator data, and the switchgear submittal.

Real power is kilowatts, kW, and it scales to megawatts, MW, on large plants. Reactive power is kilovolt-amperes reactive, kVAR, and apparent power is kilovolt-amperes, kVA, with kW equal to kVA times power factor. Frequency is hertz, Hz, 60 Hz in North America and 50 Hz in much of the world, and engine speed in rpm relates to it by the number of alternator poles. Phase angle is in degrees, zero being in phase. Droop is a percentage of frequency or voltage from no load to full load.

Synchronization

Matching an incoming set's voltage, frequency, phase angle, and rotation to the bus before its breaker closes

Synchroscope / sync-check (25)

The instrument showing phase relationship and the relay (ANSI 25) that supervises and permits the close

Droop

Frequency or voltage falling with load, commonly 2 to 4 percent, which makes parallel sets share without a comms link

Isochronous

Constant frequency regardless of load, which needs a load-sharing line to parallel sets

Reverse power (32)

Protection that trips a set being motored, when real power flows into it instead of out

Dead-bus close

The first set energizing an unpowered bus without synchronizing, governed by first-on logic so only one set does it

Load-demand dispatch

Running only the sets the load needs, starting and stopping sets on load with priority rotation for even run hours

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• What size standby generator do I need?
• What is the difference between emergency and standby power?
• What are the conditions to parallel two generators?
• What is a load bank test?
• What is a separately derived system?