Electrical
Power factor correction and capacitor banks for electrical crews
Read the kW, kVAR, and kVA triangle, size the capacitor bank to a target power factor, and keep the caps off the harmonics that blow them.
Direct answer
Power factor is the ratio of real power in kilowatts to apparent power in kilovolt-amperes, the fraction of the current doing actual work. Lagging motors draw reactive kVAR the utility supplies but cannot bill as work, so low power factor triggers penalties. A capacitor bank supplies that reactive power locally. The utility tariff governs.
Key takeaways
- Power factor is real power (kW) divided by apparent power (kVA); a capacitor bank supplies lagging displacement kVAR locally to raise it.
- Size correction kVAR = kW x (tan of present angle minus tan of target angle), aiming for 0.95 to 0.98 at typical load, not the peak.
- Utility PF penalties commonly start below 0.90 to 0.95; confirm the exact threshold and formula on the actual tariff before quoting savings.
- In buildings with drives, measure harmonics first and specify detuned reactors (tuned around the 4.2th order), or resonance amplifies harmonic current and blows the caps.
- Capacitors hold a lethal charge after disconnect; never trust the bleed resistor or wait time. Open, lock out, wait, verify dead with a tested meter, then short and ground each phase (NEC Article 460, NFPA 70E).
Power factor, and why a low one costs money
Power factor is the ratio of real power to apparent power: the kilowatts that do actual work divided by the kilovolt-amperes the system has to carry to deliver them. A power factor of 1.0 means every amp is working. A power factor of 0.7 means a third of the current you are paying to push through the wire and the transformer is doing nothing but sloshing back and forth.
The reason it costs money is the part most people skip. Induction motors, transformers, and welders are inductive loads. They pull a magnetizing current, the reactive kVAR, that builds and collapses the magnetic field every cycle. That current is real. It heats the conductors, loads the transformer, and the utility has to supply it. But it does no work, so the utility cannot bill it as energy. So they bill it another way, through a power factor penalty, and that penalty is the number that lands on the bill and starts the conversation.
The fix on most jobs is a capacitor bank. Capacitors are the mirror image of an inductive load: they supply reactive power instead of drawing it. Put the right amount of capacitance near the motors and the reactive current circulates locally between the caps and the motors instead of being dragged all the way back from the utility. The real load calculation behind the building, the connected load and demand, is a separate exercise covered in the load-calculation guide. This one is about the reactive piece the load calc does not size.
Real, reactive, and apparent power: the kW, kVAR, kVA triangle
Three quantities describe an AC load, and they form a right triangle. Real power, in kilowatts, is the horizontal leg, the power that turns the shaft and makes heat and light. Reactive power, in kilovolt-amperes-reactive (kVAR), is the vertical leg, the power that magnetizes the iron and returns to the source each cycle. Apparent power, in kilovolt-amperes (kVA), is the hypotenuse, the vector sum of the two and the total the conductors and transformer actually carry.
Power factor is the cosine of the angle between the real power and the apparent power. Pull the kVAR leg down to zero and the angle closes, the hypotenuse drops onto the real-power leg, and power factor goes to 1.0. That is exactly what a capacitor does: it cancels the inductive kVAR so the kVA shrinks toward the kW. The kW does not change. You are not making the motor use less real power. You are shrinking the apparent power the system has to deliver to run it.
One distinction matters before you size anything. The cosine-of-the-angle figure is the displacement power factor, which is what capacitors correct. On a building full of drives and electronics there is also distortion power factor from harmonic current, and capacitors do nothing for that. Worse, they can make it dangerous. Confuse the two and you install a bank that does not fix the problem and may blow itself apart, which is the resonance trap covered further down and in the harmonics guide.
PF = cosθ = kW / kVAkVA = √(kW2 + kVAR2)kVAR = √(kVA2 − kW2)- kW
- Real power, the power that does work, turns the shaft, and makes heat and light
- kVAR
- Reactive power, the magnetizing power an inductive load draws and returns each cycle
- kVA
- Apparent power, the vector sum of kW and kVAR that the conductors and transformer carry
- Displacement PF
- Power factor from the phase shift between voltage and current, the part capacitors correct
Why do utilities penalize low power factor?
Utilities penalize low power factor because they have to build and maintain the wires and transformers to carry your full apparent power, the kVA, while they can only bill the real energy, the kWh. The reactive current you pull occupies their system, ages their equipment, and earns them nothing. The penalty puts that cost back on the customer who created it, and it nudges that customer to fix it.
The most common penalty is a billing-demand adjustment. Instead of billing your actual kW demand, the tariff bills kW divided by your power factor once it falls below a threshold, which inflates the demand charge in direct proportion to how bad the power factor is. A 1000 kW plant running at 0.80 power factor gets billed as if it pulled 1250 kW, a 25 percent hit on the demand portion of the bill. Some tariffs instead bill on kVA demand directly, or add a flat reactive charge per kVAR-hour.
The threshold varies by utility and rate class, and this is the number you confirm against the actual tariff, not a rule of thumb. Many utilities start the penalty below 0.90, some below 0.85, and others hold large customers to 0.95. Industrial and large-commercial rate classes are where it bites. Small commercial and residential rates usually do not carry a power factor clause at all. Pull the tariff sheet or call the utility, because the threshold and the formula on that sheet are what govern the whole economic case for the bank.
Two facilities with the same penalty can land on different answers. A bank pays for itself in months at one site and never quite pencils out at another, depending on the tariff structure, the demand rate, and how low the power factor actually runs. Get the tariff before you quote the savings.
| Penalty mechanism | How it hits the bill |
|---|---|
| Demand multiplier (kW / PF) | Inflates billed demand when PF falls below threshold |
| kVA demand billing | Bills the full apparent power, so reactive load counts directly |
| Reactive energy charge | Flat charge per kVAR-hour of reactive consumption |
| Threshold | Commonly 0.90 to 0.95, set by the utility tariff, verify it |
What drags power factor down
Inductive loads drag it down, and the worst offenders are the lightly loaded ones. An induction motor pulls roughly the same magnetizing kVAR whether it is loaded or idling, but it only produces real kW when it is working. So a motor running at half load has a much worse power factor than the same motor at full load, because the fixed reactive draw is now a bigger share of a smaller real load. Oversized motors are low-power-factor machines by design.
The usual list, ranked by how often it is the cause: lightly loaded or oversized induction motors, then transformers carrying little load, then welders and arc furnaces, then induction heating and older fluorescent or HID lighting with magnetic ballasts. A plant that runs a fleet of motors at part load most of the day, an idling conveyor system, a shop with big welders, these are the classic low-power-factor sites.
The pattern to notice is part load. A facility designed for a peak it rarely hits runs its motors and transformers light most of the time, and that is exactly the condition that produces a bad power factor and the penalty that follows. The fix is not to load the motors harder. It is to supply the reactive power they need from somewhere closer than the utility.
How do you size a capacitor bank?
You size a capacitor bank in kVAR to move the load from its present power factor to a target power factor, and the formula is the difference of two tangents. Take the real load in kW. Multiply it by the tangent of the angle at the present power factor minus the tangent of the angle at the target. The result is the correction kVAR the capacitors must supply. The kW stays put through the whole calculation, because correction does not change the work, only the reactive piece around it.
The target is almost always just past the penalty threshold, not unity. Correcting to 0.95 is the common aim, sometimes 0.98, because chasing 1.0 buys little extra penalty relief and raises the risk of over-correcting into a leading power factor when the load drops off. Size to the typical operating load, the condition the plant sits at most of the day, not the rare peak. Size to the peak and you have too much capacitance the rest of the time.
Once you have the kVAR, the capacitance follows from the system voltage and frequency, but you order the bank in kVAR at a rated voltage, not in microfarads. One caution the formula hides: a capacitor's kVAR scales with the square of the voltage applied to it. A bank rated 100 kVAR at 480 V delivers only about 80 kVAR if the bus actually sits at 430 V. Match the capacitor's voltage rating to the real system voltage or the correction comes up short.
kVARc = kW × (tanθ1 − tanθ2)θ = cos-1(PF)kVARactual = kVARrated × (Vactual / Vrated)2- kVAR_c
- Correction reactive power the capacitors must supply to reach the target power factor
- PF1 / theta1
- Present power factor and its phase angle, before correction
- PF2 / theta2
- Target power factor and its phase angle, commonly 0.95 to 0.98
Field example: correcting a 300 kW plant to 0.95
A plant pulls 300 kW at a measured power factor of 0.78. The utility penalty starts at 0.90, so it has been on the bill for months. The target is 0.95, a comfortable margin past the threshold without over-correcting.
Work the tangents. At 0.78 the angle is about 38.7 degrees and its tangent is about 0.80. At 0.95 the angle is about 18.2 degrees and its tangent is about 0.329. The correction is 300 times (0.80 minus 0.329), which is 300 times 0.471, about 141 kVAR. Round to the standard step the manufacturer offers, commonly 150 kVAR, and confirm the result lands at or just above 0.95 rather than overshooting.
Check what it bought. Before correction the apparent power was 300 divided by 0.78, about 385 kVA. After, it is 300 divided by 0.95, about 316 kVA. That is roughly 69 kVA of apparent load taken off the service and the transformer, freeing capacity you can use for new load without upsizing the gear. On a demand-multiplier tariff that grosses billed demand up by the ratio of the 0.90 floor to the actual power factor, the penalty that billed 300 kW as about 346 kW now disappears entirely. That recovered capacity is often a bigger prize than the penalty itself.
| Quantity | Before (0.78) | After (0.95) |
|---|---|---|
| Real power | 300 kW | 300 kW |
| Apparent power | 385 kVA | 316 kVA |
| Reactive power | 241 kVAR | 99 kVAR |
| Capacitor kVAR supplied | 0 | ~141 (install 150) |
| Service/transformer load | 385 kVA | 316 kVA |
What a capacitor bank does
A capacitor bank supplies reactive power on site so the utility does not have to ship it. The capacitors and the inductive loads trade the same magnetizing current back and forth between them, cycle by cycle, instead of pulling it all the way from the substation. Everything upstream of the bank, the service conductors, the main, the transformer, and the utility line, now carries only the real current plus whatever reactive the bank did not cover.
That is the whole mechanism. The bank is a local reactive source parked next to the loads that need it. The closer it sits to the load, the more of the system it unloads, because the reactive current never has to travel the run between the load and the bank. Put the bank at the service entrance and you fix the utility penalty but the feeders and the transformer inside the building still carry the reactive current. Put it at the motor and you unload everything from the motor back.
What the bank does not do is reduce the motor's draw or fix harmonics. It cancels displacement kVAR, full stop. If the meter shows a bad power factor and the cause is harmonic distortion rather than lagging magnetizing current, a plain capacitor bank is the wrong tool and a dangerous one. Confirm the load is genuinely inductive before you reach for caps.
Fixed or automatic capacitor bank?
Fixed capacitors are hardwired and supply a constant kVAR. Automatic banks switch capacitor stages in and out with contactors under a controller that watches the power factor in real time. The choice comes down to how steady the load is. A steady load takes fixed capacitors. A load that swings takes an automatic bank.
Fixed correction belongs at a steady load, and the cleanest place is right at a motor that runs at a constant load, switched with the motor so the caps energize and de-energize with it. Base-load motors, a constant HVAC load, a transformer's no-load magnetizing kVAR, these all take fixed banks. Fixed is cheaper, has no controller or contactors to fail, and if one step fails you still have the rest. The risk is over-correction: when the steady load drops away, fixed capacitance keeps pushing kVAR into a system that no longer needs it, and the power factor swings leading.
Automatic banks belong where the load varies more than about 20 percent through the day, a manufacturing plant, a mill, a building whose connected equipment cycles on and off. The controller adds and sheds stages to hold the target power factor as the load moves, which keeps the correction matched and avoids the over-correction a fixed bank would cause at light load. The cost is the controller, the switching contactors that wear, and a tuning job at commissioning so the stages do not chatter in and out, the hunting problem covered below. Many real plants use both: fixed correction at the big steady motors, an automatic bank at the service for everything else.
| Factor | Fixed | Automatic |
|---|---|---|
| Best for | Steady, constant load | Varying load (>20% swing) |
| Output | Constant kVAR | Switched stages, matched to load |
| Location | At the motor or steady load | At the MCC or service |
| Risk | Over-correction at light load | Controller hunting, contactor wear |
| Cost and parts | Lower, no controller | Higher, controller plus contactors |
Where to put the bank
Location decides how much of the system the bank unloads. The reactive current only stops flowing on the segment between the bank and the utility, so the farther down toward the load you place the caps, the more wire and equipment you take the reactive load off of. Three positions are common, and they trade unloading against cost and flexibility.
At the motor, switched with it, gives the most unloading. The feeder, the MCC bus, the transformer, and the service all shed that motor's reactive current. It is the right call for a large motor that runs a lot. The downside is one cap bank per motor and care with the cap kVAR relative to the motor so you do not self-excite an unloaded motor on coast-down.
At the motor control center unloads everything from the MCC upstream and lets one bank serve a group of motors, a good middle ground. At the service entrance is the cheapest single installation and it kills the utility penalty, but every feeder and the transformer inside the building still carries the reactive current, so you fix the bill without relieving the internal capacity. If the goal is freeing transformer and feeder capacity for new load, push the correction down toward the loads, not up at the meter.
The harmonic resonance that blows capacitors
This is the part that ruins banks and the reason a capacitor job in a modern building is never just capacitors. A capacitor and the supply transformer's inductance form a resonant circuit, a parallel LC tank with a natural frequency. If that natural frequency lands on or near a harmonic the building is already producing, usually the 5th at 300 Hz or the 7th at 420 Hz, the tank rings. Harmonic current that was a nuisance gets amplified many times over, and it pours into the lowest-impedance path, which is the capacitors.
What you find is failed caps. Bulging or burst cans, fuses that keep opening, reactors and caps running hot, a buzzing bank, distorted voltage, and breakers that trip for no load reason. A bank that worked fine for years can start failing after a new drive or UPS goes in and shifts the harmonic spectrum into the resonance. Add raw capacitors to a building full of variable frequency drives and you have built a harmonic amplifier and pointed it at your own equipment.
The fix is a detuned capacitor bank: a reactor in series with each capacitor step, sized so the series LC resonance sits below the lowest significant harmonic, commonly tuned around the 4.2th order, under the 5th. Below its tuning point the step still delivers reactive power and corrects power factor. Above it the step looks inductive, so it cannot resonate with the system and it offers a controlled, mild path for harmonic current instead of a sharp one. On a site with real distortion you specify detuned reactors as a default, not an upgrade. Where the harmonics are heavy enough to need actual filtering, that is a harmonic-filter question, covered in the harmonics guide. The short version: measure the harmonics before you size the caps, and if there are drives, plan on detuned reactors.
Protection, switching, and inrush
Capacitors switch hard. An uncharged capacitor looks like a short circuit at the instant of connection, so energizing a step draws a large, fast inrush current, and switching one step onto a bus where another step is already energized produces an even sharper back-to-back inrush as the charged step dumps into the uncharged one. That inrush is what welds contactors, blows fuses, and cracks capacitor cans over time.
So a capacitor bank is built around the switching, not just the caps. Capacitor-duty contactors carry pre-insertion resistors or inductors that take the first bite of the inrush before the main contacts close. Capacitor-rated fuses, sized for the cap's characteristics rather than ordinary load fuses, protect each step. The conductors and overcurrent device are sized above the cap's rated current, commonly at 135 percent of rated capacitor current, because a capacitor's actual current runs above nameplate from tolerance, overvoltage, and harmonics. Confirm the exact sizing and the protection rules against the adopted code, the capacitor sections of NEC Article 460, and the manufacturer's data, since these are enforceable conductor and overcurrent requirements, not rules of thumb.
Every capacitor also needs a discharge path, a bleed resistor or discharge device that drains the stored charge after disconnection. That device is required, it is part of the protection, and it is the subject of the next section because it is also a safety item that fails silently.
The stored charge that bites the next person
A capacitor holds its charge after you open the disconnect. The bus is dead, the lockout is on, and the caps are still sitting at peak line voltage with enough stored energy to kill. This is the hazard that catches people who treat a cap bank like an ordinary panel. You do not reach into a capacitor bank because the upstream device is open. You reach in after the charge is gone and you have proven it with a meter.
Capacitors are required to carry a discharge means that bleeds the residual voltage down after disconnection. The NEC capacitor provisions, Article 460, set the requirement: the residual voltage has to drop to a safe level, on the order of 50 volts, within about a minute for the lower-voltage units the trade usually installs and a longer window for higher-voltage banks. Confirm the exact threshold and time against the adopted edition. NFPA 70E treats releasing stored energy as part of establishing an electrically safe work condition, and that is the rule that actually governs your hand going in.
Here is the blunt part. The bleed resistor is the single most common silent failure in the bank. It fails open, the caps never drain, and the next person assumes the wait time did its job and gets the full charge. So you do not trust the resistor and you do not trust the clock. After the wait, you verify dead with a meter you tested on a known source, and on larger units you apply a manufacturer-approved discharge stick to each phase and phase-to-ground before any contact. Verify dead, then short and ground. That sequence is not optional and it is not negotiable.
Over-correction and a leading power factor
Too much capacitance is its own problem. When the capacitive kVAR exceeds the inductive kVAR the load is drawing, the power factor swings from lagging to leading, and a leading power factor raises the voltage on the bus. The classic case is a fixed bank sized for full load while the plant sits at light load overnight or on a weekend. The motors that were absorbing the cap kVAR are off, the caps keep pushing, and the bus voltage climbs.
The damage from over-voltage is real and it compounds. Capacitor current rises with voltage, so an over-voltage stresses the very caps causing it, shortening their life. Sensitive equipment on the bus sees voltage above its rating. And a leading power factor can earn a penalty of its own under some tariffs, so you can over-correct your way back onto the bill you were trying to get off of. More capacitance is not safer.
The guard against it is matching the correction to the actual load. That is the whole argument for an automatic bank on a varying load: it sheds stages when the load drops so it never pushes more kVAR than the load is taking. With fixed banks, size to the minimum steady load that is actually present when the caps are energized, and switch fixed banks with the load they correct so they cannot sit energized against a system that has gone quiet.
Measuring power factor and reactive load
You measure before you size and you measure after you install, and both readings go in the file. A power quality meter or a logging power meter at the service or the MCC gives you real power, reactive power, apparent power, and power factor over time, which is what you need because power factor is not a single number. It moves with the load through the day and across the week.
Log it, do not spot-read it. A single clamp-meter reading at 10 a.m. tells you the power factor at 10 a.m., and if you size a fixed bank to that one moment you will over-correct at night and under-correct at peak. A week of logged data shows the daily swing, the minimum load when over-correction is the risk, and the typical load you actually size to. The same meter logs the harmonic spectrum, the THD and the individual orders, which is the measurement that tells you whether you can install plain caps or need detuned reactors. Running the power factor log and the harmonic log on the same instrument at the same time is the efficient move, and the harmonics guide covers reading the spectrum in depth.
After commissioning, the meter is how you prove the bank did what it was sized to do. Same meter, same point, power factor before and after, with the stages cycling, so the record shows the correction working across the load range and not just at the moment someone happened to look. Tie the result back to the utility's revenue meter where you can, since that meter is the one writing the penalty.
Large-motor plants, data centers, and the wrong-tool trap
The right correction depends on what is dragging the power factor down, and that splits hard between a motor plant and an electronics-heavy facility. Get the diagnosis wrong and you install a bank that does not help.
A plant full of motors has a lagging displacement power factor from magnetizing current, and that is the textbook case for capacitor correction. Fixed caps at the big steady motors, an automatic bank at the MCC for the rest, detuned if there are drives in the mix. The reactive load is genuinely inductive, so capacitors cancel it cleanly.
A data center is the trap. Modern IT power supplies and UPS systems with power factor correction front ends often already run near unity displacement power factor, so there is little lagging kVAR for a capacitor to cancel. What they produce instead is harmonic distortion, a distortion power factor that capacitors cannot fix and that turns a plain cap bank into a resonance hazard. A facility whose power factor looks low because of harmonics needs harmonic mitigation, an active filter or detuned passive filter, not a capacitor bank. Measure the displacement power factor and the distortion separately before you decide. The harmonics guide covers the filtering side; the rule here is that capacitors answer a lagging displacement problem and nothing else.
Commissioning the bank
Commissioning a capacitor bank is where you prove the correction works across the real load and confirm the caps are not sitting on a resonance. Start before energizing: confirm the bank rating against the design kVAR, confirm the capacitor voltage rating matches the actual bus voltage, and verify the detuned reactors are present if the harmonic study called for them. Then check the discharge devices and the protection.
Log power factor before energizing so you have the baseline, then energize and watch the correction land. On an automatic bank, exercise the controller through its stages and confirm each step switches in and out cleanly without chattering, that the contactors pull in without excessive inrush, and that the controller holds the target across a load swing rather than hunting. Tune the controller's switching delay and dead band so it settles instead of cycling, which is the single most common commissioning fault on automatic banks.
Run the harmonic check with the bank energized. Energizing capacitance shifts the system's resonant frequency, so a bank that looked fine on paper can excite a harmonic once it is live. Measure voltage and current distortion before and after energizing, and watch for caps or reactors heating, fuses opening, or distortion climbing. If the numbers move the wrong way, you have a resonance to chase before the bank goes into service, not after it has cooked a step.
What the owner has to maintain
A capacitor bank is not install-and-forget hardware, and the owner needs to know that before the first cap fails. Capacitors dry out and degrade with age, heat, and over-voltage, and they lose capacitance as they go, so a bank slowly delivers less correction than it did the day it was commissioned. The power factor drifts back down and the penalty quietly returns, often before anyone connects the rising bill to the aging caps.
The failures to look for are physical and visible. A swollen or bulging capacitor can is a cap that has overheated and is on its way out, and a burst one has already failed. Open fuses on a step mean that step is offline and the correction is short. Discolored or hot reactors, a buzzing bank, and contactors that no longer pull in cleanly are all on the list. On an automatic bank the controller itself is a maintenance item, and a controller that has lost a stage or drifted out of tune holds the wrong power factor without throwing an obvious flag.
The maintenance is a periodic look, not a deep teardown: inspect for swelling and leaks, confirm each step is carrying current, check the discharge devices, log the power factor again and compare it to the commissioning baseline, and re-check the harmonic picture if the building's load has changed. The owner who skips it finds out when the penalty comes back or a can lets go.
What to document
The record is what lets the next person service the bank without re-engineering it, and what proves the correction was sized and verified rather than guessed. Capture the bank by location, because a plant often has several. For each one, record the kVAR rating, whether it is fixed or automatic, whether it carries detuned reactors and at what tuning, the capacitor voltage rating, the target power factor, the protection and discharge devices, and the before-and-after power factor from commissioning.
Write down the harmonic measurement that justified the reactor decision, or the absence of one, because that is the call the next engineer will second-guess when a drive gets added. And keep the utility tariff threshold the design was sized to beat, so when the penalty changes or returns, the file already says what the bank was supposed to deliver.
| Field to record | Why it matters |
|---|---|
| Location | Decides how much of the system the bank unloads |
| kVAR rating and steps | The correction supplied and the switching granularity |
| Fixed or automatic | Sets the over-correction risk and the maintenance |
| Detuned reactor and tuning | Whether the bank is safe in a harmonic environment |
| Capacitor voltage rating | kVAR scales with voltage squared, so it must match the bus |
| Target power factor | What the bank was sized to deliver |
| PF before and after | Proof the correction works across the load range |
| Utility threshold sized to | The penalty the design was meant to beat |
Common mistakes
- Installing capacitors in a building full of drives without measuring the harmonics first, then losing the caps to resonance.
- Over-correcting with a fixed bank sized for full load, so the power factor swings leading and the bus over-volts at light load.
- Reaching into a bank on lockout alone, trusting the wait time or the bleed resistor instead of verifying the caps are discharged with a meter.
- Putting the whole correction at the service entrance when the goal was to free internal transformer and feeder capacity at the loads.
- Tuning an automatic controller too tight, so it hunts and the stages chatter in and out, wearing the contactors.
- Sizing the bank to a single spot reading instead of a logged load profile, so it is wrong most of the day.
- Ignoring that capacitor kVAR scales with voltage squared, so a bank rated above the actual bus voltage under-corrects.
Field checklist
Want this checklist to run itself on every job — with photo proof and a signed record crews can hand the customer? That's FieldOS.
Standards and references
The utility tariff governs the economics. It sets the power factor threshold, the penalty formula, and therefore whether and how much correction pays, and it is the first document you pull. Nothing else in this guide overrides the number on that sheet, so confirm the threshold and the billing method directly with the utility before you size or quote a bank.
The NEC, NFPA 70, controls the installation. Capacitors live in Article 460, which covers the conductor sizing above rated capacitor current, the overcurrent protection, the disconnect, and the discharge-of-stored-energy requirement, including the residual-voltage limit and the time to reach it. Treat those as enforceable requirements, confirm the exact figures against the adopted edition and local amendments, and do not cite a section number you have not checked against that edition. NFPA 70E governs the safe work practice, including releasing stored energy as part of an electrically safe work condition.
For the capacitors and shunt-capacitor application themselves, IEEE 18 covers shunt power capacitors and IEEE 1036 covers their application. Where harmonics enter the picture, IEEE 519 gives the recommended distortion limits at the point of common coupling, which is the standard behind the detuned-reactor and filter decisions and the subject of the harmonics guide. The manufacturer's data controls the specific cap, reactor, contactor, and fuse selection. Cite the standard that actually controls the point, and let the tariff and the contract documents override any rule of thumb.
Units, terms, and conversions
Power factor work uses three power units that get confused on drawings and submittals, so name them straight. Real power is watts and kilowatts, the work. Reactive power is volt-amperes-reactive, VAR and kVAR, the magnetizing power. Apparent power is volt-amperes, VA and kVA, the total the system carries.
Power factor itself is a unitless ratio between 0 and 1, sometimes written as a percentage, and it carries a direction. A lagging power factor means the current lags the voltage, the normal inductive case from motors. A leading power factor means the current leads, the over-corrected capacitive case. Correction kVAR is the capacitor's reactive output at its rated voltage, and because that output scales with the square of the applied voltage, the same can delivers different kVAR on a different bus.
- Power factor (PF)
- Ratio of real power to apparent power, kW divided by kVA, a number from 0 to 1
- Lagging / leading
- Lagging is inductive (current behind voltage); leading is over-corrected, capacitive
- kVAR
- Kilovolt-amperes-reactive, the magnetizing power a capacitor supplies or a motor draws
- Detuned reactor
- Series reactor that shifts the bank's resonance below the lowest harmonic so it cannot ring
- Discharge means
- Bleed resistor or device that drains the capacitor's residual voltage after disconnection
- Displacement vs distortion PF
- Displacement is the phase shift caps correct; distortion is from harmonics and they cannot
FAQ
What is power factor?
Power factor is the ratio of real power in kilowatts to apparent power in kilovolt-amperes, the share of the current that does actual work. It runs from 0 to 1. A low power factor means a large reactive current that heats the wiring and loads the transformer while doing no work and earning the utility nothing.
Why do utilities penalize low power factor?
Utilities size their wires and transformers for your full apparent power in kVA but can only bill the real energy in kWh. Reactive current occupies that equipment and earns nothing, so the tariff adds a penalty, often a demand multiplier of kW divided by power factor, once you fall below a threshold around 0.90 to 0.95.
What does a capacitor bank do?
A capacitor bank supplies reactive power locally, so the magnetizing current the motors need circulates between the caps and the loads instead of being dragged from the utility. It cancels lagging displacement kVAR, which shrinks the apparent power and raises the power factor. It does not change the real load and does nothing for harmonics.
Fixed or automatic capacitor bank?
Use fixed capacitors on a steady load, ideally switched with a constant-load motor, since they are cheaper and have nothing to fail. Use an automatic bank where the load swings more than about 20 percent, because its controller adds and sheds stages to hold the target and avoid over-correcting into a leading power factor at light load.
How much capacitor kVAR do I need to correct power factor?
Multiply the real load in kW by the tangent of the present power factor angle minus the tangent of the target angle. For 300 kW going from 0.78 to 0.95, that is about 141 kVAR, so you install the nearest standard step. Size to the typical operating load, not the peak, to avoid over-correction.
Can power factor capacitors cause harmonic problems?
Yes. A capacitor and the transformer inductance form a resonant circuit, and if it tunes near a harmonic the building already makes, usually the 5th, it amplifies that current and blows the caps. In any building with drives, use detuned reactors that shift the resonance below the 5th. Measure the harmonics before installing plain capacitors.
What is a detuned reactor on a capacitor bank?
A detuned reactor is an inductor in series with each capacitor step, sized so the series resonance sits below the lowest significant harmonic, commonly around the 4.2th order. The step still corrects power factor at 60 Hz but looks inductive above its tuning point, so it cannot resonate with the system and amplify harmonic current.
Is a capacitor safe to touch after the power is off?
No. A capacitor holds a lethal charge after the disconnect opens. NEC Article 460 requires a discharge means, but the bleed resistor commonly fails open. Open, lock out, wait the discharge time, then verify dead with a tested meter and short and ground each phase before any contact. Never trust the wait alone.
What happens if I over-correct power factor?
Too much capacitance pushes the power factor leading and raises the bus voltage, especially when a fixed bank stays energized against a light load. The over-voltage stresses the caps and sensitive equipment, and some tariffs penalize a leading power factor too. Match the correction to the actual load and switch fixed banks with the load they serve.
People also ask
Codes cited in this guide
This guide is written and reviewed against the published standards below. Always confirm the current adopted edition with the authority having jurisdiction.