Power Factor Correction Calculator
Size a capacitor bank in kVAR to lift a plant's power factor, and see the demand reduction that pays for it — for facility and plant staff working from a utility bill rather than a wiring diagram.
Last checked against the code
Power Factor Correction Calculator
Capacitor bank
Demand
- Apparent power nowwhat the utility meters
- —
- After correction
- —
- Reduction
- —
- Reactive powerwhat the bank takes over
- ——
How this was derived
- Phase angles
- —
- Correction factor
- —
- Required capacitance
- —
Line current
- Before
- —
- After
- —
- Freed up
- —
How this is calculated
Three powers, and only one of them does anything. Real power (kW) turns shafts and makes heat. Reactive power (kVAR) sloshes back and forth between the supply and the magnetic fields of every motor and transformer on the site, doing no work at all — and still occupying capacity in every conductor, transformer and switchgear it passes through. Apparent power (kVA) is the two combined as a vector, and it is what the utility's equipment has to be built for. Power factor is simply kW ÷ kVA.
Capacitors do not remove the reactive power; they relocate it. The motors still need their magnetic fields. A capacitor bank supplies that reactive current locally, on site, so it stops being dragged back and forth across the utility's network. The kW is untouched, the kVAR seen at the meter falls, and kVA falls with it.
The arithmetic is one triangle. Power factor is the cosine of the angle between real and apparent power, so the reactive component is kW × tan φ. Move from one power factor to another and the capacitance you need is the difference between the two reactive components:
kVAR = kW × (tan φ₁ − tan φ₂), where φ₁ = arccos(existing PF) and φ₂ = arccos(target PF).
That bracketed term is the whole of the classic printed correction-factor table — existing power factor down the side, target across the top, and nothing in any cell that this one expression does not produce.
Worked example
One full calculation with real numbers, so you can follow along and check the tool by hand.
A plant drawing 100 kW at0.75 power factor, with a tariff that penalises anything below 0.95.
The two angles. φ₁ = arccos(0.75) = 41.41°, whose tangent is 0.8819. φ₂ = arccos(0.95) = 18.19°, tangent 0.3287.
The bank. 100 × (0.8819 − 0.3287) = 100 × 0.5532 = 55.32 kVAR. That is what gets ordered — in practice rounded to whatever step size the manufacturer builds, typically 50 or 60 kVAR here.
The demand, which is the actual point. Before correction the site draws 100 ÷ 0.75 = 133.33 kVA. After, 100 ÷ 0.95 = 105.26 kVA. The reduction is 28.07 kVA, or 21.1% of metered demand — every month, for the life of the bank.
And what it frees up. At 480 V three-phase, line current falls from 160.38 A to 126.61 A — 33.77 A of feeder and transformer capacity handed back without touching a single load. On a site running close to a transformer limit, that is frequently worth more than the tariff saving.
Note the one figure that did not move: the real power is still 100 kW. The plant does exactly as much work as it did before, and consumes exactly as many kilowatt-hours. Correction is a change to how that work is delivered, not to how much of it there is.
Visual comparison
| Existing power factor | Capacitance neededkVAR | Multiplier |
|---|---|---|
| 0.60 | 100.5 | × 1.0046 |
| 0.65 | 84 | × 0.8404 |
| 0.70 | 69.2 | × 0.6915 |
| 0.75 | 55.3 | × 0.5532 |
| 0.80 | 42.1 | × 0.4213 |
| 0.85 | 29.1 | × 0.2911 |
| 0.90 | 15.6 | × 0.1556 |
Source: Derived from kW × (tan φ₁ − tan φ₂), the same expression the calculator uses
Reference tables
The classic correction-factor table: find your existing power factor down the side, your target across the top, and multiply the cell by your real power in kW to get the bank size in kVAR. It is the standard way this calculation was done before calculators, and it is still the fastest way to sanity-check one.
This copy is generated rather than transcribed — every cell is the same tan φ₁ − tan φ₂ expression the tool above runs. That matters more here than it might elsewhere: published versions round to two or three decimals and disagree with each other in the last digit, and a table that cannot diverge from the calculator beside it is worth more than one that matches a particular printed source. Cells where the target is not above the existing power factor are blank, because there is nothing to correct.
| Existing PF | Target 0.85 | Target 0.90 | Target 0.95 | Target 1.00 |
|---|---|---|---|---|
| 0.50 | 1.112 | 1.248 | 1.403 | 1.732 |
| 0.55 | 0.899 | 1.034 | 1.190 | 1.518 |
| 0.60 | 0.714 | 0.849 | 1.005 | 1.333 |
| 0.65 | 0.549 | 0.685 | 0.840 | 1.169 |
| 0.70 | 0.401 | 0.536 | 0.692 | 1.020 |
| 0.75 | 0.262 | 0.398 | 0.553 | 0.882 |
| 0.80 | 0.130 | 0.266 | 0.421 | 0.750 |
| 0.85 | — | 0.135 | 0.291 | 0.620 |
| 0.90 | — | — | 0.156 | 0.484 |
| 0.95 | — | — | — | 0.329 |
Notes and exceptions
Harmonics change the problem entirely. Everything on this page is DISPLACEMENT power factor — the phase shift between voltage and current on a linear load. Where a site runs significant non-linear load, true power factor is worse than the displacement figure and plain capacitors can make things sharply worse: a capacitor bank and the supply impedance form a resonant circuit, and if that resonance lands near a harmonic the system is producing, currents can multiply rather than cancel. Sites with substantial drive or rectifier load need a harmonic survey and detuned capacitors, not a bank sized from this calculation.
Fixed banks and automatic banks solve different problems. A fixed bank is cheap and correct at one load. An automatic bank switches stages in and out to follow the load, which is what you need where demand swings widely across a shift or a season — and is what keeps a site from going leading overnight.
Where the capacitors sit changes what they fix. At the main switchboard, correction fixes the billing but leaves every downstream conductor carrying the same reactive current. At individual motors it unloads the feeders too. Motor-mounted capacitors have a hard constraint: they must not exceed the motor's no-load magnetising current, or the motor can self-excite as it coasts down after being switched off, generating voltage back into a circuit somebody believes is dead.
Capacitors have their own code requirements. Article 460 covers their overcurrent protection, disconnecting means and the discharge requirement that drains stored charge after de-energising. A bank is not simply a load hung on a spare breaker, and none of that is modelled here.
Common mistakes
Correcting all the way to unity
A fixed bank sized for 1.0 at full load pushes the system leading the moment the load drops, and plenty of tariffs penalise leading power factor exactly as they penalise lagging. It can also raise voltage on a lightly loaded feeder. 0.95 captures most of the saving and stays safely on the right side.
Expecting the energy bill to fall
Correction does not reduce kWh, because it does not reduce the work being done. What it reduces is metered demand in kVA and any power factor penalty attached to it. A site billed purely on energy with no demand component may see no saving at all — check the tariff before pricing a bank.
Entering the power factor as a percentage
It is a ratio between 0 and 1. Typing 85 rather than 0.85 is easy, both look plausible in the field, and the result is wrong by two orders of magnitude. This calculator rejects the percentage form rather than computing from it.
Sizing from a single instantaneous reading
Power factor moves through the day with the load — a plant at 0.72 mid-shift may sit at 0.9 overnight with only the lightly loaded transformer running. A bank sized from the worst moment overcorrects for most of the week. Size from a logged period of typical operation.
Fitting plain capacitors on a site full of drives
Capacitors correct displacement power factor. Where harmonic distortion is significant — variable-speed drives, rectifiers, large LED installations — true power factor is worse than the displacement figure, and untuned capacitors can resonate with the supply impedance and make things sharply worse.
This tool provides planning estimates. Always verify final values against your local code and a licensed electrician.
Frequently asked questions
What is the difference between kW, kVAR and kVA?
kW is real power — the part doing work. kVAR is reactive power, which sloshes back and forth between the supply and the magnetic fields of motors and transformers, doing nothing useful but still occupying every conductor on the way. kVA is the two combined as a vector: kVA² = kW² + kVAR². Power factor is kW ÷ kVA, so it is simply the fraction of what you are drawing that is doing something. Correction supplies the kVAR locally from capacitors, which shrinks kVA and leaves kW untouched.
How many kVAR do I need for 100 kW at 0.75 power factor?
About 55 kVAR to reach 0.95. The multiplier is tan(arccos 0.75) − tan(arccos 0.95) = 0.882 − 0.329 = 0.553, and 100 × 0.553 is 55.3 kVAR. That correction takes metered demand from 133.3 kVA down to 105.3 kVA — a 28 kVA reduction, which on most demand tariffs is the number that pays for the bank.
Will power factor correction lower my electricity bill?
Only the parts of it tied to demand or to power factor itself. Correction does not change kWh consumed, so an energy-only charge is unaffected. What it does change is billed demand in kVA, and any explicit power factor penalty or kVA-adjusted kW charge. Read the tariff first: a site with no demand component and no PF clause has little to gain financially, whatever its power factor.
Can power factor be corrected too far?
Yes, and it is a real risk with fixed banks. Correcting to unity at full load means the site goes leading whenever the load drops — overnight, at weekends, during a shutdown — and many tariffs penalise leading power factor as readily as lagging. A leading condition can also push voltage up on a lightly loaded feeder. Targeting 0.95 rather than 1.0, or using an automatic bank that switches stages with the load, avoids it.
Does correction reduce the current in my cables?
Yes, upstream of the capacitors, and by the same percentage as the kVA reduction. That is often the second reason to do it: freeing capacity in a transformer or feeder that is running close to its limit can defer an upgrade. Downstream of the bank nothing changes — the motors still draw what they drew, which is precisely why individual motor correction and bulk correction at the switchboard do different jobs.
Where should the capacitors go?
It depends what you are buying. A bank at the main switchboard corrects what the utility meters and is the cheapest way to fix a billing problem, but everything downstream still carries the reactive current. Correcting at individual motors unloads the feeders too, which is what you want if the goal is freeing up capacity rather than avoiding a penalty. Motor-mounted capacitors must not exceed the motor’s no-load magnetising current, or the motor can self-excite when it is switched off.