What Power Factor Actually Costs a Shop

Why this matters

Power factor gets sold to customers as an energy saving and it usually is not one. It gets ignored by techs as an engineering abstraction and it usually is not that either. What it actually is, on a service call, is the reason your clamp meter and your load calculation disagree, the reason a feeder is full at two thirds of the power you thought it was carrying, and the reason a correction proposal is either genuinely worth writing or a straight waste of the customer's money depending entirely on one line of their tariff.

Getting this wrong in either direction has a cost. Recommend correction to a customer whose meter bills energy only and you have sold them nothing. Ignore it on a service that is out of headroom and you will size a generator or a subpanel from a watts figure that was never the constraint.

The one paragraph of theory you need

Alternating current in an inductive load - any motor, transformer, coil or ballast - lags the voltage. Current that is out of phase with voltage still flows in the conductor, still heats it, and still has to fit through the breaker, but it does no work at the shaft. Real power (watts) is what does work. Apparent power (volt-amperes) is what the conductor and the device carry. Power factor is the ratio of the two: watts divided by volt-amperes. A resistance heater sits at essentially 1.0. An induction motor at full load commonly runs somewhere in the high 0.8s, and the same motor lightly loaded falls well below that, because its magnetizing current stays roughly constant while its working current shrinks with the shaft load.

That last point is the one most often misread in the field, so state it precisely: low power factor on a lightly loaded induction motor is normal behaviour of a healthy motor, not a fault symptom. Power factor is a property of the load's character and its loading, not a health reading.

What power factor does not cost you

The sharpest thing to know about power factor is the list of things it is not responsible for. Every item here is something a tech has confidently blamed on it.

It is not on a residential electric bill. A standard residential meter registers real energy over time. Reactive current passes through it and is not billed. Correcting power factor at a house changes the bill by zero. Any proposal that promises a homeowner savings from a power-factor device is selling something the meter cannot see.

It is not a measure of motor condition. A motor with a failing bearing draws more current and its power factor typically improves, because the shaft load went up. Diagnosing a motor by power factor points you the wrong way.

It is not read by a plain clamp meter. A current clamp reports amps. Power factor requires sampling voltage and current together and measuring the phase angle between them, which needs a power meter or a power-quality instrument. A number written on a ticket as "PF" that came off a current clamp came off nothing.

It is not fixed by a bigger conductor or a bigger breaker. Those accommodate the current. They do not change the load's character, and after the upsize the same reactive amps are still occupying the same fraction of the new capacity.

It is not the same thing as harmonic distortion, although both put amps in a wire that do no useful work at the load. See the section below, because the distinction changes the fix completely.

What it does cost you, and where

Two places, both real.

Conductor and device capacity. Every ampacity table, every breaker rating and every transformer nameplate is a current limit. Current is set by apparent power, so a low power factor consumes capacity in proportion to 1 divided by the power factor, whatever the watts are doing. Conductor heating goes with the square of current, so the effect on temperature is larger than the effect on the ammeter.

Billed demand, but only on some tariffs. Larger commercial accounts are frequently billed with a demand component alongside energy, and where that demand is metered in kVA, or where the tariff carries a power-factor adjustment clause, apparent power is directly in the bill. Where demand is metered in kW with no adjustment clause, it is not.

The gate: when correction returns something

Per service, evaluated over a full billing period, recommend correction only when either:

• the account's tariff bills demand in apparent power (kVA) or applies a power-factor adjustment clause, or
• a feeder, transformer or generator on that service is capacity-limited and you need headroom you would otherwise have to buy with copper.

If neither is true, correction returns nothing and should not be quoted. Read the tariff before you write the proposal, not after.

The worked case: same 0.78, two accounts, opposite answers

A compressor room, single-phase 230 V, three motors running together. Clamped feeder current: 92 A. Measured blended power factor with a power meter: 0.78. Feeder conductors and breaker are good for 100 A.

Apparent power is 230 x 92 = 21,160 VA. Real power is 21,160 x 0.78 = 16,505 W, so 16.5 kW is doing work while 21.2 kVA is occupying the feeder. A tech who converts the clamp reading straight to watts reports 21.2 kW and overstates the actual load by 1 divided by 0.78, which is 28.2 percent. On a generator sizing exercise that error goes both ways at once: the engine is sized by kW and the alternator by kVA, so quoting one number for both guarantees one of the two is wrong.

Now correct the blend to 0.95 with switched capacitors at the motors. Real power does not change, so current becomes 16,505 divided by (230 x 0.95) = 75.5 A.

Account A, energy-only tariff. Billed energy is unchanged, because the 16.5 kW is unchanged. The savings on the bill are zero and saying otherwise is a lie the meter will expose. What the customer actually gets: feeder utilisation falls from 92 of 100 A to 75.5 of 100 A, which is 92 percent down to 75.5 percent, and conductor loss falls by (75.5 / 92) squared = 0.674, a 32.6 percent reduction in the heat generated in that run. If the reason for the visit was "we want to add a fourth machine and the feeder is full," correction just released roughly 16.5 A of capacity without trenching a new feeder, and that is the whole case. If the reason for the visit was the bill, there is no case and the honest answer is to say so.

Account B, demand billed in kVA. The same physical change cuts apparent power from 21.2 kVA to 16.5 kVA for the identical work performed, a reduction of 1 minus (0.78 / 0.95) = 17.9 percent in the quantity the demand charge is calculated on. Energy is still unchanged, so the effect lands on the demand portion of the bill only, never on the whole bill. Quote it that way. A proposal that applies the 17.9 percent to the customer's total bill will be wrong by whatever fraction of their bill is energy, and they will find out in one billing cycle.

Same measurement, same correction, same 0.78 to 0.95: one account has a capacity argument and no bill argument, the other has both. The tariff decided it, not the meter.

Displacement, distortion, and why the meter may be lying

Total power factor has two independent parts and mixing them up leads to buying the wrong fix.

Displacement power factor is the phase lag from inductance, and it is what capacitors correct.

Distortion power factor comes from loads that draw current in a non-sinusoidal shape: switch-mode power supplies, LED drivers, electronic ballasts, variable frequency drives and battery chargers. Those loads can sit at nearly unity displacement power factor while still filling the conductor with harmonic current that does no work. Capacitors do not fix distortion, and installing them on a harmonic-rich service can make things worse by forming a resonant circuit with the supply inductance and amplifying a harmonic.

The measurement consequence lands on your own meter. An averaging-type clamp meter is calibrated to report the RMS value of a clean sine wave, and on a distorted waveform it reads wrong, typically low on a peaky current. A true-RMS instrument reports the heating value of whatever shape is actually there. On any service with substantial electronic load, a non-true-RMS clamp will under-report the current that is genuinely heating the conductor, so use true-RMS and know which one is in your bag.

If you do correct it, the hazards that ride along

Capacitors hold a lethal charge after the circuit is opened. This is an electrical hazard in a utilization installation, so isolate under 29 CFR 1910.333(b)(2) rather than 29 CFR 1910.147, which expressly sends electrical exposure of this kind to Subpart S. Establishing an electrically safe work condition includes releasing stored energy: allow the built-in discharge means the full time stated on the equipment label, then verify at the terminals with a CAT-rated meter and the live-dead-live sequence in NFPA 70E-2021, 120.5 before your hands go in. Never discharge a capacitor by bridging its terminals with a screwdriver - it welds the tip, throws molten metal at your face, and can rupture the case. NEC Article 460 covers the discharge means and the overcurrent and disconnect requirements for capacitors.

Switch the capacitor with the motor, not ahead of it. A capacitor left connected to a motor that is coasting down can self-excite the motor and produce voltages above line, which damages the motor and surprises the next person to open the box.

Do not add correction capacitors on the load side of a variable frequency drive, and check with the drive documentation before adding them on the line side. The drive's own input circuitry is not a load that wants capacitors bolted to it.

How to verify you got this right

Measure real power, apparent power and current at the same moment, before and after, with a true-RMS power meter at the same point on the feeder and with the same equipment running. Then confirm three things close: the watts should be effectively unchanged (if real power moved, something else changed and your before-and-after is not a fair comparison), the current should have fallen in the ratio of the old power factor to the new one, and the conductor loss reduction should equal the square of that current ratio. If the watts moved, stop and find out why before you report a result.

Then check the bill one cycle later against what you claimed, on the specific line item you claimed it on. A correction that shows up on the demand line and not the energy line is the correction working exactly as described.

References

• NEC Article 460 for capacitor discharge means, overcurrent protection and disconnecting means
• 29 CFR 1910.333(b)(2) for de-energizing electrical equipment, with 29 CFR 1910.147 reserved for mechanical and stored-energy isolation outside electric utilization installations
• NFPA 70E-2021, 120.5, for establishing an electrically safe work condition including stored-energy release
• Manufacturer documentation for variable frequency drives regarding capacitor application on input and output circuits
• See related: How a Motor Draws What It Draws; Single-Phase and Three-Phase in Practice