Power Factor Penalties Explained

Some of the current flowing into your site does useful work. Some of it does not, and still has to be carried by every conductor, transformer and switch between the generator and your equipment.

That second component is reactive power, and utilities bill for it — not because they are recovering fuel, but because it occupies capacity that could otherwise serve somebody else.

The triangle

The relationship between the three quantities is geometric, and the geometry is the most useful thing to hold in mind.

Real power (kW) does the work: it turns the shaft, heats the oven, lights the space. It is what the energy charge prices.

Reactive power (kVAR) does no work at all. It flows back and forth each cycle, establishing the magnetic fields that inductive equipment needs in order to function. Motors, transformers and induction heaters all require it.

Apparent power (kVA) is the vector sum of the two: the actual current the system has to carry, multiplied by the voltage. It is what conductors, transformers and switchgear have to be sized for.

Power factor is kW divided by kVA — the fraction of the carried capacity that is doing something useful. At unity, all of it is. At 0.75, a quarter of the capacity being occupied is producing nothing.

Why the utility cares

A site drawing 800 kW at a power factor of 0.75 is drawing about 1,067 kVA. Every element between it and the generator — the service conductors, the distribution transformer, the feeder, the substation — has to be sized for 1,067 kVA even though only 800 kW of work is being done.

The same site corrected to 0.95 draws about 842 kVA for the same 800 kW of work. The utility recovers 225 kVA of capacity across its network without generating another watt.

That is the cost the penalty is recovering, and it is a real one. It is also why the mechanism is a capacity charge rather than an energy charge: nothing extra is being consumed, something extra is being occupied.

It is worth being clear about what does and does not flow. Reactive power is not lost, burned or wasted in the way the language around it sometimes suggests. It oscillates: energy moves into the magnetic field of a motor during part of each cycle and returns to the supply during the rest. Over a full cycle the net transfer is zero, which is precisely why it does no work and why the energy meter does not register it.

What is not zero is the current required to make that oscillation happen. That current flows through every conductor and every winding on the path, and it produces heat in all of them. So the utility incurs two costs from a poor power factor: capacity that cannot be sold to anyone else, and additional resistive losses across its network for output it cannot bill. Neither shows up in a kilowatt-hour figure, which is why a separate mechanism exists to recover them.

The same logic applies inside your fence, incidentally, and it is the reason correction is sometimes worth doing at a site whose tariff ignores power factor entirely. Your own cables and transformers carry that current too.

Three mechanisms, three different investment cases

This is the part that decides whether correction is worth money at your site, and it is entirely a question of tariff wording.

1. An explicit penalty. The tariff sets a threshold — commonly somewhere between 0.85 and 0.95 — and charges when measured power factor falls below it. The charge may be per kVAR of excess reactive demand, or a percentage adjustment to the bill. Correction to just above the threshold eliminates the charge; correction beyond it earns nothing further.

2. Demand billed in kVA. No penalty clause exists because none is needed. Demand is billed on apparent power, so poor power factor inflates the determinant directly and continuously. Correction reduces billed demand in proportion, and the saving is often the largest of the three cases.

3. A power factor adjustment to billed kW. The tariff multiplies measured kilowatts by a factor derived from power factor — for example, by the ratio of a target power factor to the actual one. The effect resembles kVA billing but the arithmetic differs, and the correction target that maximizes return differs with it.

4. Nothing at all. Many schedules, particularly smaller commercial ones, ignore power factor entirely. At those sites, correction has no billing benefit whatever. It may still be worth doing for internal capacity reasons — freeing up transformer headroom for an expansion is a real benefit — but it must not be justified on a bill saving that does not exist.

Which of the four applies to you is a clause in your rate schedule, and finding it is a ten-minute job that determines whether a capacitor project is worth six figures or nothing: how to read a utility tariff book.

What it is costing you

The arithmetic is straightforward once the mechanism is identified.

The contrast in that footnote is the practical point. A threshold penalty is a cliff: you pay it or you do not. kVA billing is a slope: every improvement earns. The same capacitor bank produces very different returns under the two.

Why power factor is poor in the first place

Understanding the cause matters, because some causes are cheaper to fix than a capacitor bank.

Lightly loaded motors are the classic. A motor's magnetizing current is roughly constant regardless of load, so a motor running at 30 percent of its rating draws nearly the same reactive power as at full load while doing a third of the work. A plant full of oversized motors has poor power factor structurally, and right-sizing them addresses the cause rather than the symptom.

Idling equipment. Machines energized but not producing draw magnetizing current and no useful output. Switching them off is free.

Transformers, particularly lightly loaded ones. Every transformer on site contributes magnetizing reactive power whether or not anything is connected downstream.

Induction heating, welding and older discharge lighting ballasts are all substantially inductive by nature.

The first two are worth investigating before capital is committed. Correction equipment fixes the measurement; switching off an idling machine fixes the cause and saves energy as well.

There is a diagnostic worth running before anything is specified. Compare power factor at full production with power factor overnight or at the weekend. A site whose power factor is acceptable when busy and poor when idle has a fixed reactive load — transformers, standing equipment, machines left energized — and that pattern points at operational fixes and at a fixed correction component rather than at a large switched bank. A site whose power factor is poor at full production has genuinely inductive process load, and correction is the appropriate answer.

Correcting it

Capacitors supply reactive power locally so it does not have to travel from the generator. Where they sit determines what they relieve.

At the service entrance relieves the utility's network and fixes the bill. It does nothing for the cables and switchgear inside your site, which continue to carry the same current.

At a distribution panel relieves everything upstream of that panel.

At the individual motor relieves the entire path and is the most effective technically, at the cost of many small units to install and maintain.

Automatic switched banks vary the correction with the load. Necessary where load varies substantially, because a fixed bank sized for full load will overcorrect at light load, and overcorrection is a real problem rather than a theoretical one.

Sizing the bank is where projects are decided, and the calculation is short: sizing a capacitor bank.

The risks worth knowing before you order

Capacitor banks are not inert additions, and two failure modes are common enough to plan around.

Harmonic resonance. Capacitors and the supply inductance form a resonant circuit. If the resonant frequency lands near a harmonic that the site's own equipment is producing — and variable frequency drives, rectifiers and switched-mode supplies all produce them — the result can be amplified currents, overheating capacitors and blown fuses. At sites with substantial drive load this is the first thing to check, not an afterthought: harmonics, resonance and why capacitors sometimes make it worse.

Overcorrection. A fixed bank sized for peak load overcorrects when the plant is idle, producing a leading power factor, raising voltage, and in some tariffs incurring a penalty in the other direction. Automatic switching solves it; a fixed bank on a variable load creates it.

What correction does not do

It does not reduce kilowatts. The real power doing the work is unchanged — that is the entire point of the correction, and it is why a site whose demand is billed in kW sees no demand reduction at all from a capacitor bank.

It does not meaningfully reduce billed kilowatt-hours. Lower current reduces resistive losses in your own conductors, which is a real but small effect, and it does nothing to the energy the equipment is actually converting.

Presenting a correction project as a demand or energy saving at a kW-billed site is a business case that will not survive the first bill, and it is one of the entries in six efficiency measures that do not cut your demand charge.

The order of work

1. Find the power factor clause in your tariff and identify which of the four cases applies.
2. If it is the fourth — no clause — stop, unless you need transformer headroom for an expansion.
3. Establish your actual power factor from the bill or from metering, at the times the tariff measures it.
4. Look for causes that can be removed rather than corrected: idling equipment, grossly oversized motors.
5. Assess harmonic content before specifying capacitors, particularly if drives are a significant share of load.
6. Size the bank to land inside a target band, not at maximum correction: sizing a capacitor bank.
7. Check whether a different schedule treats power factor more favorably, since this is one of the provisions that differs most between schedules: how to choose a rate schedule.

Step three deserves one caution. Tariffs differ in when power factor is measured — at the time of peak demand, as a monthly average, or continuously — and a site that is poor at the peak and good on average will be assessed very differently under the two. It is one more reason the clause has to be read rather than summarized, and the underlying units are worth being clear about first: kVA billing versus kW billing.