Power factor correction is the practice of adding capacitance to a plant's electrical system to offset the reactive power drawn by motors and other inductive loads, so the plant pulls closer to its real, working power from the utility instead of paying for power it does not use productively. Get it wrong near variable frequency drives, though, and the fix can create a harmonic resonance problem worse than the one you started with.
Power factor (PF) is the ratio of real power (kW), the power that does actual work, to apparent power (kVA), the total power the utility has to generate and deliver to supply that load. PF = kW / kVA. A PF of 1.0 (unity) means all delivered power is doing useful work. A PF of 0.75 means the utility is supplying roughly a third more current than the load actually converts into work, and that extra current has to travel through the same wires, transformers, and switchgear as the useful current.
The gap between kW and kVA is filled by reactive power, measured in kVAR, which oscillates back and forth between the source and the load without being consumed. It is not "wasted" in the sense of being converted to heat at the load, but it does occupy capacity everywhere upstream of that load.
Induction motors, the workhorses of most plants, are inherently inductive: they need a magnetizing current to establish the rotating magnetic field in the stator before they can produce any torque. That magnetizing current lags voltage by close to 90 degrees and contributes nothing to real work, but it still shows up as current the utility has to supply.
The magnitude of the effect depends heavily on load:
This is why plants with lots of intermittent or oversized motor loads, common in facilities that never right-sized equipment to actual demand, tend to see the worst site-wide power factor.
Low power factor costs a plant money in two distinct ways.
First, utility penalties. Many utilities meter kVAR or calculate a billing power factor and add a surcharge, or bill on kVA demand instead of kW demand, once a site's power factor falls below a contractual threshold. Thresholds vary by utility and tariff, commonly somewhere in the 0.85 to 0.95 range, so the exact number and penalty structure has to be read off your own utility's rate schedule rather than assumed. Utilities do this because low-PF customers force them to size generation, transmission, and distribution equipment for kVA that never turns into billable kWh.
Second, wasted internal capacity. The reactive current still flows through your own transformers, feeders, breakers, and cables. That current causes real I²R heating losses in that equipment and consumes ampacity that could otherwise serve additional load. A plant running at a poor power factor can find itself needing a transformer or service upgrade to add a new production line, when the real problem is that a large share of existing capacity is tied up circulating reactive power rather than delivering usable kW. Correcting power factor frees that headroom without touching the incoming service.
Because inductive loads draw lagging reactive current, the fix is to supply an equal and opposite leading reactive current locally, which capacitors do naturally. A capacitor bank sized correctly offsets the motor's magnetizing kVAR, so the current the utility sees looks much closer to the current actually converted into work. The required capacitor size in kVAR is the plant's real load in kW multiplied by the difference between the tangent of the present power factor angle and the tangent of the target power factor angle.
In practice, correction is applied in a few common architectures:
Over-correcting is a real risk: pushing power factor to a leading value can cause voltage rise and its own penalties, so banks are normally staged and targeted to a PF just under unity, not exactly at 1.0.
| Condition | Typical power factor (lagging) | Practical implication |
|---|---|---|
| Induction motor at full load | ~0.85 to 0.90 | Modest correction usually needed |
| Induction motor at light/partial load | ~0.2 to 0.5 | Large reactive burden relative to work done |
| Whole plant with mixed, well-loaded motors | ~0.80 to 0.90 (uncorrected) | Often near or below utility penalty threshold |
| Corrected plant target | ~0.95 to 0.98 | Below unity to avoid leading-PF penalties |
Variable frequency drives are everywhere in modern plants because they save energy on pumps, fans, and conveyors by matching motor speed to load, a topic covered in more depth alongside VFD carrier frequency selection. But VFDs are non-linear loads: their rectifier front end draws current in pulses, not a smooth sine wave, injecting harmonic currents (5th, 7th, 11th, 13th, and beyond) back onto the plant's electrical system. This is part of the same power quality picture discussed in power quality in manufacturing.
Here is the trap: a capacitor bank's impedance falls as frequency rises, while the plant's source impedance (transformers, feeders) rises with frequency. At some frequency, the two can become equal and opposite, a condition called parallel resonance. If that resonant frequency lands near a harmonic the VFDs are already injecting, typically the 5th or 7th, the capacitor bank and the system inductance amplify that harmonic current instead of absorbing it. The result can be highly distorted voltage, overheated or failed capacitors, nuisance fuse operation, and stress on any equipment sharing that bus.
IEEE Std 519 sets the reference limits utilities and engineers design to at the point of common coupling, covering both voltage distortion (THDv) and current distortion (THDi); the exact percentage limits depend on system voltage class and the ratio of available short-circuit current to load current, so they should be pulled from the standard itself or a qualified power quality study rather than approximated. The practical takeaway for any plant adding capacitors where VFDs are present:
This same electrical stress shows up mechanically too. Motors running on distorted supply or unbalanced voltage run hotter and wear insulation faster, which is why insulation resistance testing and reviewing motor insulation classes against actual operating conditions are worth revisiting whenever a plant makes a significant change to its electrical loads.
Power factor correction is an electrical-side fix, but it sits inside a bigger reliability and energy picture: right-sizing motors, fixing the loose belts and worn bearings that make drives run inefficiently, and catching electrical faults before they become downtime. Fabrico reads machine condition and OEE straight from the line with computer vision that catches what sensors alone miss, and auto-routes a work order the moment a loss is detected, all on EU-built infrastructure with EU data residency and ISO 27001, ISO 20000-1, and ISO 9001 certification. Book a Fabrico demo to see how that closes the loop between an electrical anomaly and a maintenance action.
Most utilities want to see 0.90 or higher, and many plants target 0.95 to 0.98 to build in margin against penalties. Exactly unity (1.0) is usually avoided because pushing PF to a leading value can create its own problems, including voltage rise.
It reduces kVA demand and the associated I²R losses in your own wiring and transformers, freeing capacity, but it does not reduce real energy consumption (kWh) at the load, since reactive power is not converted to work in the first place. The savings come mainly from avoided penalties and freed capacity, not from the motors themselves using less real power.
Any VFD with a standard diode or SCR rectifier front end injects some harmonic current. The risk to capacitors is not the VFD alone, it is whether the capacitor bank's resonant frequency happens to land near a harmonic the VFDs are producing on that same bus. A harmonic study identifies that risk before hardware goes in.
Both are used. Central correction at the service entrance is simpler to manage and cheaper per kVAR, but individual motor-level correction reduces reactive current flow through the plant's internal feeders too, which can matter if internal cable or transformer capacity is tight. The right choice depends on where the capacity constraint actually is.
Zaloguj
Blog