Power factor correction is normally presented as a straightforward transaction: install capacitance, raise power factor, stop paying the penalty. At a site with little electronic load that is broadly what happens.
At a site with substantial drive load, it is how a capacitor bank ends up switched off within a year, with blown fuses, hot capacitors and a supplier being asked why the equipment was defective.
It was not defective.
What the interaction is
Two things are true at once and they combine badly.
Capacitors present a low impedance to high frequencies. Harmonic currents flowing anywhere on the system will preferentially flow into a capacitor rather than back to the source, so a bank installed for power factor reasons becomes a sink for harmonics it was never intended to carry.
Capacitance and inductance resonate. The capacitance you have installed and the inductance of the supply — dominated by the service transformer — form a parallel resonant circuit at a particular frequency. At that frequency the circuit presents a high impedance to the harmonic source and circulates large currents internally between the capacitance and the inductance.
If the resonant frequency lands near a harmonic order that is actually present in meaningful quantity, those circulating currents can be several times the harmonic current the equipment is producing.
Where the resonant frequency lands
The resonant order depends on the ratio between the supply short-circuit capacity and the installed capacitor rating. Adding more capacitance lowers the resonant frequency; a stiffer supply raises it.
The practically important consequence is that the resonant point moves as capacitor steps switch in and out. A bank with four steps has four different resonant frequencies depending on how many are connected, and a system that behaves perfectly with two steps connected can resonate with three. Problems that appear intermittently, correlate with plant loading and defy diagnosis frequently have this cause.
The harmonics most often responsible are the fifth and seventh, which are characteristic of the six-pulse rectifier front end used by a great many variable frequency drives and by rectifier loads generally.
The symptoms
| Symptom | What is happening |
|---|---|
| Capacitor fuses operating repeatedly, with no fault found | Harmonic current well above the fundamental rating |
| Capacitors running hot, bulging cans, short service life | Continuous overcurrent at harmonic frequencies |
| Problems that appear only at certain plant loadings | Resonant point moving as steps switch |
| Transformer running hotter than its loading suggests | Additional losses from harmonic currents |
| Nuisance tripping of drives or protection | Voltage distortion at the point of common coupling |
| Correction appearing to work, then degrading | Failed capacitor elements reducing effective kVAR |
The last row is worth noting because it hides the problem. A bank losing elements to harmonic stress delivers progressively less correction, so the power factor penalty creeps back onto the bill months after the project was signed off. Anyone monitoring only the bill sees a saving that decays and no obvious cause.
Detuned banks
The standard solution is a reactor in series with each capacitor step, sized so the resonant frequency of the reactor-capacitor combination sits below the lowest significant harmonic present — commonly tuned below the fifth.
Two things follow. The combination cannot resonate with any harmonic above the tuning point, because it is inductive at those frequencies. And it presents a higher impedance to harmonic currents, so it stops acting as a sink.
Detuned equipment costs more than plain capacitors. It is the difference between an installation that works for its design life and one that is isolated after eighteen months, and at any site with meaningful drive load it should be the default specification rather than an option to be priced separately and then cut.
Measure before specifying
- Commission a power quality survey at the service entrance and at the main distribution panels.
- Record over a full operating cycle, including startup, full production and light load — the spectrum changes with plant loading.
- Capture the harmonic spectrum, not just a total distortion figure. Which orders are present determines the tuning point.
- Obtain the supply short-circuit capacity at the point of connection from the utility. It is needed to calculate where resonance would fall.
- Inventory the non-linear load: drives, rectifiers, uninterruptible supplies, induction heating, welding, LED drivers.
- Give the survey results to whoever specifies the correction equipment, and require the resonant calculation for every switching combination, not just for the full bank.
- Re-measure after installation. A survey before and none after is half a diagnosis.
The point about every switching combination is the one most often missed in a specification. A calculation performed for the fully connected bank tells you nothing about the intermediate states the bank will spend most of its life in.
The wider reason to care
Harmonics are a cost even without a capacitor bank in the picture. They add losses in transformers and conductors, they can force a transformer to be derated, and severe distortion affects other equipment on the same supply. A site that measures its harmonic environment for a correction project frequently finds other things worth acting on.
There is also a design-side option. Drives can be specified with front ends that produce less distortion — multi-pulse arrangements or active front ends — which reduces the problem at source rather than mitigating it downstream. Where drives are being purchased anyway, that choice is worth making deliberately, and it interacts with the power factor question in a way that is frequently misread: variable frequency drives and displacement power factor.
Where this sits in the decision
Harmonic assessment belongs before the capacitor bank is specified, not after it misbehaves. The order is: confirm the tariff actually charges for power factor, establish that a benefit exists, measure the harmonic environment, then size and specify.
Reversing the last two steps is what produces the failures described here. The billing side of the decision is in power factor penalties explained, and the sizing method, including when to insist on detuned equipment, in sizing a capacitor bank.
