Failure
Why capacitor banks overheat and fail
The capacitor is almost never faulty. It is being asked to carry current at frequencies it was never designed for, and the mechanism is predictable — even calculable before it happens.
Capacitor banks overheat because capacitive reactance falls as frequency rises, so a capacitor absorbs harmonic current in addition to its rated fundamental current. When the bank forms a parallel resonance with the supply inductance at or near the 5th or 7th harmonic, that current can reach several times the rated value.
A detuned reactor shifts the resonant frequency below the lowest harmonic present, which is why almost every industrial capacitor bank should be specified with one.
The mechanism, in one paragraph
A capacitor's reactance is Xc = 1 / (2πfC). Double the frequency and the reactance halves. At 250 Hz — the 5th harmonic on a 50 Hz system — the capacitor presents one fifth of the impedance it presents at the fundamental, so the harmonic current flowing into it is five times larger than the same voltage would drive at 50 Hz.
The capacitor is not being overloaded by the utility voltage. It is being overloaded by harmonic current that it actively attracts. This is the single most common reason for capacitor failure in plants that also run VFDs, rectifiers or switch-mode supplies.
Resonance is what turns a problem into a failure
Every network has inductance — transformer leakage, cable, and the source. The capacitor adds capacitance in parallel. Together they form a resonant circuit with a natural frequency determined by the ratio of short-circuit capacity to capacitor rating.
# S_sc = short-circuit capacity at the capacitor (MVA)
# Q_c = capacitor rating (MVAr)
# 1000 kVA transformer, 6% Z, 300 kVAR bank
Ssc = 1.0 / 0.06 = 16.7 MVA
hres = √(16.7 / 0.3) = 7.5 ← sits between the 5th and 7th
A resonant order of 7.5 is dangerously close to the 7th harmonic. At resonance the impedance seen by that harmonic current is very high, so a modest harmonic source produces a large harmonic voltage, which in turn drives a large current through the capacitor. Currents of three to ten times rated are common at this point.
Note what the formula implies: the resonance order depends on the capacitor rating you choose. A larger bank lowers the resonant frequency; a smaller one raises it. This is why sizing a bank without checking the resulting resonance order can create a failure rather than prevent one.
What you see before it fails
| Symptom | What it suggests |
|---|---|
| Cases swollen or bulging | Internal overpressure from sustained overcurrent — the end stage |
| Fuses blowing repeatedly | Harmonic current above fuse rating, or resonance |
| Audible humming, louder than usual | Harmonic current causing magnetostrictive noise in the reactor |
| Bank hot to touch, cabinet fans running constantly | I²R losses from harmonic current; check capacitor case temperature |
| Contactors pitting or welding | High inrush plus harmonic current at switching |
| Power factor reading correct but penalty persists | True versus displacement power factor — distortion is not being corrected |
| Failure within 12–24 months of installation | Almost always resonance rather than a manufacturing defect |
The last row is the one worth remembering. A capacitor that fails in its first two years, in a plant that runs drives, is a resonance failure until proven otherwise.
What detuning actually does
A detuning reactor is connected in series with the capacitor. Together they form a series resonant circuit tuned to a frequency below the lowest harmonic present, so that at harmonic frequencies the branch looks inductive rather than capacitive and the harmonic cannot be amplified.
| Detuning factor | Series resonance | Protects against | Typical use |
|---|---|---|---|
| 5.67% | 210 Hz | 5th and above | Light detuning, marginal networks |
| 7% | 189 Hz | 5th and above | Standard industrial choice |
| 14% | 134 Hz | 3rd and above | Networks with high single-phase content |
Note the voltage consequence. A detuned capacitor carries a higher voltage than the bus voltage, because the reactor and capacitor form a voltage divider at the fundamental. A 400 V network with a 7% reactor requires capacitors rated at approximately 440–480 V. Fitting 400 V capacitors behind a 7% reactor produces a second, entirely avoidable failure mode.
When capacitors are the wrong answer
Detuned capacitors are cost-effective and reliable up to a point. Beyond it, they stop being the right tool.
- Voltage THD above roughly 8%. The capacitor is carrying substantial harmonic current even when detuned. An active filter or a hybrid arrangement is more appropriate.
- Highly variable load. A stepped bank cannot track a load that swings between 15% and 100% without going leading at the low end. A static var generator provides continuous output with no switching steps.
- Multiple resonances. If the network has more than one resonant peak — common where several banks exist at different boards — a single tuned branch cannot protect all of them.
- Power factor correction is needed but the penalty is caused by distortion. Capacitors cannot correct distortion. If the tariff meters true power factor, a capacitor bank will not clear the penalty.
The economics only work if the root cause is identified first. A detuned bank costs 20–30% more than a plain one, and both cost far less than replacing capacitors every eighteen months.
Frequently asked
Why do power capacitors fail so often in industrial plants?
Because a capacitor is a low-impedance path for exactly the frequencies harmonic currents occupy. Capacitive reactance falls as frequency rises, so the capacitor absorbs harmonic current and heats up. A capacitor rated for 400 V and 50 Hz may carry several times its rated current once a 5th harmonic resonance is established.
Does a detuned reactor solve the problem completely?
It solves the resonance risk at the design frequency, which is what causes most catastrophic failures. A 7% reactor shifts the resonant frequency to around 189 Hz, below the 5th harmonic, so the capacitor never sees amplified 5th harmonic current. It does not remove the harmonic current itself, and it does not protect against higher-order amplification in a network with a second resonance.
What detuning factor should I choose, 7% or 14%?
7% (reactor impedance relative to capacitor impedance) protects against the 5th harmonic and above, and is correct for most industrial networks where the 5th dominates. 14% protects against the 3rd harmonic as well, and is needed where single-phase loads, switch-mode supplies or LED drivers produce significant triplen content. Choosing 7% on a network with high 3rd harmonic leaves that order unprotected.
Can I just replace the failed capacitor with a same-rated unit?
Only if the root cause has been ruled out. Replacing a capacitor that failed through harmonic overcurrent with an identical unit predictably produces a second failure, often within months, which is then blamed on the replacement supplier. A power quality recording taken before the replacement costs far less than the second failure.
Capacitors failing on a schedule?
Send a power quality recording covering a full production cycle, plus the transformer rating and existing bank size. We will calculate the resonance order and tell you whether detuning is enough or the bank needs replacing with a different approach.