Why Capacitor Fuses Blow: Real Failures vs Nuisance Operations

Two kinds of fuse operation

A capacitor fuse has two jobs that pull against each other. A capacitor bank maker's fuse selection note states the balance: the fuse must be small enough, or sensitive enough, to prevent case rupture, yet large enough to prevent spurious or false operation, taking account of the inrush and outrush currents of bank energization. When a fuse operates, it is either the first job working as intended or the second job failing.

A real operation leaves a failed unit or a fault behind the fuse. A nuisance operation leaves a healthy unit with a blown fuse in front of it. Both look the same from the ground: an indicator out, a leader ejected, or an unbalance alarm. The difference only shows when the unit is tested and the circumstances are checked.

Real failures: what the fuse is supposed to catch

A relay maker's paper on bank fundamentals describes the usual path to a real operation in an externally fused unit. An element failure welds its foils and shorts its section; the remaining sections carry more voltage; if a second element fails the process repeats, and successive failures within the same unit make the fuse operate. With all-film units, one element failure alone does not draw enough extra current to melt the external fuse, so a unit typically fails in stages before the fuse acts.

Other real faults sit outside the unit: bushing failures or faulty connections at the unit, arc-over within the rack, and rack-to-rack flashovers where connections are not insulated. One large utility's field experience, published with a relay maker, is that animals climbing on the bank, contamination and human error now cause most bank trips, more than arcing inside units. A university dissertation on unbalance protection adds that capacitor units most often fail from continuous overvoltage stress.

Nuisance operations: inrush, outrush and switching

Every time a bank is energized, a high-frequency charging current flows through its fuses. One fuse maker's literature calls the resulting nuisance operations sneakouts. Published magnitudes vary with the arrangement:

• A single isolated bank: generally less than 5 kA and 1 kHz, per one manufacturer's application guide.
• Back-to-back banks on the same bus: on the order of tens of kiloamperes and tens of kilohertz, per the same guide; a published 115 kV case study measured 28.1 kA peak at 5.6 kHz.
• Low-voltage banks: one contactor maker gives up to 30 times nominal current for fixed banks and up to 100 times the step current for multi-step banks.
• Low- and medium-voltage power factor correction: one fuse maker states inrush lasts less than 1/4 cycle and is typically less than 25 times the capacitor's current rating.
• Outrush: discharge into a nearby fault, which a switching tutorial calls severe for a bus fault because the inductance is on the order of tens of microhenries.

Lightning, overvoltage and resonance

Lightning drives a large charging current into a capacitor. The RUS bulletin's rule for the fuse is direct: if lightning causes excessive fuse blowing but there is no capacitor damage, fuses should be replaced with the next larger size. The same bulletin lists switching operations, arcing grounds, accidental contact with higher voltages, disturbances from other arresters, and resonance caused by motors while starting as sources of transient overvoltage.

Harmonic resonance is a different case, because a bigger fuse does not fix it. One maker's troubleshooting table says frequent unexplained fuse blowing may be a sign of a parallel resonance condition and to perform harmonic analysis. A plant engineer guide gives the estimate h = sqrt(kVAsys / kvar). Its worked example: a 1,500 kVA, 5.5% transformer fed from a 48,000 kVA utility short circuit gives a system short-circuit capacity of 17,391 kVA (1,500 / 0.055 = 27,273 kVA in series with 48,000 kVA). A 350 kvar bank then gives h = sqrt(17,391 / 350) = 7.0, right on the 7th harmonic. The RUS bulletin adds that the 9th harmonic, at 540 Hz, can have the greatest effect because transformers generate it and capacitors offer low impedance there.

A switching tutorial describes one more mechanism: voltage magnification between a switched transmission bank and distribution banks, with a 1.76 per unit example, which could lead to failure of the downstream bank.

The fuse itself: hot spots, overload, damage and undersizing

Some operations trace back to the fuse or how it was applied.

• Loose connections: one fuse maker explains that a loose button-head, cap or rod connection creates a hot spot that makes a link behave as though it is heavily preloaded, so a transient that would not normally affect it blows it. It calls this most common on capacitor bank fuses because their continuous currents are higher.
• Overloaded links: a municipal utility's protection standard says that above 150% of the minimum melt threshold, a link starts melting, either blowing eventually or weakening and operating unpredictably later.
• Element damage in current-limiting fuses: one maker's I2t test note says a nick halfway from the element edge to a hole cuts minimum melt I2t by 44% while changing resistance by only 0.75%, so a resistance check will not find it. Elements can be damaged in service by miscoordination, high current surges or clearing multiple phase faults.
• Aging: a utility and relay maker paper notes that pollution, corrosion and fluctuating climate reduce the reliability of external fuses, which must be checked and replaced periodically.
• Undersizing: the RUS bulletin says sizing must allow for the capacitor's 135% capability, overvoltage, harmonics and actual kvar being on the high side of manufacturing tolerance.

Worked example: why nameplate current is not enough

Take a 200 kvar, 7,200 V unit. Nameplate current is 200 / 7.2 = 27.8 A. Capacitor current is V / Xc, so it rises in proportion to voltage: at 110% voltage the unit draws 1.1 x 27.8 = 30.6 A, and its kvar rises by 1.1^2 = 1.21 to 242 kvar. IEEE 18-2002, as stated in IEEE's 2005 interpretation, expects units to run continuously up to 135% of nominal rms current, which here is 1.35 x 27.8 = 37.5 A, with harmonics included.

A fuse rated near 28 A on this unit would spend its life near or above its rating, and every switching transient would push its element toward the melting point. That is why published guidance applies a margin: 1.35 at 25 °C in several medium-voltage application methods, 1.25 in one utility guideline, and 1.65 to 2.5 in some low-voltage and plant guides. Each source states its own basis; none is a standard rule. The sizing article compares them.

What the indicators and the unbalance relay tell you

In an externally fused bank the first sign of an operation is usually an alarm. A regional transmission organization's capacitor guideline, PJM's, says a single blown fuse should cause the unbalance relay to operate and initiate an alarm, and that the bank should be tripped when more than one fuse blows in the same series group and the overvoltage on the remaining units exceeds 10%. A relay maker's paper adds a useful clue: when a unit fails open, as it does once its fuse clears, the neutral voltage signal is out of phase with the faulted phase voltage, and when a unit fails short it is in phase. That points the crew at the right phase before anyone climbs.

On the bank itself, the fuse's own indication finds the unit. Many capacitor fuses eject a leader or push out an indicator button when they operate, and DENCO's catalog lists its C-rated capacitor fuses with or without a blown-fuse indicator. Infrared survey is another route: a test-equipment maker notes that a blown fuse shows up as a temperature difference once the bank has been energized for at least an hour, which helps where units have no indicators.

None of these tells you why the fuse operated. They only tell you where to start measuring.

How to tell a real failure from a nuisance operation

Published troubleshooting steps line up. A test-equipment maker's guidance: if you find a blown fuse, shut down the entire bank and determine what caused it; common causes are bad capacitors, reactor problems and bad connections. One bank maker's table for a blown capacitor fuse: check for a shorted capacitor unit, check for a short circuit, and treat frequent unexplained blowing as a sign of parallel resonance. A capacitor maker's instructions say that where nothing external is visible, defective units must be found by capacitance measurement, and reconnecting without clarifying and correcting the cause is not permissible.

The RUS bulletin says to take the whole bank out of service when one phase is out, and a utility and relay maker paper notes that a blown fuse on a fused substation bank may require taking the bank out to prevent overvoltage failures of other units. One fuse maker offers an I2t inrush withstand test for suspect current-limiting fuses before they return to service.

Common mistakes

• Re-fusing and re-energizing without testing the unit. A good fuse on a failing unit will blow again or, worse, the next failure may rupture the case.
• Going up a size to stop resonance-driven blowing. The cause is a resonance; harmonic analysis and a filter or a different kvar are the fixes.
• Increasing fuse size past the case rupture limit. The fuse's maximum clearing curve must stay inside the unit's safe zone at the available fault current.
• Replacing a fuse but not the connection hardware that overheated it.
• Leaving one phase of a bank in service with the others out, or leaving a fused substation bank running with a blown fuse without checking the overvoltage on the remaining units.

Frequently asked questions