An external arc occurs inside the unbalance protection zone but outside a capacitor unit, for example along the insulator of a fuse rail or on the rack. It produces very little phase overcurrent and can spread from group to group. According to IEEE C37.99-2000, the most effective protection is a fast unbalance trip, with a delay no shorter than the maximum clearing time of the fuse.
Short answer
An external arc is a fault that occurs inside the unbalance protection zone but outside a capacitor unit: for example, along the insulator that supports a fuse rail, or on the rack. According to IEEE C37.99-2000, it can cause extensive damage within a fraction of a second, create a hazard for people and spread from group to group if the bank is not tripped quickly. Overcurrent protection hardly sees it. The most effective protection is a fast unbalance relay, with the shortest delay that coordination with the fuses allows.
What an external arc and a rack fault are
The guide defines an external fault in a bank as one that lies inside the unbalance protection zone but outside a capacitor unit (clause 8.1). It is not a punctured element inside a can. It is an arc in air or across an insulator, between parts at different potentials.
The rack fault is the typical case. If the bank is built with the phases well separated on different structures, the arc starts across a single series group. That fault produces very little phase overcurrent. If the unbalance relay does not operate, more and more series groups of the same phase become involved until the bank overcurrent protection trips or the fuses clear. By then the damage is large: many blown fuses and ruptured units (clause 7.1.5).
This is one of the reasons unbalance protection is regarded as the primary protection against arcing inside the bank. The schemes are explained in the guide on capacitor bank unbalance protection.
The IEEE C37.99 guide to capacitor bank protection places this fault within the overall bank protection.
Why overcurrent protection is not enough
The standard is direct on two points:
- Overcurrent protection is not normally sensitive to an arc inside the bank and, if it senses the disturbance, it does so late (clause 7.2.1).
- Instantaneous relays are not usually effective for rack faults because their setting must be high, so that they do not trip on inrush and outrush currents between banks (clause 7.1.5).
The physical reason is simple. An arc across one series group short-circuits a small fraction of the phase impedance. The current that flows is barely distinguishable from the service current. What does change, and clearly, is the distribution of voltage and current inside the bank. That is why it is detected with unbalance measurements and not with line current.
What overcurrent protection does cover, and what is left to other functions, is explained in overcurrent and breaker failure protection in a capacitor bank.
What protects against the arc
According to clause 7.1.5, the most effective protection is a fast unbalance relay. Two ideas apply. They are worth keeping clear, but they must be fixed in the bank study.
Short delay, with a floor. A short delay reduces the damage from rack faults. The standard mentions that delays as short as 0.05 s have been used. That delay, however, should not be shorter than the maximum clearing time of the unit or element fuse: if the relay tripped first, it would prevent the fuse from doing its job of isolating the damaged unit and giving a visible indication.
Sensitivity set for a lost unit, not for an arc. The unbalance trip is set to protect the units or elements from continuous overvoltage after a unit is lost. With that criterion, the resulting sensitivity is enough to detect the initial arc across one series group of a phase.
When the unbalance scheme is of the type that responds to system unbalance, a long delay may be needed to coordinate with the line relays. A scheme that does not respond to the system zero-sequence voltage, or that compensates for it, avoids that additional delay (clause 7.1.5).
When neutral unbalance does not see the arc
The guide warns about one specific case: an ungrounded wye bank with two series groups per phase, with the three phases stacked in a single metal structure. A phase-to-phase arc at mid-height of the rack (figure 5 of the standard) produces no unbalance in the neutral voltage, nor neutral current if the bank were grounded. A neutral relay does not respond. The fault can spread until overcurrent protection, timed or instantaneous, clears it, with considerable damage to all three phases.
The standard presents two protection methods against this type of phase-to-phase arc:
| Method | General idea | Limitation noted by the standard |
|---|---|---|
| Definite-time overcurrent, fast response | Detect an increase above normal phase current with a short delay and fast reset | It must be selective with respect to unbalances and energization |
| Unbalance current or negative-sequence current | Detect the asymmetry caused by the phase-to-phase arc, as backup | The trip must be delayed to coordinate with the rest of the system, and it may not prevent arc damage inside the structure |
In double wye banks the unbalance protection does operate for this type of fault, provided the right groups are connected to the rack. That detail is covered in the guide on ambiguous indications and faults that unbalance protection does not see.
Negative-sequence protection (46) is explained separately.
The same problem appears in ambiguous indications and faults that unbalance protection does not see, and the two-branch arrangements that make it worse are in H-bridge unbalance and in double wye unbalance.
Defective connections and case ruptures
The standard sets out an important limit (clauses 7.1.4 and 8.1): unbalance protection is not normally sensitive enough to detect a defective connection. A loose connection deteriorates until a fault occurs inside the bank, and only then does the unbalance protection operate. Inside a unit, a defective connection, which is uncommon, can raise the pressure and rupture the case before the unbalance protection operates. If a case ruptures, the protection must act quickly because of the external arc that follows.
This has a practical consequence: electrical protection does not replace inspection. Thermography, connection checks and maintenance find what no relay can see. The same applies to the high system short-circuit current inside a unit: the standard notes that unbalance schemes do not operate fast enough to prevent catastrophic failures from that cause, and that in certain single-series grounded banks an external fuse may be advisable.
What not to do
- Do not rely only on overcurrent protection against arcing in the bank.
- Do not use the line instantaneous element as rack fault protection: its setting makes it insensitive.
- Do not set an unbalance delay below the clearing time of the unit or element fuse.
- Do not assume that a neutral scheme covers a phase-to-phase arc if the phases share a structure.
- Do not leave the rack without correct equipotential bonding in double wye banks: it can defeat the response of the scheme.
- Do not re-energize a bank that tripped on unbalance without a visual inspection and unit tests.
When to contact us
If your bank has its phases in a single structure, if the unbalance trip came late or if you are not sure what your current protection covers, at Tiza Group we review the scheme and leave the relay in order. Remote relay configuration covers the settings, and if an event has already happened, we can quote the trip analysis. Write to us.
Related guides
- IEEE C37.99: capacitor bank protection guide
- Ambiguous unbalance: faults the protection does not detect
- Overcurrent and breaker failure protection in a capacitor bank
- Capacitor bank unbalance protection 60
- Capacitor bank trips: causes and diagnosis
References
- IEEE C37.99-2000, IEEE Guide for the Protection of Shunt Capacitor Banks, clauses 7.1.4, 7.1.5, 7.2.1, 8.1 and figure 5.
Frequently asked questions
Why does overcurrent protection not detect an arc inside the bank?
Because an arc across a single series group barely changes the phase current. IEEE C37.99-2000 (cl. 7.2.1) states that overcurrent protection is normally not sensitive to an arc inside a bank and, if it sees anything, it sees it late. The settings needed to ride through inrush currents make it even less sensitive.
Does a neutral unbalance relay detect a phase-to-phase fault at mid-height of the rack?
Not always. If the three phases are stacked in a single structure, a phase-to-phase arc at mid-height may not change the neutral voltage or current. That is why the standard warns against relying only on that scheme in banks whose phases are not well separated, and it mentions other backup methods.
How much delay can the unbalance protection have when there is an arc risk?
The standard states that delays as short as 0.05 s have been used, but not shorter than the maximum clearing time of the unit or element fuse. A shorter delay would make the relay trip before the fuse can isolate a damaged unit. The actual value comes from the coordination study of each bank.