Switching a capacitor bank is more demanding than other switching duties. IEEE C37.99-2000 asks that the breaker be applied within its voltage, frequency and current limits, with margins for overvoltage (110 %), capacitance tolerance (115 %) and harmonics (110 %). The energization current depends on the point on the wave at closing and on the bank; between banks on the same bus it can be very high and of high frequency. Metal-oxide arresters are better than silicon carbide ones.
Short answer
Switching capacitors is more demanding than any other switching duty. On opening, the capacitor is left charged near the peak voltage, and a restrike can produce overvoltages. On closing, an energization (inrush) current appears that can be very high and of high frequency, especially if another bank is already energized on the same bus.
IEEE C37.99-2000 covers the subject in clauses 10.1 to 10.3 and in Annex D. This guide explains the breaker margins, restrike, energization current and surge arresters. Sizing a breaker or a reactor is outside its scope.
Breaker margins
Every capacitor switching device must be applied within its maximum voltage, frequency and current ratings, including the transient inrush current and its frequency. The standard says the current rating of the device must include margins for:
| Effect | Margin |
|---|---|
| System overvoltage | up to 110 % |
| Capacitance tolerance of the units | up to 115 % |
| System harmonics | up to 110 % |
For application values, the standard refers to the breaker standards it cites, among them C37.04, C37.06, C37.012 and C37.66.
Opening: reignition and restrike
On opening, at current zero the capacitor is left charged almost to the peak line voltage. Little voltage appears across the contacts at that instant, and the capacitive arc normally interrupts at the first current zero. Half a cycle later, the recovery voltage across the open contacts approaches twice the peak value, because the source-side voltage keeps alternating.
If the gap between contacts breaks down, the capacitor tries to equalize its voltage with that of the system in an oscillating circuit. At the first peak, depending on damping, the capacitor voltage overshoots by an amount close to the difference between the two voltages before the breakdown. That overvoltage can damage equipment.
The standard distinguishes two cases:
- Reignition. The gap breaks down less than 0.25 cycle after current zero. The oscillation does not exceed the normal transient of the bank’s first energization.
- Restrike. The breakdown occurs half a cycle after interruption. It is the worst case of the first restrike. In theory the overvoltage reaches three times the phase-to-ground crest. In practice it rarely exceeds 2.5 times.
Additional restrikes can produce even higher crests. That is why they should be limited. For the type of breaker, see classes C1 and C2 and the special-purpose switch for banks and motors.
Resonance with the network caused by restrikes is covered in resonance in capacitor banks.
Energization current
What it depends on
When a bank is energized, a current transient appears. Its magnitude and frequency depend on:
- The applied voltage, that is, the point on the wave at which it closes.
- The capacitance of the circuit.
- The inductance of the circuit.
- The initial charge of the bank at closing.
- Damping, from closing resistors or other resistances in the circuit.
Isolated bank and back-to-back
In an isolated bank, the energization current is lower than the short-circuit current available at that point. It therefore does not limit the application of the switching device, although the momentary rating of other devices not built to interrupt faults must be checked.
In back-to-back switching, that is, when one or more banks are already energized on the same bus when another closes, high-magnitude, high-frequency transient currents can flow between the banks. They are limited only by the impedance of the capacitors and of the circuit between them. They normally decay within a fraction of a cycle, and the part supplied by the source is so small it is usually neglected.
Order of magnitude: Annex D equations
Annex D (informative) gives the equations for the undamped peak current. The real value is around 90 %.
- Isolated bank, no previous charge: i max = square root of (2 × Isc × I1), and frequency = fs × square root of (Isc / I1). Isc is the symmetrical short-circuit current at the point and I1 the rated current of the bank.
- Back-to-back, with the bank being closed uncharged: i max (kA) = square root of (1000 / (3 × π × fs)) × square root of (kvar1 × kvar2 / (Leq × kvarT)), with Leq in µH. The frequency is 9.5 × square root of (fs × kVLL × (I1 + I2) / (Leq × I1 × I2)), in kHz.
If the bank is charged, as in a restrike, the current can be double.
A worked example with the equations above: a 10 Mvar bank at 34.5 kV has a rated current of 167 A. At a point with 20 kA short-circuit current, the undamped peak on energization is about 2600 A, at a frequency near 660 Hz. If another identical bank is already energized and the inductance between them is 20 µH, the undamped peak rises to about 21 kA and the frequency to about 10.6 kHz. The same bank produces a transient about eight times larger when there is another one beside it.
How it is reduced
| Measure | Effect according to the standard |
|---|---|
| Series current-limiting reactors | Change the frequency and reduce the magnitude; they also reduce the outrush current to a nearby fault and extend contact life in daily switching |
| Pre-insertion resistors or inductors in the breaker | Minimize the inrush current, but do not help with the outrush current to a nearby fault |
| Synchronized closing or closing at voltage zero | Reduces the severity of the transient |
| Inductance between banks | Even a minimal amount significantly reduces the current between banks |
Reactors must have an impulse insulation level high enough that the spark gaps or arresters protecting them do not short them when the bank is energized. And the standard says grounded-wye banks should be installed as far as practical from the control room and the cable trenches.
Inrush current can also trigger resonances with other parts of the system, induce overvoltages in control cables and interfere with communications in the area. If the rate of change of current (dI/dt) or its crest exceeds the limits in the tables of C37.06-1997, consult the breaker manufacturer. And in a substation with several large banks on the same bus, check that the transient discharge of one bank into a nearby fault does not exceed the rating of any breaker on the bus.
The effects of that current on CTs, VTs and control cables are in switching transients in CTs, VTs and control circuits, and its effect on overcurrent, in overcurrent and breaker failure in a bank. Single-point or peninsula grounding, which reduces transients between banks, is covered in bank grounding, neutral and ground grid.
Surge arresters
Lightning strikes and capacitor switching produce significant overvoltages. Where lightning activity is high, or where the bank is switched many times a day, arresters can be placed on the bank. Their ability to dissipate the energy of switching is the most important factor when selecting them.
Restrikes of the switching device usually cause the highest overvoltages, especially in isolated banks. Overvoltages from magnification in resonant circuits of the system can also appear when a remote bank, cable or line is switched.
The standard says metal-oxide arresters are better than silicon carbide ones: they normally withstand more energy for the same rated voltage and absorb less energy per event. See surge arrester.
Common mistakes
- Checking only the rated current of the breaker. It must include margins for overvoltage, tolerance and harmonics.
- Ignoring the neighboring bank. The current between banks is far higher than that of an isolated bank.
- Reactors without enough insulation. The spark gaps can short them on energization.
- Arresters without checking their energy. Repeated switching demands absorption capability.
What we do at Tiza Group
In testing and commissioning we verify the switching of the bank with its breaker and arresters before it is put into operation. We do this in Colombia, or elsewhere on request. Contact us and we will quote the verification.
Related guides
- IEEE C37.99: shunt capacitor bank protection guide
- Bank transients in CTs, VTs and control circuits
- Overcurrent and breaker failure in a capacitor bank
- Grounding a capacitor bank: neutral and ground grid
- Automatic control of MV and HV capacitor banks
References
- IEEE Std C37.99-2000, IEEE Guide for the Protection of Shunt Capacitor Banks, clauses 10.1 to 10.3 and Annex D, Figure 45.
- ANSI C37.06-1997, inrush current and rate-of-change tables cited by the standard.
- IEEE Std C62.22-1997, IEEE Guide for the Application of Metal-Oxide Surge Arresters for Alternating-Current Systems.
Frequently asked questions
What is the difference between a reignition and a restrike?
Per IEEE C37.99-2000, if the gap between contacts breaks down less than a quarter cycle after current zero, it is a reignition and the oscillation does not exceed the normal energizing transient. If it happens half a cycle later, the worst case, the overvoltage can in theory reach three times the phase-to-ground crest and in practice rarely exceeds 2.5.
Why is the energization current between two banks on the same bus so high?
Because the oscillatory current between banks is limited only by the impedance of the capacitors and of the circuit between them. It is of high frequency and decays within a fraction of a cycle. Even a small inductance between banks reduces that current significantly.
Which surge arresters suit a capacitor bank?
The standard says metal-oxide arresters are usually better than silicon carbide ones, because they normally withstand more energy for the same rated voltage and absorb less energy per event. The ability to dissipate the energy of bank switching is the most important selection criterion.