Ferroresonance in a voltage transformer — why an unearthed or Petersen-coil-earthed network can develop unexplained overvoltages

The guide on the Petersen coil covers how a resonant-earthed medium-voltage network deliberately lets a single earth fault continue without tripping immediately. This article covers a different phenomenon that can occur specifically on that type of network — and on a fully unearthed (IT) network, see the guide on IT-system first-fault detection — quite independent of any actual earth fault: ferroresonance between a voltage transformer (VT) and the network's own capacitance.

Why a non-solidly-earthed neutral makes ferroresonance possible

In a network with a solidly earthed neutral, the voltage of each phase relative to earth is fixed, forced by that low-impedance earth connection. In an unearthed or resonant-earthed (Petersen coil) medium-voltage network, by contrast, the neutral is not connected to earth at all, or only through a high impedance — the neutral's potential relative to earth is then not fixed, but determined by the balance between the three phases' capacitances to earth. If an inductive voltage transformer is connected between a phase and earth on such a network (for example for voltage measurement or an earth-fault indication system), the magnetizing inductance of that VT, together with the network's capacitance to earth, forms an LC circuit — exactly the ingredients needed for resonance phenomena.

Why this is not ordinary, linear resonance

An ordinary LC resonant circuit has a single, fixed resonant frequency, determined by a fixed inductance and a fixed capacitance. A VT core's magnetizing inductance, however, is strongly non-linear: at low magnetic flux the core behaves as a relatively high inductance, but as the core approaches saturation, the effective inductance drops sharply. This non-linearity means the LC circuit can resonate not just at one fixed frequency, but — depending on a chance trigger, for example switching an unloaded cable or transformer, or a brief earth fault that clears itself — can "lock" into one of several possible, stable resonant states: for example at the network frequency itself, at a sub-harmonic frequency (a fraction of the network frequency), or at a higher harmonic. In any of these states, the voltage across the VT winding and the current through it can rise well above normal operating values, and — unlike an ordinary, brief switching transient — can remain sustained at that elevated level until the circuit is disturbed in some other way.

What ferroresonance causes in practice

Overheating and failure of the VT itself: the strongly increased magnetizing current during ferroresonance can stress the core and winding of the voltage transformer well beyond its thermal limits, potentially resulting in a burnt-out VT.
Unexplained, audible humming or vibration of the VT core, often the first observable symptom before the VT actually fails.
Overvoltage on the network phases themselves, which can stress other equipment connected to the network, unrelated to the VT where the phenomenon originated.
Confusing measurements: a voltage measurement or earth-fault indication that shows an apparent earth fault during ferroresonance, while no fault is actually present on the network, potentially causing an unnecessary and time-consuming troubleshooting effort.

The common countermeasure: a loading resistor on the open delta

The most common way to suppress ferroresonance in a VT on an unearthed or resonant-earthed network is fitting a sufficiently low-impedance loading resistor across the open-delta winding (the auxiliary winding normally used for earth-fault detection) of the VT set. This resistor damps the energy in the LC circuit enough to keep the non-linear core out of, or pull it out of, a ferroresonant state, without disturbing the VT's normal measuring or indication function. The resistor can be permanently connected (thermally rated to dissipate continuously) or switched in as soon as a ferroresonant condition is detected — a design choice that depends on the VT's allowed continuous burden and the desired damping.

Note: the exact sizing of the loading resistor, and the choice between a permanent or switched design, follow from a system study that accounts for the actual network capacitance and the type and magnetizing characteristic of the applied VT; this article covers the principle, not a ready-made sizing formula for every network configuration.

Practical relevance

When designing or reviewing an earth-fault detection system with inductive voltage transformers on an unearthed or resonant-earthed medium-voltage network, it must be explicitly checked whether a sufficiently sized damping resistor is present on the open-delta winding — its absence is a known cause of unexplained VT failures and apparent earth-fault alarms that, on closer investigation, turn out to be false.

Common mistakes

Connecting an inductive voltage transformer to an unearthed or resonant-earthed network without a damping resistor on the open-delta winding — this leaves the network susceptible to ferroresonance after every switching operation or brief earth-fault disturbance.
Confusing a ferroresonant VT failure with an ordinary insulation fault or manufacturing defect — without recognizing the phenomenon, a replacement VT can fail again without the underlying cause being removed.
Undersizing or omitting the damping resistor to make measurements "more sensitive" — a resistor that is too high in value, or missing, reduces damping precisely when it is needed most.
Treating an apparent earth-fault indication during ferroresonance as a real earth fault without further checking — this can lead to a lengthy, fruitless troubleshooting effort while the network is actually fault-free.