Transformer Inrush vs Internal Fault: How to Tell Them Apart

A transformer differential relay operates. The record shows a large differential current on one side and nothing on the other. That is either an internal fault requiring the transformer to stay out, or magnetising inrush on a perfectly healthy unit that should be put straight back in service. Getting it wrong in either direction is expensive, and the two look similar to a relay measuring only magnitude.

The short answer

Inrush is unipolar with a gap; fault current is bipolar without one. Magnetising inrush appears as a series of current pulses all on the same side of zero, separated each cycle by an interval where current is essentially zero, with rounded peaks and an envelope that decays over many cycles. Internal fault current crosses zero twice per cycle, is roughly sinusoidal apart from DC offset, and shows no gap. Confirm with context: inrush follows an energisation and comes with a voltage dip, not a voltage collapse.

Why inrush produces differential current at all

A differential relay compares what enters the protected zone against what leaves it. In normal service those balance, allowing for CT ratio and vector group compensation.

Inrush current breaks that balance legitimately. When a transformer is energised, current flows in to magnetise the core. It does not pass through to the other winding, because it is not load current. So the relay measures a large current on the energised side, nothing on the other, and computes a differential quantity that can reach several times rated current. The transformer is entirely healthy. The measurement is indistinguishable from an internal fault on magnitude alone.

How severe it is depends on the point on the voltage wave at which the breaker closes and on the residual flux left in the core. Closing at a voltage zero with residual flux in the same direction produces the worst case; closing at voltage peak with favourable remanence may produce almost nothing. This is why the same transformer energises cleanly one day and draws heavy inrush the next.

Reading the waveform

The gap is the single most reliable visual discriminator. It exists because the core is only driven into saturation during part of each cycle; for the rest, the magnetising current is negligible. A fault has no mechanism that would produce such an interval.

Second harmonic restraint, and where it fails

The asymmetric saturation that produces the gap also produces strong even harmonics, the second being dominant. Differential relays exploit this: measure the ratio of second harmonic to fundamental in the differential current, and block or restrain when it exceeds a threshold, typically set somewhere around 15 to 20 percent depending on the relay and the transformer.

It fails in both directions, and both failures show up in fault records.

• Restraint fails to block. Modern grain-oriented low-loss core steels saturate more sharply and produce inrush with a lower second harmonic fraction than the designs the classic thresholds were derived from. A restraint threshold set too high can let a healthy energisation trip. If you find a differential operation on energisation with second harmonic just below the setting, this is the likely explanation.
• Restraint blocks a real fault. A saturating CT during a genuine internal fault generates its own harmonic content, including second harmonic. That can hold off a trip that should have happened, delaying clearance. See how to identify CT saturation for the waveform evidence to look for here.

Relays mitigate this with cross-phase logic, where restraint on one phase influences the others, and with separate fifth harmonic restraint for overexcitation. When you analyse an operation, check which logic the specific relay applies, because the answer varies by manufacturer and by setting.

Cases that catch people out

• Sympathetic inrush. Energising one transformer can drive an already energised transformer on the same bus into partial saturation, producing inrush-like current in a unit nobody switched. Look for a differential current that begins exactly when a neighbouring breaker closed.
• Recovery inrush. When an external fault clears and voltage returns, the transformer re-magnetises and can draw inrush. Because it follows a real fault, it is easily misread as a fault that evolved into the transformer.
• Energising onto a fault. Both are present simultaneously. Expect bipolar current without a clean gap, plus second harmonic from the inrush component. This is the case where waveform shape earns its keep, because the harmonic ratio alone is ambiguous.
• Overexcitation. Sustained overvoltage or underfrequency drives the core into saturation symmetrically on both half-cycles, producing odd harmonics with the fifth dominant. Not inrush and not a fault, and restrained by different logic.

A practical checklist

1. Establish whether a breaker closed immediately before the current appeared.
2. Check polarity: is the current unipolar, or does it cross zero each cycle?
3. Look for the near-zero gap between peaks. Its presence points strongly to inrush.
4. Check whether the envelope decays over many cycles or stays constant.
5. Compare voltage behaviour. A dip that recovers suggests inrush; a collapse on specific phases suggests a fault.
6. Compute the second harmonic ratio in the differential current and compare it to the relay setting.
7. Check whether CT saturation is present, since it contaminates the harmonic evidence.
8. Plot the operate and restraint quantities against the relay characteristic to see how far into the trip region the operation actually was.

Doing this on your own record

Detego reads COMTRADE records from any manufacturer in the browser and computes the harmonic spectrum and the differential operate/restraint quantities against the 87T characteristic, including CT ratio and vector group compensation. Nothing installs and the file stays on your machine.

Detego AI evaluates the same evidence when it analyses a record: waveform polarity and gaps, harmonic content, the differential characteristic, and the relevant relay's own documented restraint logic drawn from its manufacturer manual. It reports what the measurements support rather than forcing an answer when the record is genuinely ambiguous, which energising onto a fault often is.

References

• IEEE C37.91, Guide for Protecting Power Transformers
• IEEE C37.110, Guide for the Application of Current Transformers Used for Protective Relaying Purposes
• IEC 60076-1, Power transformers: general