A transformer differential relay (87T) is built on one simple idea: current going in should equal current coming out. That idea works well until you close the breaker on an unloaded transformer. For a few hundred milliseconds the HV side may draw 6 to 10 times rated current while the LV side carries nothing. To the relay, that looks exactly like an internal fault.
This current is called magnetizing inrush current. The standard way to stop the relay from tripping on it is 2nd-harmonic restraint, also called harmonic blocking. This article explains where inrush comes from, why its waveform contains so much 2nd harmonic, how relays use that, how to set and test the function, and where it can fail.
Read the complete protection of transformer: Power Transformer Protection: Complete Guide (87T, REF, 24).
What is magnetizing inrush current?
Every transformer needs a small magnetizing current to set up flux in its core. In steady state this current is tiny, usually well under 1% of rated current. Inrush happens when the core is pushed deep into saturation during energization. The magnetizing current then becomes very large for a short time.
Why the core saturates
Core flux is the time integral of applied voltage:
In steady state, flux lags voltage by 90°. When voltage crosses zero, flux is at its negative or positive peak (±Φm).
Now suppose the breaker closes at a voltage zero and the core has no residual flux. Flux has to start from zero rather than from −Φm. Over the next half cycle it rises by 2Φm and reaches 2Φm. If the core still holds residual flux Φr in the same direction, the peak becomes:
Power transformer cores are designed to run near the knee of the B-H curve, around 1.6 to 1.7 T. Twice that, plus remanence, puts the core far into saturation. In saturation the core’s permeability collapses toward that of air, magnetizing inductance falls sharply, and current climbs to very high peaks.
Typical characteristics
| Parameter | Typical value |
|---|---|
| First peak magnitude | 6–10 × rated current for large units, higher for small units |
| Duration | Tens of cycles to several seconds, longer for large, low-loss units |
| Waveform | Unidirectional, peaky, with “gaps” near zero |
| DC component | Large, decays with the transformer’s L/R time constant |
| 2nd harmonic content | Often 15–60% of the fundamental, can be lower in modern cores |
| Where it appears | Only on the energized winding, so it is pure differential current |
Factors that affect inrush severity
- Switching angle. Closing at voltage zero is the worst case. Closing at voltage peak produces almost no inrush. That is why controlled (point-on-wave) switching is used to reduce inrush.
- Residual flux. Remanence can reach a large fraction of peak flux, depending on how the transformer was last de-energized. It adds to or subtracts from the inrush flux.
- Source impedance. A stronger source (lower impedance) gives higher inrush current.
- Winding energized. Energizing the inner winding (usually LV) generally gives higher inrush than energizing the outer winding, because the air-core inductance is smaller.
- Core material and design. High-permeability, low-loss steel and step-lap joints change both the magnitude and the harmonic content.
- Transformer size. Larger units have longer time constants, so inrush decays more slowly.
Three types of inrush
Energization inrush happens when an unloaded transformer is switched on. It is the classic and usually the largest case.
Recovery inrush happens when an external fault is cleared and system voltage returns quickly. The core flux shifts suddenly and the transformer draws inrush-type current again. It is usually smaller than energization inrush, but the transformer is already loaded at the time.
Sympathetic inrush happens when a transformer already in service, in parallel with one being energized, starts drawing inrush current too. The DC component of the incoming unit’s inrush flows through the shared source resistance. This shifts the flux in the running unit. Sympathetic inrush builds up slowly, can last a long time, and has caught out many relays whose harmonic blocking had a time limit.
Why inrush trips an unrestrained differential relay
The 87T relay calculates differential current after ratio and vector group compensation:
During energization, I2 (LV side) = 0, so Idiff = I1 = inrush current.
On the biased differential characteristic, the restraint current is also small, because only one side carries current. The operating point lands well above the first slope. The bias characteristic cannot tell inrush apart from an internal fault. A separate function that looks at the shape of the current is needed.
Why the inrush waveform is rich in 2nd harmonic
A symmetrical sinusoid, or a symmetrically distorted wave like a saturated-CT fault current, contains mainly odd harmonics. Even harmonics appear when a waveform is not symmetrical between its positive and negative half-cycles.
Inrush current is heavily asymmetrical. Current flows as tall pulses in one direction only during the saturated part of each cycle. For the rest of the cycle it is almost zero, which creates a dead angle or gap. Fourier analysis of this waveform gives a large DC component, a fundamental (50 Hz) component, a significant 2nd harmonic (100 Hz) component, and smaller 3rd and higher harmonics.
Internal fault current, on the other hand, is mostly sinusoidal at fundamental frequency, with a decaying DC offset. Its 2nd harmonic content is usually low.
The principle in one line: if the 2nd-harmonic-to-fundamental ratio in the differential current is high, the relay treats it as inrush and blocks. If the ratio is low, the relay treats it as a fault and trips.
How 2nd-harmonic restraint works
The basic criterion
For each phase, the relay calculates the fundamental (I1f) and 2nd harmonic (I2f) of the differential current, usually with a one-cycle DFT. It then checks:
Typical setting: 15% of the fundamental. Common range is 10–20%. Several manufacturers ship with 15% as the default. Some utilities use 12% or lower for modern low-loss transformers.
“Restraint” vs “blocking”
The two terms are often used interchangeably, but strictly they differ. With harmonic blocking, the differential element is blocked completely when the ratio exceeds the setting. With harmonic restraint, the harmonic current is added to the restraint quantity, which raises the operating threshold. The original electromechanical relays used this method. Most modern numerical relays use blocking logic on a ratio, but the term “2nd-harmonic restraint” is still used everywhere, including in this article.
Per-phase blocking vs cross-blocking
This is the most important setting decision, and the one most often misunderstood.
In a three-phase transformer, the three phase currents have different inrush patterns because each phase’s flux starts at a different point on its wave. One phase may show 40% 2nd harmonic while another shows only 8%. After delta-side compensation (phase-to-phase current subtraction), harmonic content can be diluted even further.
| Method | How it works | Trade-off |
|---|---|---|
| Per-phase (independent) blocking | Each phase is blocked only by its own harmonic content | Secure for internal faults, but a low-harmonic phase can trip on inrush |
| Cross-blocking | Inrush detected in any phase blocks all three phases | Very secure against inrush, but a fault during energization may be delayed |
| Averaging / sum methods | Uses the sum or average of harmonic content across phases | A middle-ground approach |
Modern relays limit how long cross-blocking stays active. For example, SIPROTEC transformer differential relays have a cross-block time setting in cycles. Within this time, inrush in one phase blocks all phases. After it expires, each phase relies on its own harmonic content. A short cross-block time (a few cycles) covers the first peaks, where the low-harmonic-phase problem is worst, without holding the relay blocked for long.
The unrestrained differential stage (87H / Idiff>>)
Harmonic blocking has a weakness. A heavy internal fault can saturate the CTs, and saturated CT secondary current also contains even harmonics. If 2nd harmonic rises above the setting, the restrained 87T stage may be blocked during a real fault.
The solution is the unrestrained (high-set) differential stage, often labelled 87H, Idiff>>, or I-DIFF>>. This stage has no harmonic blocking and no bias, operates on instantaneous or fundamental differential current, and is set above the maximum expected inrush so that it cannot trip on energization.
Typical practice is to set it above the highest expected inrush fundamental and above the maximum through-fault current that could appear as false differential current. A common starting point is around 1/uk of rated current, where uk is the transformer impedance in per unit. For uk = 12.5%, that gives 8 × In. Typical settings fall in the 6–12 × In range. Confirm the value against energization records whenever you can.
Worked example
| Transformer | 50 MVA, 132/33 kV, Dyn11, uk = 12.5% |
|---|---|
| HV rated current | 50,000 / (√3 × 132) = 218.7 A |
| Harmonic settings | 2nd harmonic = 15%, cross-block = 3 cycles |
| Differential settings | Idiff> = 0.3 In, Idiff>> = 8 In |
Case A: Energization, phase L1
Fundamental in Idiff = 2.4 In, 2nd harmonic = 0.84 In. Ratio = 0.84 / 2.4 = 35%, above 15%, so L1 is blocked ✔
Case B: Energization, phase L3 (low-harmonic phase)
Fundamental = 1.1 In, 2nd harmonic = 0.10 In. Ratio = 0.10 / 1.1 = 9%, below 15%.
With per-phase blocking only, L3 is not blocked. Idiff = 1.1 In is above Idiff>, so the relay trips on inrush ✘
With cross-blocking, L1’s detection blocks L3 too, so the relay stays stable ✔
Case C: Internal fault on LV winding
Fundamental = 5.0 In, 2nd harmonic = 0.20 In. Ratio = 4%, not blocked, so the relay trips ✔
Case D: Severe internal fault with CT saturation
Fundamental = 14 In, 2nd harmonic ≈ 3.0 In (21%). The restrained stage is blocked, but 14 In is above Idiff>> = 8 In, so the unrestrained stage trips instantly ✔
These four cases show why a reliable scheme needs three things together: a sensible harmonic threshold, cross-blocking, and a properly set unrestrained stage.
The modern transformer problem: low 2nd harmonic inrush
Older transformers routinely showed 30% or more 2nd harmonic during inrush. Modern transformers built with high-permeability grain-oriented steel, step-lap core joints, and lower design flux density can produce inrush with 2nd harmonic below 10% in one or more phases. There are documented cases of correctly set 15% relays tripping on energization.
Practical responses include:
- Lowering the threshold to 10–12%, which costs some speed during internal faults with CT saturation.
- Using cross-blocking for the first few cycles.
- Using additional inrush detection methods offered by modern relays: waveform or dead-angle detection (inrush has flat, near-zero gaps that fault current does not have; ABB and others use this alongside harmonic detection), DC component or asymmetry detection, and flux-based methods that estimate core flux from voltage.
- Point-on-wave controlled switching to reduce inrush at the source.
If your relays have both harmonic and waveform-based inrush detection, enabling both gives the best security.
2nd harmonic vs 5th harmonic: don’t mix them up
Both functions appear in the same relay menu, but they cover different conditions.
| 2nd harmonic | 5th harmonic | |
|---|---|---|
| Detects | Magnetizing inrush | Overexcitation (high V/f) |
| Waveform cause | Asymmetrical saturation | Symmetrical saturation |
| Typical setting | 15% | 30–35% |
| Why block | Inrush is not a fault | Overexcitation is handled by 24 (V/Hz) protection |
Overexcitation saturates the core equally in both half-cycles, so it produces odd harmonics (3rd and 5th) rather than 2nd. The 3rd harmonic is mostly removed by delta windings and zero-sequence elimination, so the 5th harmonic is used instead. See the overexcitation (24) section in our transformer protection pillar guide.
Testing 2nd-harmonic restraint with OMICRON CMC 356
The harmonic restraint test confirms that the relay blocks at the set ratio and that cross-blocking works. In OMICRON Test Universe, use the Diff Harmonic test module, or a State Sequencer / Ramping test with harmonic signals added.
Test 1: Harmonic blocking threshold (per phase)
- Inject current into one side only (for example HV, L1) so that all of it appears as differential current.
- Choose a fundamental differential current safely above pickup, for example 2 × Idiff>, and below Idiff>>.
- Superimpose 2nd harmonic at a ratio above the setting, for example 25%. The relay should not trip.
- Ramp the 2nd harmonic ratio down in small steps (0.5%) until the relay trips.
- Record the trip ratio. It should match the setting within the manufacturer’s tolerance.
- Repeat for L2 and L3, and at a second fundamental level, for example 4 × Idiff>.
Watch the matching factors. With single-phase injection on a delta-compensated or zero-sequence-eliminated side, the differential current the relay sees may differ from the injected value by a factor (often 1/√3 or 2/3, depending on vector group and compensation). The Diff Harmonic module accounts for this when the transformer data is entered correctly. In manual State Sequencer tests you need to calculate it yourself. Our article on differential stability conditions explains these matching factors.
Test 2: Cross-blocking
- Inject fundamental differential current in L1 and L2.
- Add 2nd harmonic (above setting) only to L1. L2 carries pure fundamental.
- Cross-block enabled: neither phase trips during the cross-block time. After the time expires, L2 should trip.
- Cross-block disabled: L2 trips immediately, and L1 stays blocked.
Test 3: Unrestrained stage is not blocked
Inject differential current above Idiff>> with 2nd harmonic well above the setting, for example 30%. The relay must trip instantly through the unrestrained stage. This confirms harmonic blocking is not wired into 87H.
Testing tip: some relays calculate the harmonic ratio against the fundamental, and others against the total RMS. Check the manual so your expected values are correct.
Field troubleshooting: 87T tripped on energization
If a differential relay trips during transformer energization, work through these checks in order:
- Pull the disturbance record. Confirm current is on the energized side only and that the waveform shows inrush features: unidirectional pulses with gaps.
- Compute the 2nd harmonic ratio per phase, using SIGRA, OMICRON TransView, or the relay’s own fault analysis. Note which phase had the lowest ratio.
- Check settings: harmonic threshold, cross-block enabled, and cross-block time.
- Check Idiff>>: was it set below the inrush fundamental peak?
- Check CT and vector-group compensation. Wrong matching can create false differential current that adds to inrush. Our CT article covers CT ratio and saturation basics.
- Look at the system: was a parallel transformer already in service (sympathetic inrush)? Was the unit recently DC-tested? Winding resistance tests leave high residual flux, so energization right after testing can produce very severe inrush.
The last point is a common real-world cause. Where possible, demagnetize the core after winding resistance or other DC tests before energizing.
Key takeaways
- Inrush current is caused by core saturation during energization. It is worst when switching at voltage zero with residual flux in the same direction.
- It appears on one side only, so the differential relay sees it as an internal fault.
- Its asymmetrical waveform contains strong 2nd harmonic. Fault current does not.
- Relays block the 87T stage when I2f/I1f exceeds a setting, typically 15%.
- Cross-blocking protects against the low-harmonic phase problem. Keep its time short.
- The unrestrained stage (Idiff>>) must never be harmonic-blocked, and it must be set above maximum inrush.
- Modern low-loss transformers can show low 2nd harmonic. Consider waveform-based detection and lower thresholds.
- After DC testing, demagnetize the core before energizing.
Frequently asked questions
What is the typical 2nd harmonic restraint setting for transformer differential protection?
15% of the fundamental differential current is the most common setting and a typical manufacturer default. Values from 10% to 20% are used depending on transformer design and utility practice.
Why is 2nd harmonic used to detect inrush and not 3rd harmonic?
Inrush current is asymmetrical between half-cycles, which produces even harmonics, mainly the 2nd. The 3rd harmonic appears in both inrush and other conditions, and it is largely removed by delta windings and zero-sequence filtering, so it is not a reliable indicator.
How long does magnetizing inrush current last?
From a few cycles to several seconds. Large transformers with high X/R ratio and sympathetic inrush cases can last considerably longer.
Can inrush current trip overcurrent relays too?
Yes. Instantaneous overcurrent (50) elements on the transformer feeder can pick up on inrush. Many numerical relays include an inrush restraint function for overcurrent elements based on the same 2nd harmonic principle.
What is cross-blocking in transformer differential protection?
Cross-blocking means that if inrush is detected in any one phase, the differential element is blocked in all three phases. It prevents tripping in a phase whose own 2nd harmonic content is too low to block it.
Does the unrestrained differential stage use harmonic blocking?
No. Idiff>> operates without harmonic blocking or bias, so that severe internal faults with CT saturation trip without delay.
