When the 87T trips and the transformer is healthy, the differential current is almost always manufactured by the measurement: a CT with the wrong polarity or ratio, a poorly set vector group compensation, zero sequence not removed during an external ground fault, CT saturation on a through fault, or a tap changer far from its nominal position. The oscillography of the external fault and the differential current at normal load point to the cause.
Symptom
The 87T transformer differential tripped during a fault on a line or feeder, or even at normal load, and the transformer shows no damage: the Buchholz did not operate, the electrical tests are good and the oil has not changed. For operations this is a misoperation due to lack of security: the protection operated for a condition outside its zone. This is complemented by SAT tests of a transformer differential 87T.
The 87T compares the currents entering and leaving the transformer. Before comparing them, the relay corrects three things: the magnitude (CT ratios and nominal voltages), the phase shift of the vector group (for example 30° in a Dyn1) and the zero sequence of grounded-wye windings. If any correction does not match what is connected in the field, the relay sees a differential current that does not exist.
Possible causes, from most to least likely
| Cause | When it shows | What gives it away |
|---|---|---|
| Zero sequence not removed | External ground fault on the grounded-wye side | Trips only on external ground faults; differential current similar to the fault 3I0 |
| Wrong vector group compensation | At load and on any through fault | Differential current in steady state, proportional to load |
| Reversed CT polarity | At load | Large differential current at load: ≈2 pu in one phase with no compensation, ≈1.15 pu in two phases with delta compensation, spread over all three if zero sequence is removed |
| CT ratio or nominal voltage entered wrong | At load, grows with load | Constant percentage difference |
| CT saturation on a through fault | Close-in external faults with a high DC component | Clipped secondary current on one side, difference only during the first cycles |
| Tap changer at an extreme position | At high load and on through faults | Differential current that varies with tap position |
| Physical CT connection different from what was assumed | After changes or after wiring was restored | CTs in delta or wye different from the settings; compensated currents on both sides 30° out of phase (differential ≈ 0.5 pu of load) |
| Overexcitation | High voltage or low frequency | High fifth harmonic; see the 24 overexcitation protection |
What to check first
- Differential and restraint current at normal load. Almost all numerical relays show them in their metering. With the transformer healthy and loaded, the differential current should be small compared with the restraint current. If it is not, the problem is in the connection or the settings, not in the fault that caused the trip.
- Oscillography of the external fault: in which phases the differential appeared, whether it matches the fault 3I0, and whether any secondary current looks clipped or distorted.
- Compensation settings: vector group, angle reference, zero-sequence removal per winding, CT ratios and nominal voltages. Compare each value with the transformer nameplate and the CT connection drawings.
- Drawings and actual CT connection: wye or delta, star point toward the transformer or toward the bus, a single ground per secondary circuit. See current transformer terminals and secondaries.
- Tap changer position at the time of the trip and its full range.
- CT sizing: accuracy class, knee point and connected burden against the maximum external fault current.
- Is there appreciable differential current compared with restraint at normal load?
- Yes: How does that differential current behave?
- Large at load: Reversed polarity of a CT (≈2 pu in one phase with no compensation, ≈1.15 pu in two with delta, or spread over all three if zero sequence is removed)
- Proportional to load: Vector group, CT ratio or nominal voltage entered wrong
- 30° out of phase after compensation (≈ 0.5 pu of load): CT connection in wye or delta different from the setting, or vector group set wrong
- Varies with tap position: Tap changer at an extreme position
- No: Under what condition did it trip?
- External ground fault on the grounded-wye side: Zero sequence not removed
- Close-in external fault with a high DC component: CT saturation; look for clipped current on one side
- High voltage or low frequency: Overexcitation; check the fifth harmonic and function 24
- Yes: How does that differential current behave?
How to confirm
- Load stability test (in-service phasor check): magnitude and angle of the currents on each side, with the angle expected from the vector group. It is the definitive check of polarity, ratio and compensation.
- CT tests: polarity, ratio and excitation curve and knee point to confirm whether the CT could saturate for that fault.
- Replay of the oscillography in the relay with secondary injection, with the present settings and with the corrected ones.
- Simulation of an external ground fault in the differential test: by injecting a through single-phase fault, the relay must remain stable with the correct zero-sequence removal.
How to keep it from coming back
- Run the load stability test at every commissioning and after any change of wiring, CTs or relay. It is the only test that validates the actual connection and the compensation setting together.
- Simulate external faults, single-phase and three-phase, with secondary injection during the differential test, not only internal faults: that verifies zero-sequence removal and the slope.
- Size the CTs for the maximum external fault and its DC component, and confirm that the CTs on all sides have similar characteristics.
- Document the compensation settings with the transformer nameplate, the CT drawings and the vector group, in a single document that is updated with every change.
What not to do
- Do not raise the slope or pickup to stop it tripping: that covers up a connection error that is still there, and the protection loses sensitivity to real internal faults.
- Do not reverse polarity “in the settings” without confirming in the field which CT is backwards; the error may be in another phase or on another side.
- Do not leave the 87T out of service without an active, agreed backup protection.
- Do not re-energize without ruling out an internal fault if there is any doubt: check the mechanical protections and the oil first.
When to call us
If your 87T tripped on an external fault and you cannot find the cause, our team reviews remotely the oscillography, the compensation settings and the differential metering under load. We work with relays from the main manufacturers. We tell you what to change and what to verify in the field. See remote relay configuration, available worldwide. If the CTs or the differential must be checked on site, we do protection testing and current transformer testing in Colombia, or elsewhere on request. Write to us.
Related guides
- Diagnosing protection trips: method
- Transformer trip on energization
- 87T transformer differential protection
- Restricted earth fault 87N
- Zero-sequence current
- On-load tap changer
- Current transformer terminals and secondaries
- SAT Tests of a Transformer Differential 87T
References
- IEEE C37.91-2021, Guide for Protecting Power Transformers.
- IEEE C37.110-2023, Guide for the Application of Current Transformers Used for Protective Relaying Purposes.
- IEC 61869-2, current transformers.
- Blackburn and Domin, Protective Relaying: Principles and Applications.
Frequently asked questions
Why does the 87T trip on an external ground fault?
Because a grounded-wye winding contributes zero-sequence current that does not appear on the delta side. If the relay does not remove zero sequence on that side, it measures it as differential current and trips for a fault outside the zone.
How do I know whether the problem is CT polarity or CT ratio?
Look at the differential current at normal load. With a reversed CT, a differential current appears that depends on the vector group compensation: about 2 times the load (in per unit) in one phase with no compensation, about 1.15 times in two phases with delta compensation, or spread over all three phases (close to 1.33, 0.67 and 0.67) if zero sequence is removed; with a wrong ratio, a difference proportional to load appears, always with the same sign.
Can the tap changer make the 87T trip?
On its own it should not, because the slope of the characteristic covers the ratio mismatch at the extreme taps. It can contribute when the slope was set without considering the full range of the tap changer and there is also a through fault with some saturation.