How to Identify CT Saturation in a Fault Record
By Yarden Jordan · August 21, 2026
Current transformer saturation is the most common reason a fault record misleads the engineer reading it. The relay did what its measurements told it to do; the measurements were wrong. This guide covers how to recognise saturation in a recording, why it distorts each class of protection differently, and how to separate it from the two things it is most often confused with: DC offset and transformer inrush.
The short answer
CT saturation shows up as a change in the shape of the current wave, not its size. The secondary current follows the primary faithfully for part of each cycle, then collapses toward zero as the core saturates, then recovers abruptly when the flux reverses. On screen this reads as a waveform with a missing or flattened section each cycle and a sharp edge where it recovers. The voltage channels stay comparatively clean, and the distortion usually appears on some phases and not others.
Why a CT saturates
A current transformer works by driving flux around a magnetic core. The secondary current reproduces the primary only while that core stays in its linear region. Push the flux past the knee point and the core can no longer sustain the rate of change required, so the secondary output collapses regardless of what the primary is doing.
Three factors push a CT past its knee point during a fault:
- Fault current magnitude. A CT sized for load current can be driven well past its rating by a close-in fault.
- DC offset in the fault current. This is usually the dominant cause. The unidirectional component drives flux in one direction without the reversal that would otherwise reset it each cycle. A system with a high X/R ratio produces a slowly decaying offset, and the flux accumulates.
- Remanent flux. Flux left in the core from a previous fault or from interrupted DC can bias the core so that a much smaller fault saturates it. This is why a CT that behaved on one event can misbehave on the next.
Burden matters too. A long cable run or a heavily loaded secondary circuit raises the voltage the CT must develop, and therefore the flux, for the same primary current.
Reading the waveform
Open the record and look at the current channels through the fault. The signature has three parts, and all three should be present before you call it saturation.
- A faithful first portion. Saturation is not instantaneous. The core takes time to accumulate flux, so the first part of the first fault cycle is usually accurate. That window is the reason high-speed protection can still work on a saturating CT.
- A collapse toward zero. Once past the knee, output falls away. The peak that should be there is truncated or missing entirely.
- An abrupt recovery. When the flux comes back out of saturation, the secondary current snaps back. That sharp edge is the most distinctive visual feature and is what separates saturation from every smooth-envelope phenomenon.
Severity rises through the fault rather than being constant. Early cycles are often usable and later ones are not, because flux keeps accumulating while the DC offset persists.
Harmonic evidence
Saturation produces a broad spectrum rather than one dominant order, with even and odd harmonics both present. The conventional indicator is the ratio of second harmonic to fundamental, which relays and analysis tools compute from a sliding DFT.
Treat that ratio as supporting evidence, not proof. Transformer magnetising inrush also produces strong second harmonic, which is precisely why second-harmonic restraint exists in differential relays. Harmonic content tells you the current is distorted. It does not tell you why. Waveform shape and system context do.
What saturation does to each protection function
| Function | Effect of a saturating CT |
|---|---|
| Overcurrent (50/51) | Measured fundamental is lower than the true primary current, so pickup may be delayed or missed and IDMT timing runs long |
| Distance (21) | Apparent impedance Z = V/I inflates because the current denominator is understated, pushing the measurement out of zone and causing under-reach |
| Differential (87) | If CTs on one side saturate and the other side does not, a false differential current appears. A recognised cause of transformer and busbar differential misoperation on external faults |
| Directional (67) | Phase angle of the distorted current shifts, so the direction decision can be degraded, though voltage polarisation gives some immunity |
| Fault location | Reactance-method estimates depend on accurate current, so distance results are unreliable and should not be quoted from a saturated record |
The practical rule: if the CTs saturated, do not treat any current-derived number from that window as a measurement. Use the pre-saturation window, or state the limitation in the report.
Telling it apart from what it resembles
| Phenomenon | Distinguishing feature |
|---|---|
| CT saturation | Wave shape distorted with truncated peaks and an abrupt recovery edge. Severity grows through the fault |
| DC offset alone | Wave still sinusoidal with rounded peaks, displaced from zero, envelope decaying smoothly per the system X/R |
| Transformer inrush | Unipolar pulses separated by a gap of near-zero current, peaks rounded rather than clipped, present on energisation rather than during a fault |
| Overexcitation | Symmetric distortion on both half-cycles with odd harmonics dominant, no clipping plateau, driven by overvoltage or underfrequency |
DC offset and saturation appear together constantly, because the offset is what drives the core into saturation. The question to ask is not which one is present but whether the peaks are still rounded. Rounded peaks mean offset without saturation. Truncated peaks with a sharp recovery mean the core has gone.
A practical checklist
- Plot all three phase currents on a common scale through the fault.
- Compare the affected phase against the unaffected ones. Saturation rarely hits all three identically.
- Check whether peaks are rounded or truncated, and look for the abrupt recovery edge.
- Check whether distortion worsens cycle by cycle rather than staying constant.
- Check the voltage channels. Clean voltage with distorted current points at the CT.
- Compute the harmonic spectrum and note the second harmonic ratio as supporting evidence.
- Identify the last usable cycle and take your measurements from before it, or qualify the result.
Doing this on your own record
You can work through the checklist above on a real recording in Detego's COMTRADE viewer. Drop in a .cfg and .dat pair or a .cff file and the waveforms, harmonic spectrum and per-phase comparison are available without an install or an account. Parsing and computation run in the browser, so the file stays on your machine.
Detego AI also examines current distortion as part of its fault analysis, and will report when a measurement chain problem makes a record unreliable rather than analysing it as though it were clean. That distinction matters more than it sounds: a confident answer derived from a saturated CT is worse than no answer.
References
- IEEE C37.110, Guide for the Application of Current Transformers Used for Protective Relaying Purposes
- IEEE C37.91, Guide for Protecting Power Transformers
- IEC 61869-2, Instrument transformers: additional requirements for current transformers