Negative sequence

What it is (precise)

The symmetrical-component decomposition — the positive-sequence entry works through the frame — resolves any three phasors into three balanced sets, and negative sequence is the one with reversed phase order: equal magnitudes, 120° apart, with phase b leading phase a instead of lagging it. The physical consequence follows from the winding geometry.

Feed a three-phase winding a set of currents whose time order is reversed while the spatial order of the windings stays fixed, and the resultant air-gap MMF wave rotates at synchronous speed in the direction opposite to the normal field. A perfectly balanced a-b-c system contains positive sequence only; any departure in magnitude or angle puts some negative sequence into the mix.

What separates it from the third component is the return path. The three phasors of a negative-sequence set sum to zero, so the component flows freely in any three-wire circuit — no neutral, no earth connection required. Zero sequence is the opposite case: it needs a fourth path, which is why unbalance in a delta-connected or ungrounded three-wire circuit maps entirely into negative sequence with no zero-sequence current at all, and why the two components answer different design questions — the zero-sequence entry owns the ground-path story.

The steady-state measurement of unbalance, including the IEC-style ratio of negative- to positive-sequence voltage, belongs to the phase-unbalance entry; this page follows the component itself, from what creates it to what it destroys and what protection builds on it.

What creates it

In normal operation, anything that loads the three phases unevenly: single-phase loads spread unequally across the phases, traction supplies, arc furnaces, and the asymmetric per-phase impedances of untransposed overhead lines. Series unbalances make it too — a broken conductor, or a breaker with one stuck pole, unbalances the circuit without drawing any extra current, which is exactly what makes them hard to detect by other means.

Faults are the strong sources, and the fault types rank cleanly. A balanced three-phase fault produces no negative sequence at all — every ampere is positive sequence. A line-to-line fault is the richest: the boundary conditions force the negative-sequence current to equal the positive-sequence current in magnitude, |I₂| = |I₁|.

A single-line-to-ground fault splits its current into equal thirds, I₁ = I₂ = I₀, and a double-line-to-ground fault produces all three components. Negative sequence is therefore the one component every unsymmetrical fault contains, whether or not ground is involved — zero sequence appears only where a ground path exists, which is why protection built on I₂ covers fault types that residual ground elements can never see.

Why rotors pay: the double-frequency sweep

In a synchronous machine the negative-sequence field rotates backward at synchronous speed while the rotor turns forward at synchronous speed, so the field sweeps past the rotor at twice synchronous speed and induces rotor-body currents at exactly twice line frequency — 100 Hz on a 50 Hz system, 120 Hz on 60 Hz — together with a double-frequency pulsating torque that fatigues shafts and couplings, and an average braking torque.

In an induction motor, whose rotor runs just below synchronous speed, the induced frequency lands just under twice line frequency — the near-twice figure the phase-unbalance entry quotes, along with the motor-derating consequences.

The heating concentrates where the machine can least afford it. Skin effect at double frequency confines the induced currents to thin surface layers of the rotor forging, the slot wedges and the retaining-ring contact zones — paths of high effective resistance at mechanically critical parts, so the damage mechanism is thermal-mechanical rather than a bulk temperature rise. The machine's negative-sequence reactance is small, roughly the average of the two subtransient reactances, so even a modest negative-sequence voltage drives significant current.

Capability is specified two ways: a continuous I₂ limit of roughly 5-10% of rated current, and a short-time withstand (I₂,pu)²·t ≤ K, with K around 5-10 for large directly cooled cylindrical rotors and up to around 30-40 for salient-pole machines. The direction is worth memorising: the larger and more highly utilized the machine, the less negative sequence it withstands — the short-time process is near-adiabatic, the same logic as I²t fuse damage.

A BESS plant meets this physics from the supply side. Its own rotating machines are the thermal-management fleet — compressors, pumps, fans — which inherit whatever standing unbalance sits on the auxiliary supply, and any synchronous plant sharing the network inherits the negative sequence the wider system produces during faults and unbalanced operation.

The protection built on it

Device 46 in the ANSI/IEEE C37.2 numbering — the negative-sequence overcurrent relay the phase-sequence entry names — exploits the defining property: balanced load produces no I₂, so load current simply vanishes from the measured quantity.

The pickup can therefore sit at roughly 10-20% of rating, far below full load, making the element a sensitive backup for phase-to-phase faults, where the negative-sequence current equals the positive-sequence current in magnitude. On generators the element does thermal accounting rather than simple pickup: it integrates (I₂,pu)²·t against the machine's K and must operate below the capability curve, tripping before the rotor's withstand is spent.

The ratio of negative- to positive-sequence current adds a second trick: broken-conductor detection. A series open phase produces a large I₂/I₁ ratio with little or no overcurrent — one phase open feeding a balanced load on a three-wire circuit gives a ratio of exactly 100%, a grounded four-wire load about 50% — and because I₂ and I₁ both scale with load, the ratio is load-independent, so the element works even on lightly loaded lines, sitting far above the roughly 2-10% standing unbalance of a healthy feeder; typical pickups sit near 20%.

Directional and phase-selection logic in modern line protection lean on the same component, preferred over zero-sequence quantities partly because the strong mutual coupling of parallel lines that corrupts zero-sequence measurements barely touches negative sequence.

Inverters, and what IEEE 2800 asks of them

A synchronous machine feeds a bolted fault with subtransient current of roughly 3-8 times rated — trapped flux and thermal mass decide, and physics answers the same way every time.

A PCS is limited by semiconductor thermal ratings, with millisecond thermal time constants and no overload mass, to roughly 1.0-1.3 pu — up to about 1.5 pu in some designs — and its fault current's magnitude, phase and sequence content are control decisions rather than physical responses. That order-of-magnitude gap is what breaks classical overcurrent grading in inverter-dominated systems, and the sequence content is where it bites protection hardest.

A legacy grid-following inverter controlled in a balanced positive-sequence frame regulates its negative-sequence current toward zero even during unbalanced faults — which makes it a very high negative-sequence impedance, close to an open circuit in the negative-sequence network. Every protection scheme in the previous section is starved by that behaviour: device-46 elements, phase selection and negative-sequence directional logic all lose the signature they were built on.

IEEE 2800-2022, the US standard for transmission-connected inverter-based resources, therefore requires the injection of negative-sequence reactive current during unbalanced faults, proportional to the negative-sequence terminal voltage — turning the plant into a low, predominantly inductive negative-sequence impedance, the way a machine's X₂ behaves, so the fault signature protection expects reappears. Germany's VDE-AR-N 4110/4120 impose the analogous requirement through a negative-sequence k-factor; IEEE 1547-2018, governing US distribution-connected DER, contains no such requirement.

Two traps live in the implementation. The phase requirement is convention-dependent — an angle quoted without stating whether current is measured into or out of the resource can be implemented backwards, which would emulate a capacitive element and amplify the unbalance instead of damping it. And the injected I₂ shares the same total-current ceiling as the positive-sequence current, so the standards also define the priority split between them during the fault.