Zero sequence

What it is (precise)

Any unbalanced set of three-phase currents or voltages can be rewritten, exactly, as the sum of three balanced sets. The positive sequence is the one the plant is designed around: three equal phasors displaced 120°, rotating in the normal phase order. The negative sequence is its mirror — equal magnitudes, 120° apart, rotating in the reverse order. The zero sequence is the odd one out: three phasors of equal magnitude with no displacement between them at all.

Its value is one third of the phasor sum, I₀ = (Ia + Ib + Ic)/3, which makes two things immediate: a perfectly balanced set contains none at the fundamental, and anything that breaks the cancellation — a ground fault, a single-phase auxiliary load landing on one phase, a saturated transformer's triplen magnetizing current arriving in phase on all three — creates some.

The reason a storage engineer meets the decomposition is the short-circuit and ground-fault study. The study models the plant three times over — a positive-, a negative- and a zero-sequence network — and the first two look like the one-line diagram with impedances written on it. The zero-sequence network does not.

Because in-phase current only flows where a neutral or earth connection closes the loop, that network is drawn from winding connections and neutral treatments rather than from ratings: a through-path where a grounded neutral exists, an open circuit where none does, a connection to the reference bus inside a delta. Two transformers with identical MVA, impedance and ratio but different letters on the nameplate produce different zero-sequence networks — the first hint that the vector group is not a detail.

The ground-fault component

The most common fault on a collection system is one conductor to ground, and its arithmetic runs entirely through the zero sequence. For a single-line-to-ground fault the three sequence networks connect in series, and the current in the faulted phase is 3·I₀ — so the zero-sequence impedance between the fault and its ground sources decides how many amps flow.

That is also the quantity protection measures: a ground relay reads the residual sum of the three phase CTs, or a core-balance CT around all three conductors together, and both see exactly 3·I₀ while staying blind to balanced load current. Ground-fault protection on the MV collection system is therefore a zero-sequence design problem — the settings, the CT arrangement and the expected fault level all come out of the zero-sequence network, not the load-flow model.

The same network sets what happens to the phases that did not fault. On a bus with a low-impedance zero-sequence path, a ground fault draws heavy current, relays clear it fast, and the healthy phases barely move. On a bus with no path at all, the fault current is nearly nothing — and the two healthy phases rise toward line-to-line voltage, √3 times their normal value, stressing every meter of cable insulation and every surge arrester until the fault is found.

Neither behaviour is wrong in itself; unmanaged is wrong. Interconnection requirements commonly demand an effectively grounded system, arrester ratings and cable insulation levels assume one answer or the other, and the answer is created — or lost — by the transformer connections the project specifies.

The triplen-harmonic component

Zero sequence is not only a fault quantity; a whole class of harmonics is made of it in normal operation. The third harmonic sits 3 × 120° = 360° apart across the phases — in phase by construction, a full turn — so it and every odd multiple of three (3rd, 9th, 15th…) are zero sequence by definition.

Where balanced fundamental currents cancel in a shared neutral, triplens add — three conductors' worth in one wire — which is how a neutral overheats on a system whose phase currents all read within rating. The neutral-current entry follows those amps; the point here is which component of the spectrum they are.

For a storage plant the triplen sources are the single-phase auxiliary electronics and anything saturating magnetically — a transformer energized near the knee of its magnetizing curve generates them too; the PCS itself, a balanced three-wire converter, cannot drive triplen line current and contributes only through imperfection: unbalance, dead-time distortion, or a provided neutral path.

The design consequence sits in the winding letters: a delta traps zero-sequence current as a circulating current inside the closed winding, so a step-up transformer's delta keeps triplen content on its own side instead of exporting it to the medium-voltage system — one of the standing reasons the connection scheme includes a delta at all.

What does cross to the point of interconnection is what the harmonics limits in the connection agreement are written against, so the vector group is doing power-quality work before any filter is specified. The total-harmonic-distortion entry covers the compliance metric; the transformer connection decides which part of the spectrum ever reaches the meter.

Which paths pass it, block it, trap it

Three rules cover the hardware. A grounded wye passes zero sequence: each winding runs from a line terminal to the neutral, the neutral is tied to earth, so in-phase current flows through the windings and returns through the ground connection. An ungrounded wye blocks it: the same windings, but the neutral floats, so there is no return path and no zero-sequence line current at all.

A delta traps it: the closed loop of windings gives in-phase current a path to circulate around, but there is no neutral for it to leave through, so nothing reaches the lines. Blocking and trapping look identical from outside — no zero-sequence current in the delta-side lines — but the trap is doing work: the circulating current is real amperes in real copper, and it is exactly what lets the delta act as a sink.

In the zero-sequence network, then, a transformer is an open circuit at a delta or ungrounded-wye winding and a sink at a grounded-wye winding — the framing the transformer-vector-group entry carries in full, with the IEC and EPRI sourcing behind it. The sink behaviour is useful enough that IEC names a winding after it: the stabilizing winding, a supplementary delta added to a star-star transformer for the stated purpose of decreasing its zero-sequence impedance.

The same logic runs all the way to the inverter terminals. A PCS connects to its transformer through three wires, and vendor rules such as SMA's — where a low-voltage neutral terminal exists, it must not be grounded — deliberately leave the inverter-side system with no zero-sequence path, so ground-fault current on the collection system has one less branch for the study to trace.

How it shows up in specs, studies and contracts

The zero-sequence decisions on a project hide inside two documents. On the single-line diagram: the vector group of every transformer and the earthing of every neutral, because together they are the zero-sequence network. A YNd11 step-up grounds the collection system directly through its MV wye; a Dyn11 unit puts the delta on the MV side and forces a separate grounding transformer onto the bus to restore the earth reference the winding choice removed.

In the grounding study: whether the resulting network is effectively grounded — the classical IEEE C62.92-family criterion caps X0/X1 at 3 and R0/X1 at 1, both positive, a specification written directly on the zero-sequence impedances the winding letters created. The grounded-wye entry owns the winding decision itself; the working point here is that the study and the nameplate letters are the same choice seen twice, so changing one without re-running the other is how grounding schemes go wrong on paper.

Two traps connect the concept to protection and to data requests. The delta zero-sequence trap is also a measurement blind spot: for a ground fault on the wye side of a transformer, no zero-sequence current reaches CTs on the delta side, so a relay there cannot see the fault — backup and reach schemes have to respect where the letters cut the network. And zero-sequence impedance is its own datasheet line: for cables and overhead lines the return path runs through earth, sheaths and ground wires rather than the other phase conductors, so Z0 differs from Z1 and cannot be estimated from it.

A collection-system ground-fault study needs zero-sequence data requested explicitly — every cable, every transformer, every neutral resistor — or it gets built on placeholders. The commissioning echo is short: the grounding scheme declared on the SLD is verifiable with a meter, and it should be verified before the first ground fault runs the test for you.