Neutral current

What it is (precise)

In a four-wire wye, each phase current flows out on its own conductor and the neutral returns the vector sum of all three. Write the balanced set — three equal currents 120° apart — and that sum is zero at every instant, which is why the single-phase-vs-three-phase entry can state flatly that a perfectly balanced four-wire wye carries no neutral current at all.

Unbalance breaks the cancellation in proportion to itself: hang a single-phase load on one phase with the other two idle and the neutral returns that load's full current, every amp of it. The general case is the phasor residue of the three phase currents, and where that residue comes from — unequal aux loads, single-phase equipment landed unevenly across phases — is the phase-unbalance entry's territory.

Harmonics break the cancellation by a different mechanism, and only some of them do. Shift a third-harmonic current by the fundamental's 120° and it moves 3 × 120° = 360° — a full turn, back where it started. So the third-harmonic currents of the three phases arrive at the neutral in phase and add arithmetically: the neutral carries three times the per-phase triplen content. The same holds for every multiple of three — 3rd, 9th, 15th — while the 5th, 7th, 11th and 13th keep their displacement and cancel like the fundamental.

The consequence is stark once you work it through: with each phase carrying third-harmonic content at 30% of its fundamental, the neutral carries 90% of a phase's fundamental current, and above roughly 35% third-harmonic content the neutral carries more current than any phase conductor. Why electronic loads produce that spectrum is the harmonics entry's subject; what matters here is where those amps end up.

The compact way to book all of this is symmetrical components: neutral current is three times the zero-sequence current, I(N) = 3 × I0, because the neutral is the only path zero-sequence current has. That component view — and what blocks it, notably delta transformer windings — belongs to the zero-sequence entry; this page stays with the amps in the wire.

Where the neutral actually is in a BESS plant

Walk the one-line and look for a fourth wire. The MV collection system is commonly three-phase three-wire — no neutral conductor at all, as the single-phase-vs-three-phase entry warns anyone hoping to tap one. The PCS-to-transformer LV loop is a three-wire power circuit in most storage architectures.

The working neutral in a BESS plant lives in one place: the low-voltage auxiliary distribution — the station-service or auxiliary transformer's wye secondary, typically 230/400 V, feeding the boards that keep the plant alive. That is where HVAC and thermal-management controls, battery management supplies, UPS units, lighting, heaters, fire-detection and communications panels all connect, and most of them connect line-to-neutral.

The aux board is also the worst-case load mix a neutral can face, because it stacks both failure modes of the cancellation. The loads are single-phase, so however carefully the contractor spreads them across phases, some standing residue is inevitable — and the fire-alarm and comms panels a different discipline connects last were never in anyone's balance calculation.

And the loads are electronic: control cabinets, chargers, servers and UPS front ends are switched-mode rectifier supplies whose input current is rich in exactly the third-harmonic content that adds in the neutral. The large aux motors — liquid-cooling pumps, chiller compressors, HVAC fans — mostly run on three-phase drives and keep their distortion off the neutral; it is the accumulation of small single-phase electronics, multiplied across every container and cabinet on a multi-hundred-MWh site, that loads the fourth wire.

Sizing the fourth wire

The habit of sizing the neutral smaller than the phases descends from an assumption — balanced, linear load — that an auxiliary board full of rectifier supplies does not satisfy. The triplen arithmetic above is the reason: the neutral can carry as much current as a phase conductor, or more, while every phase sits comfortably inside its rating.

So the defensible default for a BESS aux board is a neutral at full phase-conductor size, and anything smaller needs a justification written down — a measured or calculated harmonic spectrum of the actual load mix, not the observation that the phase currents balance.

The neutral is also the conductor the protection scheme cannot see. Overcurrent devices sit in the phases; a neutral overloaded by triplen current trips nothing, because no phase is overloaded. Nor is the neutral fused or switched on its own — opening it while the phases stay live is worse than the overload, as the next paragraph shows.

The heat has to go somewhere upstream too: triplen current crossing the aux transformer's wye winding circulates in its delta winding — the blocking behaviour the transformer entry describes — and that circulation is extra winding heat the transformer's loading calculation should include.

The failure mode worth respecting is the broken neutral. Lose the neutral — a loose lug, a corroded joint, a mis-pulled conductor — and the star point of the load floats: the 230 V line-to-neutral voltages redistribute according to the load impedances on each phase, and the lightly loaded phase climbs toward the 400 V line-to-line value.

The casualties are the plant's most sensitive equipment, since it is precisely the 230 V controls, BMS supplies and UPS inputs that live on that board. The bess.engineer floating-neutral visual animates the mechanism. Practical consequence: the neutral lug belongs on the same torque-check and thermographic-survey schedule as the phase connections, because it works as hard as any of them.

The measurement habit

Commissioning should record four currents per aux board, not three. Clamp the neutral alongside the phases and write it into the test record, because the number is diagnostic on its own: a neutral current close to what the phase-current imbalance predicts is an unbalance problem, fixed by moving loads between phases — which is why the auxiliary schedule should arrive with phase assignments on it, not just a kVA total.

A neutral current well above what the imbalance explains is harmonic, and the fix is a spectrum measurement and a look at which electronic loads dominate, not another round of load-shuffling that cannot touch it.

Two habits keep the measurement honest through operations. Use a true-RMS clamp meter: an average-responding meter is calibrated for sine waves and misreads the peaked, distorted waveform a triplen-rich neutral carries. And trend the reading, because the aux system is the part of the plant that changes — heaters added after the first winter, HVAC replaced, control gear swapped at a retrofit — and every change lands on a conductor with no alarm on it.

A plant sitting idle is also the most single-phase it ever gets: with the PCS fleet at zero, the aux board is the only load running, so standby is when the neutral fraction peaks, not the full-power hour the studies model.

Common pitfalls

The first trap is letting the plant's headline balance stand in for the aux board's. The power train is a balanced three-phase machine; the 230/400 V board behind the station-service transformer is a collection of single-phase electronics, and the cancellation that protects the one does not protect the other. The related error is reading three comfortable phase ammeters as proof the board is fine — the whole point of the triplen mechanism is that the neutral can be the most heavily loaded conductor on the board while the phase readings show nothing unusual.

The second trap is conflating the neutral with the protective-earth conductor. They meet at exactly one bonding point, and their jobs are opposite: the neutral carries load current every hour of the plant's life, the PE conductor carries current only during a fault. A neutral-to-earth connection made anywhere downstream of the bond puts a share of that continuous load current onto the earthing system — a standing error that corrupts ground-fault measurement and RCD behaviour, and the kind of defect a commissioning clamp reading on the PE conductor catches in minutes.