Why a Neutral Can Carry More Than You Expect

Why this matters

The neutral is the only current-carrying conductor in most buildings that is sized on an assumption about cancellation, carries no overcurrent device of its own, and can end up carrying more current than any of the phase conductors beside it in the same raceway. Nothing trips when it does. There is no protective device watching it, so the failure is thermal and slow: a neutral running hotter than its siblings, terminations discolouring at the panel, insulation aging out early, and eventually a fault in the worst-cooled section of the run. On the way there it produces neutral-to-ground voltage at the equipment, which is the thing that actually upsets sensitive loads and gets misdiagnosed as a grounding problem.

Before you touch a shared neutral

This one leads because the hazard is immediate rather than thermal.

A neutral shared by more than one circuit is dangerous to open while any of those circuits is energized. Break it and the loads on the two or three circuits end up in series across the line-to-line voltage, so a lightly loaded branch can be driven toward the full line-to-line value, up to about 208 V on a 208/120 wye and up to 240 V on a 120/240 single-phase system, against equipment rated 120 and the "neutral" you are holding becomes an energized conductor at whatever the load divider produces. This is why the adopted NEC requires simultaneous disconnection of all ungrounded conductors of a multiwire branch circuit at the point of supply, in Article 210.

So: de-energize EVERY circuit sharing that neutral before you land, lift or torque it, not just the one you are working on. 29 CFR 1910.333(a)(1) requires the circuit de-energized before an employee works on or near exposed live parts unless the employer can demonstrate that de-energizing introduces additional or increased hazards or is infeasible due to equipment design or operational limitations, and neutral work is never the case that qualifies. Lock and tag under 29 CFR 1910.333(b)(2), the standard that owns electrical work because 1910.147 excludes exposure to electrical hazards from work on conductors and equipment in electric utilization installations at (a)(1)(ii)(C), with 29 CFR 1926.417 as the construction counterpart. Prove the neutral dead live-dead-live per NFPA 70E-2021 120.5 - the neutral specifically, not just the phases - in the edition your employer's electrical safety program or your authority having jurisdiction has adopted.

What the neutral was sized on

On a three-phase, four-wire wye system, the three phase currents are 120 degrees apart. With balanced, linear loads they sum to zero at the neutral point, so the neutral carries nothing. That cancellation is the entire basis for the common practice of running a neutral the same size as the phases and never thinking about it again.

   wye source loads
   ----------- A ------------------> load A
   ----------- B ------------------> load B
   ----------- C ------------------> load C
        |                               |
        +---------- neutral ------------+

   fundamental currents in A, B and C are
   120 degrees apart and cancel in the neutral
   third-harmonic currents are in step with
   one another and add in the neutral instead

Two things break the cancellation, and only one of them is the one most techs were taught.

Break one: unbalance, which has a ceiling

With linear loads that are simply unequal, the neutral carries the vector sum of what does not cancel. This is real and it is worth measuring, but it has a bound worth carrying: the worst case is one phase loaded and the other two at zero, in which case the neutral carries exactly what that one phase carries. For linear loads at broadly similar power factor, which is the ordinary building case, the neutral current cannot exceed the largest phase current. That bound is why a same-size neutral is a defensible default for linear systems, and it is the assumption everything else on this card violates.

Break two: triplen harmonics, which have no such ceiling

Loads with rectifier front ends - switch-mode power supplies, LED drivers, electronic ballasts, small drives, most modern office and control equipment - draw current in pulses near the voltage peaks rather than as a sinusoid. That non-sinusoidal current contains harmonics, and the harmonics-defined card owns what a harmonic is. The part that matters here is the SEQUENCE.

The third harmonic, and its odd multiples the ninth and fifteenth, are zero-sequence: the third-harmonic component of phase A, phase B and phase C are all in step with each other, not 120 degrees apart. Being in step, they do not cancel at the neutral point. They add arithmetically. Balanced loading, which is what protects the neutral at the fundamental, does nothing at all for the triplens, and that inversion is the whole finding: the more evenly you distribute nonlinear single-phase load across the three phases, the better the fundamental cancels and the more triplen current you have collected in one conductor.

Take a system carrying only a fundamental and a third harmonic, balanced across the three phases, and call the third-harmonic current a fraction of the fundamental. Each phase conductor carries the root-sum-square of the two, and the neutral carries three times the third alone. The ratio of neutral current to phase current is therefore three times that fraction, divided by the square root of one plus the fraction squared. At a fraction of 0.707, that expression equals 1.73, so the neutral carries about 1.73 times what each phase carries. It keeps climbing beyond that with higher harmonic content, toward three times for a load that is nearly all third harmonic and nothing else.

That relationship is derived for balanced phases, a fundamental and a third only, and third-harmonic components perfectly in step. Real installations come in lower, because a population of different devices does not have its harmonics in perfect step, so treat the calculation as an upper bound and settle the real number with a clamp meter.

Worked case: three circuits and one hot neutral

Three 120 V branch circuits fed from a 208/120 V wye panel, one on each phase, sharing a neutral in the same raceway. Each circuit serves a bank of electronic equipment and each measures 16.0 A with a true-RMS clamp. The load is balanced across the phases, so at the fundamental this is the textbook cancellation case and the neutral should read close to nothing.

Assume, as an illustrative value, third-harmonic content at 60 percent of the fundamental on each circuit.

Split the phase current into its parts. Each phase carries the root-sum-square of the fundamental and the third, so 16.0 equals the fundamental times the square root of one plus 0.60 squared. The square root of 1.36 is 1.166, so the fundamental component is 16.0 divided by 1.166, which is 13.72 A. The third-harmonic component is 0.60 times 13.72, which is 8.23 A.

Add the triplens in the neutral. Three times 8.23 is 24.7 A.

Read it against the phases. The neutral carries 24.7 A while each phase carries 16.0 A, a ratio of 1.54. Check it against the relationship above: three times 0.60 is 1.8, divided by 1.166 is 1.544. It closes.

Now read what that means for the installation. The neutral is the same size as the phase conductors, sitting in the same raceway, carrying 54 percent more current than any of them, with no overcurrent device anywhere on it. Nothing in the panel will ever report this. The branch breakers see 16.0 A each and are content.

And the raceway got worse at the same time. Because the major portion of the load is nonlinear, that neutral counts as a current-carrying conductor for adjustment purposes, so the raceway went from three current-carrying conductors to four and picked up the 80 percent adjustment step (the ambient-and-bundling card works the counting rules). The conductor that gained the most current is the same conductor that pushed the whole raceway into a derating step, and neither event generated a work order.

What the direction check would have caught. Had the same three circuits been serving linear load - resistance heat, incandescent lighting, motors - balanced at 16.0 A each, the neutral would have read near zero rather than 24.7 A. Same wires, same currents on the phases, opposite reading on the neutral. If a neutral reading surprises you, the load type is the first thing to establish, not the connections.

What the code already assumes about this

Two provisions in the NEC edition your authority having jurisdiction has adopted exist specifically because of the above, and they are worth knowing as evidence that the effect is real rather than exotic.

Article 220 permits a reduced neutral on some feeders, and explicitly does NOT permit that reduction for the portion of the load that consists of nonlinear line-to-neutral loads. Article 310 requires the neutral to be counted as a current-carrying conductor for adjustment where the major portion of the load is nonlinear. If you find a reduced neutral on a feeder serving electronic load, you have found something that was sized under an assumption that no longer describes the building, and that is a design question for the engineer of record rather than a field correction.

Where the current goes after the neutral

The triplens do not stop at the panel. On a delta-wye transformer, zero-sequence current arriving at the wye neutral circulates in the delta primary winding rather than passing upstream, which is a feature (it keeps the harmonics off the utility) and a cost (that circulating current heats the winding with no corresponding load on the secondary). The system-effects card owns the transformer side; the reason it belongs in a technician's head is that an overheating transformer feeding a panel of electronic load is often reporting this and not reporting an overload.

How to verify you got this right

Clamp the neutral with a true-RMS instrument, not an averaging one. An average-responding, RMS-calibrated clamp reads a distorted waveform low, and reading a harmonic problem with a meter that cannot see harmonics is the most common way this gets missed. Take the reading under the energized-work gate above, with boundaries and PPE selected under NFPA 70E-2021 130.5 and 130.7 as adopted.

Compare the neutral against the three phases in the same minute. The two useful outcomes: a neutral well below the phases is a normally balanced linear system; a neutral at or above the largest phase current is triplen-dominated by definition, because unbalance alone cannot produce it. That single comparison separates the two causes without a power quality analyser.

Read neutral-to-ground voltage at the far end of the run, at the equipment, not at the panel. The neutral's own drop under this current is what shows up there, and it is what sensitive equipment reacts to. A neutral-to-ground reading that climbs when the electronic load comes on, and falls when it goes off, has told you where the current is and roughly how much conductor it is crossing.

Feel nothing with your hands. Check conductor and termination temperature with a non-contact infrared instrument from outside the restricted approach boundary, or through an infrared window if the gear has one.

References

• NEC Article 210 (multiwire branch circuits and simultaneous disconnection), Article 220 (neutral feeder load and the nonlinear-load restriction on reduction) and Article 310 (counting the neutral as a current-carrying conductor), in the edition your authority having jurisdiction has adopted
• 29 CFR 1910.333(a)(1) and 1910.333(b)(2), with the 1910.147(a)(1)(ii)(C) exclusion; 29 CFR 1926.417 for construction
• NFPA 70E-2021, 120.5, 130.5 and 130.7, as adopted by your employer's electrical safety program or your authority having jurisdiction
• See related: What a Harmonic Actually Is; What Non-Linear Loads Do to a System; Why Neutral and Ground Are Not the Same