What Happens When a Phase Is Lost

Why this matters

Losing one leg of a three-phase supply does not shut anything down, which is the whole problem. Motors that were already turning keep turning, and the equipment they drive keeps producing, so nobody calls until something burns. Meanwhile the conductor that opened is not dead: running motors feed voltage back onto it, and a tech who trusts a voltmeter reading rather than a proving sequence can take hold of a conductor that reads normal and is energized from the wrong direction.

Before anything else: the open conductor is energized

Treat every conductor in a suspected single-phasing event as live until it is proven dead with the load side isolated. The sequence that gets this right is: open and lock the supply disconnect under 29 CFR 1910.333(b)(2), then stop and lock out every downstream motor that could still be turning, under 29 CFR 1910.147 for the unexpected-startup and stored-rotational-energy hazard, then prove the conductors dead with the live-dead-live sequence at NFPA 70E-2021, 120.5, using a meter and leads rated CAT III or better at or above the voltage present. 29 CFR 1910.147 does not govern the electrical exposure itself, which it excludes at (a)(1)(ii)(C); the construction counterpart for lockout and tagging of circuits is 29 CFR 1926.417.

The order matters. Proving a conductor dead before the motors coast to a stop can give you a false dead reading during the coast-down and a live conductor thirty seconds later if anything restarts.

source open point load
  L1 ------------------------------------- motor T1
  L2 ------------------------------------- motor T2
  L3 --------------- X  ------------------ motor T3
                     |
    voltage found on the load side of the
    open comes from the motor still turning,
    not from the source

What happens to a motor that was already running

A three-phase induction motor that is turning when a line opens keeps turning. It is now being fed from two conductors, and it behaves like a single-phase machine whose rotor is already up to speed: the forward-rotating field component still dominates, so it produces torque, just less of it.

To hold the same shaft load on less effective voltage, the current in the two remaining lines rises. For a machine that was loaded near its rating, expect the remaining line currents to land in the region of 1.7 to 2 times the balanced value, with the exact figure set by how heavily it was loaded and by whether the winding is connected wye or delta. That range is a field expectation rather than a specification, and the manufacturer's data owns any specific machine.

Where the heat goes is the part that decides the outcome. With one line open, the current in the machine decomposes into nearly equal positive-sequence and negative-sequence components, and the negative-sequence component meets the rotor at close to twice line frequency. A sibling article covers that mechanism in full; the consequence here is that a large share of the extra loss lands in the rotor, which is generally not where a motor's internal thermal protection is watching. A line current sitting just under nameplate during single-phasing is not a safe condition, because the composition of that current is nothing like the balanced current the nameplate describes.

What happens to a motor that was stopped

It will not start. With one line open there is no rotating field at standstill, only a pulsating one, and a pulsating field produces zero net starting torque. The machine hums, draws locked-rotor current on the two remaining legs, and sits there until something opens.

That difference is a useful diagnostic in a building. Equipment that was running keeps running and gets hot; equipment that tried to start after the event hums and does not move. A site where one machine is loud and hot and a second one is buzzing and stationary is describing an open leg, not two coincidental faults.

Why the protection may not act

Three-leg overload protection on a starter generally does respond to single-phasing on a loaded motor, because the current rise is large. Three things regularly stop it:

• The machine was lightly loaded. A motor running at half its rating before the event sees the same rough multiple applied to a small number, and the result can sit under the trip threshold indefinitely while the rotor cooks.
• The protection is on fewer than three legs. Older starters with two overload elements can be blind to the specific leg that opened, depending on which one it was.
• The overload is protecting the conductors and the motor by current magnitude only. It has no way to see the composition of the current, so it does not know that 97 percent of nameplate current with a large negative-sequence component is not 97 percent of nameplate loading.

A dedicated phase-loss or phase-monitor relay is the device that watches for this condition directly, and it is worth having on any three-phase machine whose loss stops a process or is expensive to rewind. It stops the next event; it does nothing about the one you are standing in.

The call

A small production shop, mid-morning, three complaints in one visit: a rooftop unit that had tripped its overload twice, an exhaust fan that hums and does not turn, and a heater bank producing noticeably less than usual. Three symptoms, three different pieces of equipment, and the shop's first assumption was three separate faults on a bad day.

The rooftop unit's nameplate full-load current is 12.0 A and it had been running balanced at 10.4 A, which is 87 percent of its rating. The current on the two remaining legs measured 18.0 A, and 18.0 divided by 10.4 is 1.73, sitting right at the bottom of the 1.7 to 2 range stated above. Its overload, set to the 12.0 A nameplate, trips somewhere above that setting, so 18.0 A opened it, which is why this machine announced itself and the others did not.

The exhaust fan's nameplate is 3.4 A and it had been running lightly loaded at 1.9 A. On two legs it measured 3.3 A, a multiple of 1.74, essentially the same behaviour. But 3.3 A is under its 3.4 A nameplate and under its overload setting, so nothing opened and nothing ever would have. That motor was going to run until the rotor failed, and the only reason it stopped was that somebody switched it off and it could not restart.

Two machines, the same event, the same multiple, opposite protective outcomes. The one that tripped was the one that was working hard.

Reading it three ways

The voltage readings on this job are worth walking through in the order they were taken, because each one alone would have sent a tech somewhere different.

With the equipment running, line to line at the panel: 240 V, 231 V and 228 V. The two low readings are 96 percent and 95 percent of the healthy pair, which reads as a mild imbalance rather than an open conductor. That voltage is being manufactured by the motors that are still turning, feeding back onto the open leg through their own windings, and it is why "I have voltage on all three legs" is not evidence that all three legs are connected.

With every motor stopped but the panel still energized and the resistive heater bank still connected, the same two readings fell to 121 V and 119 V, roughly half of the 240 V healthy pair. With no rotating machine to regenerate it, the open conductor is simply being pulled toward the midpoint between the two live ones through the passive load, and half-voltage on two of the three pairs is the classic signature of an open conductor with something bridging it.

With the heater bank disconnected as well, the readings became unstable and meter-dependent, because nothing was connecting that conductor to anything and a high-input-impedance meter was reading a capacitively coupled phantom.

Three sets of readings on the same conductors, all correct, only one of them interpretable. The rule that falls out of it: take voltage readings on a suspected open leg with the rotating load stopped, and know what passive load is still connected while you read.

Confirming and correcting

Find the open before you replace anything downstream. The candidates run from the utility side inward: a utility primary fuse, a service fuse, a loose or burned lug at the service or a panel, a failed contact on one pole of a contactor or disconnect, and a broken conductor. On a fused disconnect, a single blown fuse is the most common single cause and the fastest thing to check, and it deserves the same question a tripped breaker deserves: it opened because of a condition, so replacing it without finding out why puts the same event back on the calendar.

Inspect the connections thermally where the equipment allows it. A high-resistance connection that is on its way to opening runs hot under load before it fails, and finding it during a load period is a different job from finding it after it opens. Where the enclosure has a rated viewing port, take that read with the covers on; removing a cover to see the joints turns a survey into an arc-flash exposure governed by 29 CFR 1910.333(a)(1), which permits energized exposure only where the employer can demonstrate that de-energizing introduces additional or increased hazards or is infeasible, with boundaries and PPE established on the basis in NFPA 70E-2021, 130.5 and 130.7, binding through your employer's electrical safety program or a contract rather than on its own.

Assess every motor that ran through the event before restoring service, not just the one that tripped. Insulation resistance and winding resistance leg to leg, taken with the machine isolated, locked out and proven dead, will identify a winding that has already been damaged. Motor shops report a characteristic pattern on machines that single-phased: on a wye-connected winding the damage typically shows in the two coil groups that were in series across the remaining line voltage, and on a delta-connected winding it typically concentrates in one group. Treat that as a strong indication when a shop reports it rather than as a test you can run yourself in the field.

Say plainly what the second machine's condition is. The lightly loaded fan in the case above ran with a heavily distorted current for an unknown period and its protection never saw a thing. It may be fine. It may have spent most of its insulation life. That is an honest uncertainty to hand the customer along with a recommendation to monitor it, and it is a much better position than the one you occupy after it fails three weeks later on a supply you certified as repaired.

References

• 29 CFR 1910.333(a)(1) and (b)(2) for energized work and de-energizing; 29 CFR 1910.147 for the unexpected-startup and stored-rotational-energy hazard of driven machinery, with the electrical carve-out at (a)(1)(ii)(C); 29 CFR 1926.417 for the construction counterpart
• NFPA 70E-2021, 120.5, 130.5 and 130.7, applied through an employer electrical safety program or contract
• Manufacturer motor and starter data for the specific machine, which owns its overload arrangement, its winding connection and its permissible loading
• See related: What a Phase Imbalance Does to a Motor; How to Measure and Interpret Phase Imbalance; Single-Phase and Three-Phase, and Why the Difference Matters