Investigating a 13.2 kV Motor Lug Failure and Trip

The 13.2 kV motor tripped because a localized T3 connection failure became an arcing fault during acceleration, not because the winding tests had necessarily missed a bulk insulation defect. The melted C-phase lug explains the heat source; the 51G and A-phase high-dropout flags describe the fault currents seen by separate protection elements. The pump rub could have prolonged high starting current and added thermal stress, but it could not melt a sound lug by itself.

Trip sequence and operating quantities

The number that matters is the current through the defective joint and how long that current persisted. The motor was rated 3,500 hp, 13.2 kV, 324 rpm, and 60 Hz, with 153 A full-load current and a recorded locked-rotor current of 989 A. It tripped 2.5 seconds after energization.

The protection CT ratio was 40:1. Therefore, the reported A-phase HDO flag means the relay saw more than 30 A secondary, equivalent to 1,200 A primary, for longer than 0.1 seconds. It does not identify the physical faulted connection, and it does not prove that C-phase current exceeded the same threshold.

Lug heating and arc development

A crimp with excessive resistance generates heat according to P = I²R. At 989 A, a fixed connection resistance produces about 41.8 times the heat generated at 153 A because (989/153)² ≈ 41.8. This is heat, not logic: a joint that survives normal load can deteriorate rapidly during a start.

The T-leads were 2 AWG with 1/0 lugs installed using a hydraulic crimper. The connection was considered compatible with the lug manufacturer's recommendations, but the actual conductor range, lug catalog data, die index, crimp count, strand preparation, and completed-crimp dimensions must be checked. A nominally permitted conductor-to-lug combination can still have low barrel fill, an incorrect die, incomplete compression, damaged strands, contamination, or mechanical loading.

The T3 barrel was melted beneath apparently intact Raychem tape. That appearance is plausible. The heat and arc can be concentrated inside the wrapped joint, while the outer insulation retains its general shape or develops only a small puncture. Soot and molten copper can then escape through a pinhole and deposit elsewhere in the terminal box. The black residue near T5, T1, and the B-phase coupling-capacitor jumper therefore does not locate the initiating fault by itself.

The likely sequence is a resistive T3 joint, rapid I²R heating during acceleration, loss of conductor continuity or an arc inside the wrapping, and a subsequent phase-to-ground or capacitive ground-current event. Separate metallurgical examination is needed to distinguish poor compression, oxidation, fatigue, strand damage, and arc erosion.

Relay flag interpretation

An open C-phase connection during acceleration changes the motor's sequence currents and redistributes current in the remaining energized circuit. Depending on winding connection and the location at which T3 opened, A-phase current can rise above the normal locked-rotor value. The A-phase HDO flag establishes only that its measured current crossed 1,200 A for the relay's operating interval.

The absence of an instantaneous flag places a useful boundary on the event: the measured currents either stayed below the 2,000 A instantaneous pickup, failed to remain within its operating region long enough, or were not captured by that element. Relay calibration and target operation should be checked before using the absence of a flag as a precise current limit.

The 51G element was supplied by a window CT enclosing all three phase conductors. Such a CT measures residual current, approximately IA + IB + IC. Ordinary phase-current imbalance or negative-sequence current still returns through the phase conductors and does not, by itself, create residual current. A 51G operation points instead to current returning outside the CT window, relay or CT circuit error, transient spill, or unbalanced phase-to-ground capacitive current.

An arc from T3 to grounded metal is the most direct explanation. Unequal current through grounded surge capacitors could also contribute if the terminal voltages became unbalanced, but that path must be evaluated from the capacitor connections and ratings. The negative-sequence device, a GE 12KJC51E2A set to slope 125% and time dial 4, did not provide a reported flag; its curve and reset behavior determine whether the short event should have operated it.

Pump-load relationship

The later start showed current falling from locked-rotor current to 140 A, then alternating between 140 A and 190 A over several minutes while abnormal noise was present. Inspection found the impeller rubbing the bowl after anti-rotation pins loosened outside the bowl. That mechanical defect explains the post-start load oscillation.

A rubbing pump increases accelerating torque demand and can lengthen the interval at high current. It can therefore accelerate the failure of an already resistive lug. It does not create intense heat at one lug unless that connection already has materially greater resistance than the other phase connections.

The missing time-overcurrent indication does not exclude pump involvement. The trip occurred after 2.5 seconds, while the time-overcurrent element had a 200 A pickup and time dial 2. Its operating time must be read from the applicable current-time curve at the actual current. A faster HDO or ground event can clear the motor before the time-overcurrent element reaches its trip point.

Inspection and repair procedure

1. Isolate the motor and apply the site's medium-voltage work controls. Do not re-energize with an unidentified melted connection, conductive residue, or damaged insulation in the terminal box.
2. Photograph the terminal box before cleaning. Record the positions of T1 through T6, soot patterns, arc marks, insulation punctures, tie locations, coupling-capacitor jumpers, and the panel pock-mark.
3. Preserve the failed T3 lug, loose copper, tape layers, and a clearly labeled sister connection. Avoid cleaning or cutting the specimens before the laboratory defines its examination sequence.
4. Inspect every termination, including T4, T5, and T6, for barrel deformation, exposed strands, discoloration, cracked insulation, strand withdrawal, and mechanical strain. T2 already showed two or three apparently broken strands and deserves specific examination.
5. Verify the 2 AWG conductor and 1/0 lug combination against the exact lug manufacturer's data. Record lug identification, conductor construction, die code, tool calibration status, crimp locations, and required number of compressions.
6. Measure each connection directly at the motor using a four-wire low-resistance method. Separate cable resistance from joint resistance by placing potential leads immediately across each connection, and compare equivalent phase joints at a common temperature.
7. Inspect the failed and sister lugs using radiography or computed imaging before sectioning, followed by metallographic cross-sections where justified. Look for voids, incomplete strand consolidation, incorrect compression geometry, oxide layers, fatigue, local melting, and the boundary between thermal damage and arc damage.
8. Replace suspect lugs using a controlled preparation and crimp process. In the recorded repair, T1, T2, and T3 were replaced, residue was removed, and the motor was reterminated and retaped; the remaining similarly installed joints still require a documented disposition.

Electrical and mechanical verification

After the repair, the motor passed polarization-index testing, insulation-resistance testing above 5,000 MΩ, a bridge test, and a DC step-voltage test to 24 kV DC without nonlinear behavior. These results support acceptable post-repair phase-to-ground insulation condition. They neither reconstruct the pre-repair fault nor test a phase-to-phase path when only phase-to-ground connections were exercised.

Verification should combine static tests with a captured start:

1. Repeat direct phase-resistance and joint-resistance measurements at the motor. Compare phases and retain the raw readings, test current, lead locations, and conductor temperature.
2. Check the IAC66M3A, IAC77A, and device 46 circuits, including CT ratios, polarity, secondary continuity, pickup calibration, targets, and reset condition.
3. Capture all three phase currents and residual current with enough time resolution to resolve the first 2.5 seconds. Record terminal voltage and speed when suitable instrumentation is available.
4. Confirm that current decays from the 989 A locked-rotor region toward the expected running value without returning repeatedly to 190 A. Investigate any current oscillation together with vibration, noise, pump loading, and speed.
5. Verify the pump repair, including the anti-rotation pins and impeller-to-bowl condition, before declaring the motor event closed.
6. After stable operation, inspect accessible termination temperatures using the site's approved medium-voltage thermography method. Compare phases under similar current rather than judging one absolute surface temperature through different insulation thicknesses.

Recurring diagnostic pitfalls

Frequently asked questions