By Jay Prigmore II, Ph.D., P.E. Last reviewed · Prepared and reviewed by a licensed Professional Engineer
Investigation of medium- and low-voltage switchgear failures, breakers that failed to interrupt, and protection that operated when it should not have, or did not operate when it should. The relay usually recorded the answer. The work is reading that record correctly and making the physical evidence agree with it.
By Jay Prigmore II, Ph.D., P.E. Last reviewed · Prepared and reviewed by a licensed Professional Engineer
Correct protection operation limits the damage; it does not explain the initiation. A matter in which “the relay worked” often becomes a design, manufacturing, or maintenance dispute rather than ending.
Modern protective relays keep event reports and oscillography that record current and voltage on every phase through the fault, sample by sample. Read correctly, they establish when the fault began, how it evolved between phases, which devices saw it, and how long clearing took. Read carelessly (with CT inputs mapped to the wrong location, or an event report confused with the sequence-of-events log), they produce confident conclusions that do not survive the other side’s expert.
The as-found setting file matters as much as the event report. It has to be exported before anyone re-commissions the equipment, and compared both to the coordination study and to what the relay actually did. Where the maintenance history is in dispute, whether the last relay test was primary or secondary injection can decide the question.
At about 1 a.m., one side of a double-ended medium-voltage switchgear lineup tripped on an 86 lockout. The load transferred to the alternate source and none was lost. The event log showed a zone-selective interlocking (ZSI) bus trip.
The relay record did most of the locating. Only the two relays on the incoming main saw fault current, about 13.5 kA RMS recorded by both, while the tie and feeder relays did not, which places the fault inside that bus zone. The waveforms show the fault beginning as a line-to-line A-C fault and developing into a bolted three-phase fault within about 2 ms. The relay asserted the ZSI trip in roughly 70 ms and the main breaker opened in another 50 to 60 ms, for total clearing near 120 to 130 ms. The slightly longer declaration time is consistent with the relay determining fault direction while bus voltage had collapsed to near zero on all three phases. The protection operated as designed.
The physical evidence then had to agree with the record, and explain it. Arcing was found where the voltage-transformer leads come off the main bus, with damage consistent with an A-C initiation. The root cause was a cable installed longer than the manufacturer’s specified length. The extra length forced it to curl to within an inch of the energized medium-voltage bus, inside the minimum air gap the manufacturer requires.
Partial discharge across that reduced gap degraded the cable insulation continuously for about 15 months, until the insulation could no longer withstand the voltage difference and the fault initiated. The arc flash was the result of a cable that did not maintain the manufacturer’s minimum clearance. Comparing the installation against the manufacturer’s specified cable length and air gap (and not the damaged unit alone) is what moved the analysis from where the fault started to why.
Download the printable checklist and evidence log (PDF, 2 pages) →
Often, yes. The relay record establishes the fault zone, timing, and phase sequence independently of the physical damage, and the damage pattern, arc tracks, and an undamaged exemplar section narrow the initiating location. Where two or more hypotheses genuinely cannot be ruled out with the evidence available, “undetermined” is a valid engineering conclusion, and saying so is more defensible than choosing one.
Because correct operation limits the damage but does not explain why the fault began. Once the protection is shown to have worked, the dispute usually moves to the insulation failure itself (design, manufacture, installation, or maintenance) and to who was responsible for each.
Not the damaged sections. All potentially interested parties should be put on notice and given the opportunity to attend a joint inspection before anything is disturbed. Service can usually be restored around the evidence with a temporary re-feed, so the choice between getting the facility back online and preserving the evidence is rarely as stark as it first appears.
Direct principal access. Conflict check and retention letter typically within one to two business days.