The neutral that wasn’t.

§1Facts of the loss

The Mill Creek Tube and Pipe Mill is a 78,000 square-foot single-story manufacturing facility on West 12th Street built in 1962 and expanded twice (1971 and 1989). The plant manufactures structural steel tubing in sizes from 1.0 to 8.0 inches outside diameter on three production lines. Electrical service is 480Y/277 V three-phase three-wire-with-grounded-neutral from a utility-owned 1,500 kVA pad-mount transformer at the west end of the property; service entrance is in the southwest corner of the building, feeding a main switchboard rated 2,500 A and serving five 480 V motor-control centers (MCC-1 through MCC-5) distributed across the production floor.

At approximately 14:00 EDT on 9 February 2026, the first-shift maintenance lead, Charles Brigham, observed unusual interior lighting flicker as he passed MCC-3 on his return from a morning materials draw at the receiving dock. By 14:01 the lights on the production-line-2 area had dimmed and Mr. Brigham smelled hot insulation in the MCC-3 vicinity. At 14:02:38 (timestamped by the building automation system’s electrical sub-meter) the southwest 480 V switchboard’s phase-A current spiked from 622 A to approximately 1,580 A in less than 50 ms, followed by activation of the smoke alarm in the MCC-3 cubicle at 14:03:11. Mr. Brigham initiated the manual pull station at 14:03:24 and evacuated the floor. The Millcreek Township Fire Department arrived at 14:09 EDT.

No injuries. Production loss: 14 days at $284k/day average revenue. Structure damage: limited to the southwest MCC vault and the cable tray serving it. Total claim: $4.2 million. The insurer, Erie Insurance Exchange, retained the firm on the evening of the loss with subrogation pursuit as the goal — the prior electrical contractor on the building (a since-closed Erie firm we name in the sealed report) had performed a 2018 service-entrance refurbishment that is the suspected origin point of the failed neutral.

§2Scene investigation

The scene was released by the Millcreek Township Fire Department at 19:15 EDT on 9 February. Presque Isle PFE’s engineers were on scene at 06:45 EDT on 10 February under the suppression authority’s and the carrier’s combined release.

The first photographic set documented the eight cardinal-direction approaches per NFPA 921 §17.5.4. The second set documented the service entrance on the building exterior, including the utility metering compartment with the seal intact and the transformer pad. The transformer pad was undisturbed and the secondary cable trough was clean of any fire damage; the source of the fire was clearly downstream of the metering point.

The third set, conducted with the dead-front of the southwest main switchboard removed, was the dispositive evidence. The neutral bar termination at the service entrance — a lay-in lug rated 1,000 kcmil — showed clear evidence of historic overheating. The lug body was discolored bronze, the copper neutral conductor at the termination point had recrystallized to a coarse grain structure visible without microscopy, and the lay-in screw was rotated approximately 1/3 of a turn below the manufacturer’s torque specification. The neutral conductor itself was intact downstream of the lug, but in the lug body it had become a high-impedance connection.

Arc-mapping per NFPA 921 §9.11 followed in MCC-3, where the fire originated. The primary arc location was at the line-side terminal of a Cutler-Hammer Magnum DS frame breaker rated 800 A serving production line 2. The bus stab at the line-side phase-A terminal showed parallel arc damage to ground; the parallel arc developed when the available short-circuit current at the MCC bus was insufficient to clear an initial line-to-neutral fault and the fault progressed to ground through the arc carbon.

§3Exhibits

The plant’s maintenance group preserved nine months of monthly clamp-meter logs on the southwest service entrance, taken as part of a 2024 reliability-centered-maintenance program. The logs (Exhibit C-001 through C-018, retained in the firm’s evidence vault) showed phase-A current consistently 8–18% above phase-B and phase-C from June 2025 through January 2026. The unbalance had been noted in the May 2025 entry and assigned to a follow-up “motor balancing study” that was never opened.

Exhibit C-019 is the recovered lay-in lug from the service-entrance neutral bar — bagged, sealed, and retained in the evidence vault under chain of custody. Exhibit C-020 is a 6-inch section of the neutral conductor at the lug body, retained alongside. The conductor section was sectioned at Intertek’s Mentor, OH metallurgical lab on 12 March 2026; the cross-section shows the coarse-grain recrystallized structure consistent with prolonged operation above the conductor’s rated temperature.

§4Hypothesis testing

The standing R-01 hypothesis grid (see R-01 §4) was applied to the matter. The eight working hypotheses were tested as follows:

• H1 · Loose connection at the MCC-3 breaker line-side terminal. Tested negative — the MCC-3 terminal at the failure point showed no evidence of pre-existing glowing-contact damage; the terminal’s arc damage was acute and consistent with a single high-current event.
• H2 · Series arcing through a degraded conductor. Tested negative — no evidence of multiple-site arc damage along the feeder.
• H3 · Parallel arcing. Tested as the immediate fault mode at MCC-3 — yes — but as a consequence of H4 progression rather than as the originating cause.
• H4 · Overload heating. Tested as the underlying cause, sustained by Exhibit C-001 through C-018 showing 8–18% phase unbalance for nine months. The unbalance was the consequence of the degraded service-entrance neutral redirecting unbalanced single-phase load current through the phase-A conductor system. The phase-A bus and feeder cable had been operating at approximately 110–115% of their ampacity for nine months before the day of the fire.
• H5 · Surge-induced damage. Tested negative — Vaisala NLDN search of an 8 km radius and ±60 minute window around the fire returned no strokes; no utility SCADA-logged surge events on the affected feeder.
• H6 · Failed appliance. Tested negative — fire origin is at MCC-3, not at any production-line appliance.
• H7 · Non-listed wiring. Tested negative — no evidence of non-listed conductors in the affected feeder.
• H8 · Lightning. Tested negative — see H5.

The cross-referenced R-06 investigation by Adekunle confirmed the degraded neutral mechanism. Resistance-to-earth at the service-entrance grounding electrode system was 6.4 Ω (within NEC 250.53 limits, ruling out earth-return as the unbalance-conducting path). The unbalance current returned through the phase-A conductor system itself, accelerating the conductor’s thermal degradation through ordinary I²R heating.

§5Finding

The originating cause of the fire at the Mill Creek Tube and Pipe Mill on 9 February 2026 was sustained overload heating of the southwest 480 V service-feeder phase-A conductor over a nine-month period preceding the fire, the overload itself produced by a high-impedance termination of the service-entrance grounded neutral conductor at the main switchboard’s lay-in lug. The high-impedance neutral termination was installed during the 2018 service-entrance refurbishment by the prior electrical contractor (named in sealed exhibit). The fire was preventable by reading of the plant’s own monthly clamp-meter records, which had recorded the phase-current unbalance for at least nine months prior to the fire. Sealed PE report · 4 May 2026 · Lindqvist-Vasilakou (R-01) & Adekunle (R-06)

The finding supports subrogation pursuit against the 2018 electrical contractor under Pennsylvania’s twelve-year statute of repose for construction-defect claims (42 Pa C.S. §5536). The contractor’s insurance carrier was placed on notice on 12 May 2026; settlement negotiations are pending.

§6Aftermath

The plant returned to production on 28 February 2026 after a temporary service-entrance refurbishment by a different electrical contractor. The temporary service is to be replaced under permanent refurbishment in Q3 2026; specifications for the permanent work were drawn by a separate consulting engineering firm in accordance with NEC 250 and IEEE 142 §4. Presque Isle PFE was not retained for the replacement design and is not in a position to comment on its sufficiency; we declined the retainer due to the conflict with our ongoing subrogation work.

The matter became the subject of a feature in the Erie Times-News’s 23 March 2026 industrial-safety section under the byline of T. Vespasian; the article correctly characterized the firm’s findings and the working-distance question. The Filed Record’s Vol. VIII No. 2 was prepared in part on the strength of the lessons this matter teaches: a plant’s own preserved electrical data, taken seriously, can prevent a $4.2 million claim. Three other Erie-area industrial customers have since opened standing R-04 retainers with the firm based on the publication of this case.

§7References