Line 7 Investigation

00Summary

The short version, before the detail.

01What we found

Tap any finding to see what it means.

02Supply data at the utility main

What the analyzer recorded, 8–13 August.

Utility main — line-to-line and line-to-ground voltageMEASURED

Figure 1. Line-to-line voltage with one-second min/max band (top) and line-to-ground voltages (bottom). Orange marker: controlled capacitor switching test, 10 Aug. Red marker: storm sag, 11 Aug. Tap to enlarge.

03Line 7 single-phase transformers

At least four fed from the Line 7 bus duct. The unit below is the first load directly downstream of the machine fuse that blew, and the only one of the four showing distress.

Line 7 transformer — enclosure and resin extrusionUNALTERED SITE PHOTOGRAPHS

Figure 2. Left: nameplate and enclosure, with resin runs down both sides. Right: resin extruded at the base. The enclosure top is clean, so the material originates inside the unit rather than dripping from above.

04Line 7 machine control cabinet

Where the blown fuse sits, and where the transformer is mounted relative to it.

Control cabinet — blown fuse and transformer positionUNALTERED SITE PHOTOGRAPH

Figure 3. Red: location of the blown fuse. Green: position of the 5 kVA transformer, mounted above the machine (out of frame).

05Working theory

Tap a marked part of the diagram to see what it does in this theory.

Each transformer is a single-phase unit connected phase-to-phase across the 480 V ungrounded delta. A winding connected this way presents series inductance between the two phases it bridges. To a fast transient — such as those produced by utility capacitor switching — that inductance acts as a choke, impeding transient current between those two phases, while the third phase, with no winding attached, is unimpeded. The result is an uneven distribution of transient energy across the three phases of the Line 7 bus.

This would account for each observed symptom

06Next steps

Filter by who owns each item.

00Report identification

480 V service entrance — engineering report.

Evidence basis

01Executive summary

Monitoring at one-second resolution with simultaneous voltage, current, power and harmonic capture, supplemented by waveform records, a controlled switching test, severe weather data and site correspondence.

02Scope, instrumentation and method

2.1 Measurement configuration · 2.2 Measurement boundary · 2.3 Analytical method

A Fluke 1775 analyzer was installed in the utility CT and metering section of Switchboard A, ahead of the A1 and A2 system mains. 322 parameters at one-second aggregation, retaining minimum, mean and maximum sub-interval values. Waveform capture ran concurrently on a separate trigger with sub-millisecond resolution.

Figure 1 — Measurement location. The analyzer was installed in the marked utility CT / metering section, immediately adjacent to the A1 SYSTEM and A2 SYSTEM mains. All Gear A distribution, including Line 7, is downstream of this point.

WHY THIS LOCATION MATTERS The analyzer sits ahead of every panel, feeder and machine on Gear A. Anything arriving from the utility was captured, so conclusions about external causes are strong. Anything generated inside the facility downstream of the switchgear was not captured, so the absence of a recorded disturbance does not exclude an internal cause — it points to one.

ANALYZER

Fluke 1775

AGGREGATION

1 s

PARAMETERS

322

SUB-INTERVAL

min / mean / max

WAVEFORM TRIGGER

sub-millisecond

RECORDS

519,161

2.3 Analytical method

03Session overview

Six days at the 480 V service entrance.

Six-day session overview, 480 V service entranceMEASURED

Figure 2 — Session overview. Panel 1: L-L voltage with one-second min/max envelope · Panel 2: L-G voltages, balanced throughout · Panel 3: phase currents with one-second maxima · Panel 4: true power factor. Markers: 1 capacitor energized (DarkHawk) 8/10 10:25:46 · 2 de-energized 10:41:38 · 3 third-party capacitor close 8/10 16:04:47 · 4 73.2% sag 8/11 10:30:09 · 5 logger removed 8/13 15:39. Tap to enlarge.

04Power factor

True and displacement power factor track closely, confirming fundamental reactive demand rather than harmonic distortion. Reactive power runs 200–400 kvar against 140–600 kW real load.

Load by hour of day (left) and power factor across the session (right)MEASURED

Figure 3 — Load profile and power factor. The shaded band on the left is the observed range at each hour; the line is the hourly mean. True power factor never reaches 0.90 at any point in the six-day record.

Figure 4 — Reactive demand against real load, with constant power factor references. The measured population sits well below the 0.85 line across the full load range.

BASIS OF ASSESSMENT This utility bills on peak demand rather than power factor, so no billing penalty arises from the measured value and none is claimed here. Power factor is assessed on electrical grounds only: elevated reactive current increases conductor and transformer heating for the same useful output, consumes capacity in the service and feeders, and increases voltage drop under load. A billing structure which does not price reactive demand is consistent with heavier reliance on utility-side capacitor banks for voltage support, which is relevant to Section 5.4.

CAUTION ON CORRECTION Any power factor correction installed at 480 V on this system must incorporate series detuning reactors. Undetuned capacitance at the low-voltage bus is the specific mechanism by which the utility switching transients documented in Section 5.4 would be amplified further.

05Capacitor switching analysis

5.1 Controlled test, 10 August 2026

A controlled test was performed with monitoring in place. A failed capacitor cell was found and re-fused. All three cells (600 kvar total) were gang-closed, held for approximately 16 minutes, then opened. The recorded data corresponds to this sequence to the minute, providing independent verification of both the instrument clock and the switching times.

600 kvar utility capacitor bank — energize / de-energizeMEASURED

Figure 5 — Capacitor energization and de-energization. Shaded region is the energized interval. Panel 2 shows the transient peak at closing. Panel 3 shows the small but repeatable distortion shift while the bank was in service.

JELSERT · Event ID 2252 · 10 August 2026, capacitor energizationUNALTERED INSTRUMENT RECORD

Figure 6 — Waveform record, event 2252, capacitor energization 10:25:46. Peak excursion reaches approximately 718 V against a 679 V nominal peak — 1.06 per unit against a 2.0 per unit industry concern threshold. High-frequency ringing is visible on all three phases.

Instrument trend view · 10 August 2026 · 600 kvar bankUNALTERED INSTRUMENT RECORD

Figure 7 — Trend view of the same operation. Voltage steps up at close, holds flat for approximately sixteen minutes, and steps back down at open. Cursors bracket the energized interval. Voltage distortion is unchanged; current distortion rises slightly and reverts.

5.2 Back-to-back switching

The station capacitors were already energized at the time of the test. The gang-close therefore constituted back-to-back switching, and no inrush reactors are installed. IEEE 141 §8.12.2.3 addresses this case directly: transient voltages are comparable to single-bank switching, but because the banks are electrically adjacent the inrush currents may be substantially higher, limited only by bus, switch and instrument transformer inductance typically in the range of 100 to 200 microhenries.

5.3 Third-party switching event, 16:04:47

The bank was left open on departure. At 16:04:47 the same day a second switching operation occurred that was not performed by DarkHawk. Waveform record 2256 classifies it as a waveshape deviation of 299.5 V over 0.717 ms, triggered on phase C-A. The associated steady-state step was approximately +8 V, roughly three times the magnitude of the 600 kvar operation. The significant characteristic is rate of change rather than magnitude: a comparable notch depth occurring in approximately one ninth of the time represents an appreciably steeper dv/dt, which is the parameter governing stress on metal-oxide varistors and rectifier semiconductor junctions. Overshoot nevertheless remained near 1.03 per unit.

JELSERT · Event ID 2256 · 10 August 2026, 16:04:47UNALTERED INSTRUMENT RECORD

Figure 8 — Waveform record, event 2256, 16:04:47 on 10 August. Notch depth 299.5 V over 0.717 ms — comparable depth to Figure 6 at roughly one ninth of the duration.

5.4 A recurring contributor

The controlled test characterizes one bank under one operation and does not characterize the exposure. Several capacitor banks operate on this feeder, switching on schedules outside the customer's knowledge or control. Their aggregate effect is a repeated transient duty on every piece of connected equipment in the building.

06Disturbance inventory

Twenty-three seconds exhibited a within-second excursion exceeding 8 V. Depth is the lowest line-to-line one-second minimum as a percentage of 480 V nominal. Filter or sort the table.

6.1 Distribution and severity

Disturbances per day vs recorded precipitationMEASURED

Figure 9 — Disturbances per day with recorded precipitation. Seventeen of twenty-three events fall on 11 August, the day of the only confirmed convective passage near the site.

Figure 10 — All twenty-three measured disturbances plotted against the approximate ITIC lower susceptibility limit. Twenty-two lie above 90% of nominal. One event reaches 73.2%.

WHAT THE SEVERITY DISTRIBUTION MEANS Only one disturbance falls into the range where drive undervoltage protection typically operates. Published guidance places most PWM drive undervoltage trip points between 70% and 85% of nominal DC bus voltage, placing the 11 August event at the margin. The remaining twenty-two are shallow enough that correctly configured equipment should ride through them without response. The customer reports no equipment response to the 11 August event.

22Above 90% nominal · of 23 events

1In drive-trip range · 73.2%, at the margin

6.2 Largest current excursion

JELSERT · Event ID 1835 · 10 August 2026, 02:52:35UNALTERED INSTRUMENT RECORD

Figure 11 — Waveform record, event 1835, 02:52:35 on 10 August. Peak current 2347 A with pronounced DC offset decaying over approximately ten cycles, followed by sustained elevated load. This is a large motor or line start — a normal facility operation, and the largest current excursion in the session.

07Storm fault sequence and recloser confirmation

NOAA severe weather records storm cell detections near the site at 10:32:50 and 10:36:03 on 11 August. The site recorded 2.83 inches of precipitation that day with gusts to 53 mph. The electrical disturbance cluster spans 10:28 to 11:10, bracketing the detections.

Storm fault sequence, 11 August 2026MEASURED

Figure 12 — Storm fault sequence, 11 August. Red marker is the 73.2% sag. Green and purple dashed markers identify two independent R2-R30 recloser sequences. The lower panel is the peak across all three phases, reaching 1361 A at the sag; Table 4 lists the phase-A value for the same second, 1247 A.

7.1 Recloser sequence identification

R2-R30 reclosing on this feeder: initial trip, first reclose ~2 s, second reclose ~30 s. The plant did not lose supply, so the faults were on adjacent circuits sharing the substation bus — observed here as a sag at each fault-on interval. Two sequences are identifiable:

7.2 Fault location inference

Twelve of the seventeen events on 11 August, including the deep sag, exhibit the B-C phase pair as the lowest voltage. Repeated faults on a single phase pair through a convective passage indicate one recurring fault location — typically vegetation contact or a tracking insulator.

08Line 7 surge protective device failures

8.1 Scope of the failures

8.2 Reported timing

Two consequences. It eliminates loss of supply as an explanation — per manufacturer literature the LED extinguishes only when the thermally protected MOV assembly has disconnected at end of life, so these units require replacement, not reset. And it sets a firm upper bound on the failure time: already failed by approximately 20:00 on Friday 7 August, with production running. Line 7 then ran until approximately 06:50 the following morning with two of three machines unprotected. No fuse failure, breaker operation or damage was reported. Site inspection found bus bar fuses intact and the fuses, breaker and drives inside Rennco 1 and 3 serviceable.

8.3 Device suitability

The installed devices are confirmed by the customer as Mersen Surge-Trap STXR480D05 — the delta variant of the series. Manufacturer data lists this series as rated for 480 V three-phase delta and high-resistance-grounded wye systems, with maximum continuous operating voltage of 550 V line-to-ground and 1100 V line-to-line. IEEE 141 §6.5.1 and industry practice require MCOV at line-to-line voltage on ungrounded systems, because a single ground fault imposes full line-to-line voltage on the unfaulted phases for an indefinite period.

8.4 Selective failure

The analyzer sat upstream of all facility distribution, so anything reaching Line 7 from outside passed through it first. Nothing did on either candidate date. An external origin is excluded; the cause lies downstream of the main switchgear.

8.5 Leading explanation

Reported: one L1 fuse, in the Rennco bucket, first occurrence. A metal-oxide varistor failing shorted draws fault current through the affected phase only. If landed on the load side of the bucket fuses, that failure opens exactly one fuse and disables the device in the same instant — one phase, one fuse, one dead device. IEEE 141 §6.6 notes arrester failure entails very low impedance with current approaching fault magnitude.

09Conclusions and recommendations

9.1 Hypotheses evaluated against the measured record. Filter by status.

9.2 Recommended actions

10Reference drawings and site views

Distribution drawings prepared by e-Hazard (arc flash study, 19 August 2025), reproduced to document the path between the measurement point and the affected equipment.

Figure 13 — Substation A distribution sections, showing feeder breaker identification.

Figure 14 — Switchboard nameplate. Square D QED Power Style, 480 V, 60 Hz, 2000 A. System designation 3 phase, 3 wire — documentary confirmation of the ungrounded delta configuration on which §8.3 depends.

11Open items — information required

The findings are complete to the limit of the information available. These remain outstanding and all require action or information from Jel Sert personnel, ordered by their effect on the conclusions.

11.1 Tier 1 — required to close the Line 7 investigation

11.2 Tier 2 — required to complete the facility picture

11.3 Tier 3 — useful, non-blocking