9.1 Three-Phase Electrical Fault Patterns

Why Three-Phase Patterns Dominate Electrical IR

Most industrial and commercial power systems are three-phase. That symmetry is a gift to the Level II thermographer: under balanced load and comparable geometry, the three phases of a feeder, breaker, contactor, or motor terminal block should show similar surface temperatures. Pattern analysis means you do not evaluate one hot lug in isolation—you compare like to like and then classify severity with a standards-based ΔT.

Level I work often stops at “phase C looks warmer.” Level II work must answer: How much warmer? Compared to what? At what load? Which failure mode fits the pattern? Exam items and field reports both hinge on that discipline.

The Similar-Component Rule

NETA-style electrical severity (the P1–P4 convention used throughout this guide and the published curricula) is built on temperature difference versus a similar component under similar load, not merely versus ambient air. For three-phase equipment, the “similar component” is almost always the other phases on the same device or an adjacent identical device carrying comparable current.

Procedure habit: Record phase temperatures A/B/C, note load current per phase if available, note emissivity and RAT settings, then compute ΔT = T_suspect − T_reference (usually the coolest or median healthy phase under the same conditions).

Pattern Family 1 — Single Hot Phase

When one phase is clearly hotter than the other two, and the other two agree with each other, treat the pattern as a localized anomaly until proven otherwise.

Most common causes:

1. High-resistance connection — loose lug, under-torqued bolt, oxidized crimp, fretting corrosion, damaged strand, or partially engaged stab. Heating follows I²R: even a modest extra resistance can create a large hotspot at high current.
2. True load imbalance — one phase carrying substantially more current than the others (single-phase loads, open or high-impedance neutral effects on related systems, unequal process loads on multi-motor feeds).
3. Less common electrical internals — deteriorating winding section, poor fuse clip contact on one pole, single-pole contactor contact wear.

You cannot decide connection vs load imbalance from IR alone. The Level II response is:

1. Capture quantitative temperatures with correct parameters.
2. Measure or request phase currents (clamp meter / power meter) under the same operating state.
3. If currents are balanced and one terminal is hot → connection (or contact) problem.
4. If the hot phase also carries higher current → imbalance or overload on that phase (still may need connection checks downstream of the measurement point).
5. Never write “normal phase variation” as the conclusion without numbers.

Pattern Family 2 — All Three Phases Hot

When all three phases are elevated to a similar degree relative to historical baselines, identical lightly loaded equipment, or absolute insulation/conductor ratings, think system-level drivers:

• Overload — current above continuous rating or prolonged high duty cycle.
• High ambient / enclosure heating — poor ventilation, blocked louvers, failed fan, solar load on outdoor gear, adjacent process heat.
• Restricted cooling path — dirty heat sinks, lost natural convection in packed wireways, closed covers that normally run open only if designed that way (do not create unsafe configurations).
• Harmonic heating (advanced plant context) — additional heating in neutrals and some magnetic components; still document with measured load quality when available.

Diagnostic contrast: A single loose C-phase lug does not make A and B match C’s extreme temperature. Symmetric heating across A/B/C argues against a single-point connection fault as the primary story.

Pattern Family 3 — Two Hot, One Cool (and Other Asymmetries)

Two elevated phases with one cool phase can indicate lost load on the cool phase (open conductor, blown fuse, open contactor pole) or a measurement artifact (emissivity difference, reflection, different surface finish). Always:

• Confirm the cool phase is truly energized and loaded.
• Check that surface finishes and IR setup match across poles.
• Correlate with electrical measurements and operations status.

Load Similarity Is Non-Negotiable

Pattern comparison is invalid if the phases are not under similar load during the capture. A “cool” phase that is simply unloaded will look healthy next to a loaded hot phase and produce false negatives or false positives depending on how you frame the report.

• Prefer inspections near steady-state load (not during motor start or large load steps).
• NFPA 70B (2023) §7.4.4–7.4.5: inspect at normal circuit loading, and where that is not feasible at not less than 40 percent of nominal circuit loading, when assigning standard severity; higher load (often preferred toward ~70–100% where safe and practical) increases sensitivity to I²R faults.
• Document percent load. A modest ΔT at low load can be more serious than the same ΔT near full load because heating scales with the square of current.

Mapping ΔT to Priorities (Experience-Based criteria)

Use the Experience-Based bands published in Infraspection Standard §10.1 versus a similar component under similar load:

Example: Phases A/B measure 42 °C and 43 °C on lugs; phase C measures 68 °C at balanced 65% load. ΔT ≈ 25–26 °C → P2. Root-cause work still needs current confirmation, but the priority is already clear from pattern + ΔT.

Never Call Imbalance “Normal”

Exam and field failure mode: the thermographer sees a few degrees of phase-to-phase difference and writes “normal imbalance.” Problems with that phrase:

• “Normal” is not a severity classification.
• Small ΔT values still sit on the P4 monitor band if they exceed the healthy peer by 1–10 °C and are real (not noise).
• Palette stretch can hide or exaggerate differences; numbers decide, not color drama.
• True electrical imbalance may be acceptable to operations for a short period, but it is still a condition to measure and report, not a reason to skip documentation.

Better report language: “Phase C lug 8 °C above phases A/B at 58% load and balanced current (±3%). Classified P4; recommend re-torque/inspection at next opportunity and trend at next route.”

Measurement Quality Checklist for Phase Work

1. Emissivity appropriate to the surface (oxidized lug vs shiny plating); use tape/paint reference when metals dominate the error budget.
2. RAT measured or controlled—bus compartments are reflective cavities.
3. Focus and IFOV — the lug face must fill enough detectors; do not report temperature on a sub-resolution bolt head from across the room.
4. Same angle and distance across phases when practical.
5. Paired visual image and clear equipment ID (panel, cubicle, phase marking).

Worked Field Scenarios

Scenario A — Connection fault. MCC starter for a process pump: A=51 °C, B=50 °C, C=79 °C; currents within 2%. ΔT ≈ 28 °C → P2, high-resistance C-phase connection until proven otherwise. Do not leave as “monitor only.”

Scenario B — Load imbalance. Same temperatures, but C-phase current is 40% higher than A/B. Report both the thermal ΔT and the current imbalance; maintenance may rebalance loads and still inspect the C-phase connection because higher current stresses any weak joint.

Scenario C — Overload / ambient. A/B/C all ~75 °C in a sealed outdoor panel on a 40 °C day at 90% load, with identical spare panel at 55 °C under lower load. Pattern is global, not single-phase. Address load, cooling, and enclosure conditions; do not “fix” one lug.

Scenario D — False alarm. One phase looks hotter only because that lug is painted (ε ≈ 0.95) and peers are bare metal (ε left at 0.95 on camera). Correct emissivity or apply reference emitters; re-measure before writing a P-priority.

Level II Decision Flow (Mental Model)

1. Are A/B/C under similar, adequate load (≥40% guidance for standard severity)?
2. Is the pattern single-phase, two-phase, or all-phase?
3. What is quantitative ΔT to the best similar reference?
4. What do currents say about connection vs imbalance?
5. Assign P1–P4 (or absolute limits if those govern) and recommend action in plain maintenance language.
6. Schedule re-inspection after repair to close the finding.

Quick Answer: Single hot phase under balanced current → high-resistance connection (or contact) fault until proven otherwise; single hot phase with high current → imbalance/overload on that phase; all three hot together → overload, cooling, or ambient. Always compare similar components under similar load, map ΔT to P4 1–10 / P3 >10–20 / P2 >20–40 / P1 >40 C°, and never dismiss phase differences as “normal” without measured evidence

Criteria used in Chapters 9–14. Where these chapters quote a four-level priority, they use the Experience-Based criteria published in Infraspection Institute's Standard for Infrared Inspection of Electrical Systems & Rotating Equipment §10.1 — P4 = 1–10 C°, P3 = >10–20 C°, P2 = >20–40 C°, P1 = >40 C°, with Priority 1 most severe. ANSI/NETA Table 100.18 uses different bands (1–3 / 4–15 / >15 C° against a similar component) and assigns no priority numbers of its own; see Section 8.1. Always state in the report which criteria set you applied. .