Skip to content
Jinki
Request assessment
 · 04 MIN READ

From Hot Spot to Work Order: NFPA 70B-Style Severity for Aerial Thermal Findings

NFPA 70B made electrical thermal inspection a standard. How its severity thinking translates — and breaks — when findings come from the air.

NFPA 70B’s 2023 revision did something quiet but enormous: it turned electrical maintenance from advice into a standard. The current edition — NFPA 70B, Standard for Electrical Equipment Maintenance, whose 2026 edition is now in effect — put program requirements behind the work that reliability teams were already doing, infrared inspection among them.

That shift matters to anyone producing or consuming thermal findings, because a standard changes the question. It is no longer “did anything look hot?” It is “does this finding, documented this way, support the maintenance program you have to defend?”

What Changed in NFPA 70B

For decades, 70B was a recommended practice — a well-regarded guide no one was obligated to follow. The 2023 edition became a standard, with the language that implies: programs, documentation, and defined responsibilities. Its stated purpose is preventive maintenance for electrical, electronic, and communication systems — the equipment in industrial plants, commercial buildings, and institutional facilities — aimed at preventing failures and injuries.

For inspection work, three implications matter. Maintenance programs need to be documented, not improvised. Survey intervals follow equipment condition rather than one universal calendar. And the inspection record becomes part of the evidence chain — which is exactly where aerial thermal findings either earn their place or don’t.

”Condition of Maintenance” in Plain English

The standard’s framing is condition-based: equipment assessed into condition classes, with survey and maintenance effort following the assessment. Read plainly, the program — not the scan — owns the outcome. A scan is an input.

This is where recurring aerial coverage contributes cleanly: a dated, radiometric, comparable record of exterior asset condition over time. A single flight produces an observation. A year of flights produces a condition of record — evidence of what changed, when, under what conditions — which is the form maintenance and audit conversations actually need.

The Severity Ladder — and Why “Hottest Pixel” Is Not Severity

Conventional severity ladders sort findings by ΔT — the temperature difference between a component and a valid reference (a similar component under similar load, or the same component against its own baseline). Bands of ΔT map to priority: act now, schedule, watch.

The trap is treating temperature as the score. It is not, for three reasons:

  • ΔT needs load. The same connection reads differently at 20 percent load and at 90 percent. A ΔT without the load context is an unfinished measurement.
  • Absolute temperature misleads. A 70°C reading can be alarming on one asset class and ordinary on another; the Infraspection Institute’s electrical inspection standard — which documents temperature limits and a proven methodology for prioritizing exceptions — exists precisely because equipment classes behave differently.
  • Role matters. The same ΔT on a redundant feeder and a single point of failure are different business problems.

Which is why the professional practice, from ASTM E1934 to the severity conventions thermographers train on, ranks findings by interpreted condition — not by raw temperature.

What Changes at Altitude

Aerial thermal inherits the severity framework but not all of its inputs. Stated before any ranking:

  • No cover removal. Ground IR surveys often open enclosures to image connections directly. Aerial surveys image exterior surfaces as found — enclosures, radiators, bushings, conductor terminations, insulator strings.
  • Unknown load. Flight timing rarely aligns with peak load, and load state may only be available from SCADA afterwards. Findings are provisional until load context is attached.
  • View factor and distance. Measurement confidence depends on the target’s size in the frame — the distance-to-spot-size relationship every radiometric imager lives by. A small target from far away is a different measurement than the same target from close range.
  • Weather and sun state. Solar loading, wind, and wet surfaces all change what a surface radiates. Capture conditions belong in the record, not the footnotes.

None of this disqualifies aerial findings. It defines what has to accompany them: conditions, caveats, and a proposed — not final — priority.

An Aerial-Compatible Rubric

A defensible aerial finding sorts into three proposed classes:

  • ACT NOW. Clear anomaly on a critical, single-point asset, verified present on the visual layer, conditions favorable. Escalate to the qualified person immediately.
  • SCHEDULE. Measured ΔT against a comparable reference, consistent across passes, asset important but not single-failure. Recommend a targeted ground inspection within a maintenance window.
  • MONITOR. Ambiguous or condition-dependent signature. Track against baseline; re-scan under better conditions before spending anyone’s time.

Every class carries the same sentence: confirmation belongs to the qualified person responsible for the equipment. The aerial product proposes priority; the program disposes.

Documentation That Survives Audit

What makes a finding usable months later is the metadata around it: the radiometric payload (measurement-capable, not just imagery), ambient and wind conditions, emissivity assumptions, the asset identifier, and the comparison against prior passes. A finding ID that threads from report to work order to closure is what turns a thermal image into a maintenance record — which is the currency 70B-style programs trade in.

When NOT to Classify

Some signatures should never become severity classes: moisture on a surface, solar-driven reflections, known reflective materials, and residual heat after shutdown. These get flagged as reclassify or re-scan, not colored red. Inflating a class to look thorough is a failure mode with real costs — it burns credibility the first time maintenance finds nothing.

The data centers vertical covers how aerial thermal programs run over critical facilities; thermal imaging economics covers why recurring coverage pays; the intelligence scan covers the assessment itself; and a sample report shows how findings and conditions are documented.

QUESTIONS ANSWERED 05 ANSWERED
Does NFPA 70B require infrared inspections?

The 2023 edition moved 70B from a recommended practice to a standard and put program language behind condition-based maintenance, of which infrared inspection is a core method. Survey frequency is driven by equipment condition class rather than one universal interval. Programs should confirm which edition governs them.

Can an aerial thermal survey satisfy an NFPA 70B program?

Partially. Aerial surveys cover exterior asset classes at scale — service entrances, transformers, exterior switchgear, risers, bus — without de-energizing anything. They cannot remove covers or see connections inside enclosures; ground-based IR routes remain essential there. Aerial coverage complements the program, and the qualified person still owns severity.

Why is 'hottest pixel' a bad severity metric?

Because temperature without context says almost nothing. Severity depends on the temperature difference against a valid reference — the ΔT — the load at the time, ambient conditions, and the asset's role. A hot component at high load can be normal; a modest ΔT on a lightly loaded critical circuit can be serious. Ranking by raw temperature produces false alarms and missed priorities alike.

Who decides the final severity class?

The qualified person responsible for the equipment. An inspection provider — aerial or ground — documents findings, measurements, and conditions; the severity call that drives maintenance action belongs to the engineering authority over that asset.

What does an aerial thermal finding include?

Each finding carries its location on the asset map, thermal and visual imagery, measured temperature values where radiometric data allows, the conditions at capture (load state where known, ambient temperature, wind), and a proposed priority class — with the caveat that verification belongs on the ground.

Put this intelligence on your facility.

Severity-ranked findings, 48-hour turnaround, a mutual NDA before we see your site.

CMD ESC

↑↓ NAVIGATE · ↵ GO · CMD+K TOGGLE