The Thermal Image Shows Something, But Normal Operation Explains It (Decision Tree)

Why this matters

The fastest way to lose a customer's trust is to quote a repair off a thermal image that turns out to be nothing, a warm compressor that is just running, a bright terminal that is just reflecting the sun. The second fastest way is the opposite mistake: waving off a real hot spot because "that's probably normal" and having it fail two weeks later. This tree separates a genuine fault from a shot that just needs context.

Start here: is the component under load right now

The single most common source of a false alarm is shooting a component that is not doing its normal job at the moment of the scan.

• If the component was just running and just shut off, expect residual heat with no fault behind it. Wait for it to reach a stable state, or note the time-since-shutoff and re-shoot later.
• If the component is idle and cold when it should be under load (a standby pump, an off-cycle compressor), you cannot read anything meaningful yet. Come back when it is doing work, or induce the load if it is safe to do so.
• If the component is actively under a known, steady load right now, move to the next check. This is the condition a thermal reading is trustworthy in.

Check for a reflection before you trust the color

Reflections mimic real heat and are the single most common false positive in the trade.

1. Move the camera angle 20 to 30 degrees and re-shoot the same spot. A reflection shifts or disappears with the angle; a real heat source stays put.
2. Look for a plausible reflected source nearby: sunlight through a window, a space heater, another hot component, even the tech's own body. If one is present and the hot spot lines up with it geometrically, it is very likely a reflection.
3. If the hot spot survives the angle change and no reflected source lines up, treat it as real and continue.
4. If it disappears or shifts, it was a reflection. Re-shoot from the clean angle and re-evaluate from a false-positive-free baseline.

Check whether the surface is shiny or matte

Bare polished metal (copper bus, unpainted steel, aluminum housings) has low emissivity and reads unreliably, sometimes cooler than it actually is, sometimes picking up ambient reflection instead of its own heat.

• If the suspect surface is shiny bare metal, place a strip of matte tape or flat paint on it, let it stabilize, and re-shoot. Trust the taped reading over the bare metal reading.
• If the surface is already matte, painted, or a standard plastic housing, the raw reading is close to accurate and you can proceed to the comparison step.

Compare to an identical reference under the same load

A single hot reading with nothing to compare it to is not a diagnosis, it is a data point.

• If there is an identical neighboring component under the same load (the other two phases on a breaker, the matching bearing on the other end of a shaft, a duplicate unit nearby), compare directly. A close match to the reference means the heat is normal operating heat for that load. A meaningfully wider gap than the reference points to a real developing fault, and the gap should widen further under a moderate load increase if you can safely test that.
• If no direct reference exists, compare to the component's own baseline from a prior visit or the manufacturer's normal operating range if documented. No baseline and no reference means you log the reading, note the load condition, and plan a follow-up scan rather than call it a fault on one data point.

If it still reads hot after ruling out state, reflection, and emissivity

At this point a genuine delta under matched conditions is a real finding, not an artifact. Move to diagnosing the specific component (a loose connection, a failing bearing, restricted flow) rather than continuing to second-guess the thermal read itself. Document the load, the delta, and the reference used; that record is what justifies the repair to the customer and protects you if the finding is challenged later.

Recap

1. Confirm the component is under real, steady load at the moment of the shot.
2. Rule out reflection by changing the camera angle.
3. Rule out a shiny-surface false reading by checking emissivity or adding a matte reference patch.
4. Compare to an identical or historical reference under matched load.
5. Only call it a fault once state, reflection, and emissivity are ruled out and the delta holds up against a real comparison.

References

• NFPA 70B (electrical equipment maintenance, includes infrared inspection guidance)
• Manufacturer documentation on thermal imaging camera emissivity settings and calibration
• See related: What a Hot Spot in a Thermal Image Actually Means