
What infrared thermography is
Infrared thermography uses a thermal camera to measure the radiation emitted by a component's surface and display it as a temperature image. In predictive maintenance it finds hot spots (loose connections, bearings with friction, degraded insulation) before they become failures, without contact and while the machine is running.
The value is not in the colorful image but in the comparison: a spot is an anomaly only when it is hotter than it should be relative to an identical component under the same load.
Visual map of thermography in maintenance
The map shows the camera settings, the load correction, the severity table and the panel example.

How to measure correctly
The camera measures radiation, and converting it to temperature depends on settings. Get them wrong and you get a nice-looking wrong number:
- Emissivity (ε): how much the surface emits compared with an ideal body. Black electrical tape and painted surfaces sit near 0.95; polished copper or aluminum can be below 0.1. Shiny metal looks cold in the image even when hot. Fix: stick a piece of black tape next to the spot, or use a table value for the material and surface condition.
- Reflected temperature: shiny surfaces reflect hot or cold sources around them. Measure it and enter it in the camera, or change the angle to remove the reflection.
- Distance and resolution: the measured spot must be larger than the smallest detail the camera resolves at that distance. A small, distant target reads lower.
- Load: record each phase's current at the time of measurement. With no load, a bad connection does not heat up.
- Conditions: avoid wind, direct sun and rain on the target outdoors, and let the equipment stabilize under load before measuring.
One example shows how much emissivity matters. A surface at 80 °C with a true ε of 0.70, read with the camera set to 0.95, appears at about 54 °C (simplified calculation, ignoring reflection): an error of roughly 26 °C low.
Delta T and severity
The most common criterion compares the suspect spot with a reference:
ΔT = T of hot spot − T of reference (identical component, same load)
In three-phase panels, the reference is the average of the other two phases. With no identical component, compare with ambient temperature or the manufacturer's limit. A typical classification (reference values in the style of NETA specifications; your company should fix the limits in its own procedure):
| ΔT over similar component | Severity | Action |
|---|---|---|
| 1 to 3 °C | Possible deficiency | Record and reinspect on the next route |
| 4 to 15 °C | Probable deficiency | Repair at the next planned shutdown |
| Above 15 °C | Major | Repair as soon as safely possible |
The component's absolute limit (insulation, seal, grease) always applies, regardless of ΔT.
Load correction
Power dissipated in a resistive connection grows with the square of the current (P = I² × R). A connection showing 8.5 °C at 60% load will run much hotter at full load. Estimate:
Corrected ΔT = measured ΔT × (rated I ÷ measured I)²
It is an estimate, not a measurement: it assumes the defect's resistance stays constant and that the heat comes only from Joule heating. Many procedures apply the correction only above roughly 40% of rated load and always document both values, measured and corrected.
Example: a 400 A distribution panel
Fictional example. Ambient 25 °C, phase current 240 A on a 400 A breaker (60% of rated). Temperatures measured at the connection of breaker Q1: phase A 40 °C, phase B 41 °C, phase C 49 °C.
- Reference = average of phases A and B = (40 + 41) ÷ 2 = 40.5 °C.
- Measured ΔT = 49 − 40.5 = 8.5 °C (4 to 15 band: probable deficiency).
- Load factor = (400 ÷ 240)² = 2.78.
- Corrected ΔT = 8.5 × 2.78 = 23.6 °C (above 15: major).
The correction changes the decision from "next shutdown" to "act soon". The ΔT over ambient is 49 − 25 = 24 °C. The table summarizes three points on the same panel:
| Point | Load | Measured ΔT | Factor | Corrected ΔT | Severity |
|---|---|---|---|---|---|
| Breaker Q1, phase C | 60% | 8.5 °C | 2.78 | 23.6 °C | Major |
| Busbar joint | 90% | 12 °C | 1.23 | 14.8 °C | Probable (next shutdown) |
| Contactor K2, phase B | 80% | 2 °C | 1.56 | 3.1 °C | Possible (reinspect) |
The most common cause of these anomalies is an under-tightened or oxidized connection. The typical fix is to de-energize safely, clean, retighten to the specified torque and rescan under load to confirm that ΔT has dropped.
Where to apply it and how to integrate it
- Electrical: panels, breakers, fuses, busbars, motor terminations, transformers. This is the most common application.
- Mechanical: bearings, couplings, belts, pulleys and gearboxes. A hot bearing usually signals an advanced failure; combine it with vibration analysis, which sees the defect earlier.
- Process: steam traps, piping and insulation, refractory, tank and reactor levels.
Put thermography routes into the maintenance plan, with frequency set by asset criticality (RCM and FMEA results show where it detects the failure mode). Record findings as work orders in the CMMS, closing the loop with reinspection. Thermography is a predictive maintenance technique, and historical ΔT data feeds AI predictive maintenance models. If the result calls for stopping the plant, the repair can go into the planned shutdown. For bearings and gears, cross-check with oil analysis.
A useful report has: visible and thermal photo of the same spot, asset ID, each phase's load, emissivity and reflected temperature used, measured and corrected ΔT, severity, recommended action and deadline.
Safety and competence
Inspecting an energized panel requires an electrical work procedure: follow your site's electrical safety program (NFPA 70E in the US) with qualified staff and PPE rated for arc-flash risk. Infrared inspection windows let you scan with the panel closed. Whoever interprets the images needs training and practice; ISO 18434-1 (general procedures) and ISO 18436-7 (personnel qualification) are the reference for the method and for thermographer training.
Common mistakes and limits
- Measuring without load. A panel at 10% load hides defects that show up at 80%.
- Default emissivity on shiny metal. The result underestimates temperature.
- Judging by image color. The palette and auto-scale change; what counts is the measured value and ΔT.
- Comparing different components. The reference must be identical and under the same load.
- Not rescanning after repair. Without the second image, you do not know whether the problem is fixed.
- Expecting to see inside the equipment. The camera sees the surface; ordinary glass and acrylic block thermal infrared radiation.
Frequently asked questions
What does thermography detect in maintenance?
Hot spots caused by abnormal electrical resistance, friction, misalignment, poor insulation and flow blockage.
Which ΔT is considered critical?
In NETA-style criteria, above 15 °C over a similar component under the same load is major. Each company should set its own limits in a procedure.
Do I need to correct ΔT for load?
It is advisable when the load is partial, multiplying by (rated I ÷ measured I)². It is an estimate and is usually applied above about 40% load.
Why does emissivity matter?
Because the camera converts radiation to temperature using that value. Polished metal has low emissivity, and with a wrong setting the reading comes out far below the real one.
Does thermography replace vibration analysis?
No, they are complementary. Thermography is strong on electrical systems and heat points; vibration detects mechanical defects before the bearing heats up.
How often should I inspect?
It depends on criticality and failure history. Many plants run an annual route on panels and a higher frequency on critical assets.
Sources
- ISO 18434-1: Condition monitoring and diagnostics of machines — Thermography — Part 1: General procedures.
- ISO 18436-7: Condition monitoring and diagnostics of machines — Requirements for qualification and assessment of personnel — Part 7: Thermography.
- InterNational Electrical Testing Association (NETA). Maintenance Testing Specifications for Electrical Power Equipment and Systems.
- HOLST, G. C. Common Sense Approach to Thermal Imaging. SPIE Press / JCD Publishing.
- MOBLEY, R. K. An Introduction to Predictive Maintenance. Butterworth-Heinemann.
Keep studying: trusted sources and links
- ABRAMAN: Associação Brasileira de Manutenção e Gestão de Ativos (in Portuguese)
- SMRP: Society for Maintenance & Reliability Professionals
- Wikipedia: Maintenance
Save this address: Manufatura Enxuta is your school for lean manufacturing. Come back whenever you need to study, with a complete article, a visual map and trusted sources. Follow one of the learning paths or check the A–Z glossary.
Want to measure what really matters?
Download the free e-book on performance indicators and track the losses in your operation.
Download the e-book
