Thermography for Predictive Maintenance

Thermography, or infrared thermography (IRT), is a predictive maintenance technique that uses a thermal camera to see heat patterns on equipment. Because most developing faults get hot, a loose electrical connection, a failing bearing, a blocked cooler, an infrared image reveals them before failure, with no contact and no shutdown.

A thermal camera turns invisible infrared radiation into a picture where color maps to surface temperature. That single capability spans two worlds most plants treat separately: the electrical side, where hot spots warn of loose connections and overloads, and the mechanical and process side, where heat exposes friction, blocked flow, and lost insulation. This guide covers thermography broadly across both; for the electrical-panel discipline specifically, see infrared electrical inspection.

What is thermography in maintenance?

It is the use of infrared imaging to find equipment problems by their heat signature, as one of the core tools in a predictive maintenance program. Alongside vibration analysis and oil analysis, thermography is a condition-monitoring method: it reads a symptom, surface temperature, that changes as a fault develops, so you can act on evidence instead of a calendar. The international standard for its use on machinery, ISO 18434, treats it exactly this way, as a condition-monitoring and diagnostic technique.

What makes it powerful is speed and reach. A technician can scan a whole switchgear lineup, a row of motors, or a bank of steam traps in minutes, from a safe distance, while everything runs under normal load. No other condition-monitoring method covers that much equipment that fast. That is why thermography is often the first predictive technology a plant adopts on the road from reactive to proactive maintenance.

It is also non-invasive, which matters more than it sounds. You do not have to stop the line, open a housing, or de-energize a circuit to get a reading, in fact you must not, because most electrical and mechanical faults only reveal their heat under normal load. That lets a plant inspect equipment it could never justify shutting down, and it means a scan can be repeated on a route without ever disrupting production.

How does infrared thermography work?

Every object above absolute zero emits infrared radiation, and the hotter it is, the more it emits; the camera measures that radiation and converts it to a surface-temperature map called a thermogram. The technician then looks for anomalies, a component hotter than its neighbors under the same load, or hotter than it was last time. Comparative thermography, judging a part against a similar part or against its own baseline, is the most common and most reliable approach.

The catch is that the camera measures radiation, not temperature directly, and turning one into the other depends on emissivity how efficiently a surface radiates heat. A dull painted panel has high emissivity and reads accurately; bare shiny metal has low emissivity, radiates poorly, and mirrors the temperature of things around it, so it reads wrong unless you correct for it. Reflection, viewing angle, distance, and load all bend the reading too. Getting these right is the difference between a diagnosis and a guess, and it is why thermography is a skill, not just a camera.

What faults can thermography find?

Anything where a developing problem changes surface temperature, which covers a surprising share of plant equipment. On the electrical side it finds loose or corroded connections, overloaded circuits, unbalanced phases, and failing breakers, problems that heat up long before they trip or burn. On the mechanical side, heat from friction exposes misaligned couplings, over- or under-lubricated and failing bearings, and overloaded gearboxes and belts.

It also reads process and building faults that have nothing to do with rotating iron: blocked or fouled heat exchangers and coolers, plugged pipes, tank and vessel liquid levels, failed steam traps, refractory and insulation breakdown, and roof or envelope moisture. Because the camera sees the whole scene, a single walk-down can flag issues across electrical, mechanical, and process systems at once, which is why it earns a place on the condition-based maintenance route.

How do you run a thermography program?

A useful program is a repeatable route with baselines, severity criteria, and a path from finding to work order. Here is the sequence:

1. Pick the targets by criticality. Scan the equipment whose failure hurts most first, main switchgear, critical motors and drives, key exchangers. Rank with your reliability data rather than scanning everything equally.
2. Set inspection conditions and load. Electrical and mechanical faults only show up under load, so scan equipment running at a meaningful, repeatable load. Note ambient temperature and load each time so images compare fairly.
3. Capture baselines. Image each target when it is known-good to establish its normal pattern. Most diagnoses are comparisons, this component versus its twin, or versus its own baseline, so the baseline is what later readings mean anything against.
4. Correct for emissivity and reflection. Set the right emissivity for each surface, watch the viewing angle, and account for reflected sources. This step is what separates a real temperature from a misleading one.
5. Apply severity criteria. Grade each finding by temperature rise over a reference, a similar component, ambient, or the maximum rating, and assign a priority. Consistent criteria keep "hot" from meaning different things to different techs.
6. Turn findings into work orders. Every actionable exception becomes a CMMS work order with the image, the delta-T, load, and location attached, then flows into planning and scheduling. A finding with no work order is just a photo.
7. Trend and re-scan. Re-image on a set interval, and after every repair to confirm the fix. Rising temperature over successive scans is the early warning; a flat trend is your all-clear.

What are the limits and pitfalls of thermography?

The biggest limit is that thermography only sees surface temperature, so it catches faults that produce external heat and misses those that do not. A crack that has not started rubbing, an early bearing defect that is not yet warm, or a sealed problem behind insulation may be invisible, which is why thermography complements vibration and oil analysis rather than replacing them. It also cannot see inside energized enclosures without an IR window or an open panel, and opening panels is its own arc-flash hazard requiring the right procedures and PPE.

The most common way results go wrong is human: wrong emissivity setting, a reflection mistaken for a hot spot, scanning at low load so a real overload stays cool, or comparing images shot at different loads and ambients. These are all avoidable with training and consistent conditions, which is exactly what ISO 18434 and a disciplined route enforce. Thermography rewards a program and punishes a one-off scan.

The findings only pay off when they turn into action, and that is where connection matters. When each thermal exception becomes a tracked work order with its image and delta-T attached, instead of a JPEG in someone's phone, the fault gets fixed and the trend gets built. Harmony connects inspection findings to work and asset history so a rising hot spot surfaces and gets scheduled rather than forgotten (see the platform), layering onto the systems a plant already runs with no rip-and-replace. A connected plant is walked through in the CLS case study.

Where does thermography fit in the maintenance system?

Thermography is one instrument in the predictive maintenance toolbox, feeding a broader condition-based strategy. Its findings become CMMS work orders, its trends inform reliability decisions, and its severity grades help planning and scheduling put the most urgent repairs first. Used well alongside vibration and oil analysis, it turns temperature, the most visible symptom of a failing machine, into early, actionable warning across the whole plant.