← Go back

Thermal Imaging Inspection: Heat Talks First, Smoke Talks Last

14 min
Published:
October 30, 2026

Thermal imaging inspection is the practice of scanning equipment with an infrared camera to find abnormal temperature patterns, such as a loose connection heating up or a bearing running warm, before they turn into failures. It is a condition-based maintenance technique: it shows the state an asset is in today, so a team can repair it on its own schedule instead of the failure's.

This guide comes from Holoindustry, a spatial computing company based in Budapest, Hungary, that builds the ThermalGlass family of wearable thermal imaging devices for industrial inspection and firefighting. It covers what thermal inspection finds, where it fits in a maintenance strategy, the standards that apply, where it falls short, and how to turn findings into repairs.

Let the Smoke Out: Why Heat Is the Earlier Warning

Ask an old electrician how electronics work and you may get the trade's favorite joke. They don't run on electricity. They run on smoke. You can prove it, too: let the smoke out, and they stop working.

It is 3:12 a.m. on a Tuesday, and Line 2 has just let its smoke out. Once the equipment is safely isolated, the night shift finds the culprit: a scorched lug on a 400 A breaker feeding the main conveyor drive. Melted insulation, discolored copper, and a smell that hangs around the motor control center all morning. By sunrise there is an emergency electrician on site, a part on an expedited courier, and a production manager doing painful arithmetic about missed shipments.

The scenario is illustrative, but anyone who has worked nights in a plant will recognize it. Smoke is the last thing a failing part tells you. That lug did not start failing at 3:12 a.m. For weeks it had been running a little hotter under load each day. The warning was there all along, in a wavelength nobody was watching.

Before the smoke, there is heat. The U.S. Department of Energy's O&M Best Practices Guide describes exactly this mechanism: a loose or corroded electrical connection runs abnormally hot because of increased resistance, and the pattern shows up in a thermal image before the connection is hot enough to cause equipment failure or a fire.

What Is Thermal Imaging Inspection?

Thermal imaging inspection, also called infrared thermography, is the use of an infrared camera to turn the heat that surfaces emit into an image, and to read that image for signs of deterioration. Everything above absolute zero emits infrared radiation in proportion to its temperature, so a camera that sees in the infrared can show where a component is warmer or cooler than it should be.

Two kinds of instrument do the work. A simple imager produces the picture only, which is enough when the pattern itself tells the story, such as missing insulation. A radiometric camera also measures temperature, which matters when the size of a difference decides whether there is a problem (DOE FEMP, Chapter 6).

The purpose of an inspection, in the words of ASTM E1934, the U.S. standard guide for examining electrical and mechanical equipment with infrared thermography, is "to identify and document exceptions." The international counterpart, ISO 18434-1, sets out the general procedures for using thermography in machinery condition monitoring, including how to establish severity criteria for what the camera finds.

Reactive, Preventive or Condition-Based: Where Thermal Inspection Fits

Thermal inspection belongs to condition-based (predictive) maintenance: it schedules work on evidence of an asset's actual condition, not on a calendar. Most plants mix three approaches, and the U.S. Department of Energy's guide defines them clearly (DOE FEMP, Chapter 5):

  • Reactive maintenance is "run it till it breaks." It has a place: for small parts, non-critical or redundant equipment and items unlikely to fail, letting them run to failure can be the right call.
  • Preventive maintenance is performed on a time- or run-time-based schedule. It suits equipment subject to wear and known failure patterns, but it cannot tell how one particular asset is actually aging, and it still lets catastrophic failures through.
  • Predictive or condition-based maintenance uses measurements that detect the onset of degradation, so its causes can be corrected before serious deterioration. It suits critical equipment and failures that occur at random rather than with age.

Reliability engineers picture a developing fault with the P-F curve. P is the point where a potential failure first becomes detectable; F is functional failure, the moment the part stops doing its job. The time between them is the window to investigate and act. The idea goes back to the 1978 Reliability-Centered Maintenance report by Nowlan and Heap, which names "inspect an item to detect a potential failure" as one of only four types of scheduled maintenance task. How long that window lasts depends on the failure mode, the operating conditions and how the fault is detected.

Vibration analysis, oil analysis, ultrasound and electrical testing catch different clues, and none always fires first. The DOE guide marks infrared thermography against almost every equipment category in its comparison table, from pumps and motors to circuit breakers, transformers and piping. The useful question for each asset is which warning can be detected reliably. For a great many electrical and mechanical faults, heat is a strong answer.

What Heat Gives Away: Electrical, Mechanical, Process and Building Faults

Thermal inspection finds faults that change a surface temperature, across four broad domains. ASTM E1934 explains the two basic patterns: warm exceptions in electrical equipment usually come from increased resistance, and in mechanical equipment from friction; cool exceptions usually point to failed components.

  • Electrical. What may be wrong: Loose or deteriorated connections, overloads, load imbalance, faulty or mismatched components. What thermal inspection may reveal: A local hot spot, or an unexpected difference between comparable components. A failed component may show up cooler.
  • Mechanical. What may be wrong: Bearing wear, misalignment, belt slip, poor lubrication. What thermal inspection may reveal: Unusual heating around a bearing housing, coupling, belt or gearbox, best read as a trend against its own baseline.
  • Process. What may be wrong: Failed steam traps, passing valves, blocked lines, refractory damage. What thermal inspection may reveal: Temperature patterns that do not match the expected operating state.
  • Buildings. What may be wrong: Insulation gaps, air leakage, roof moisture. What thermal inspection may reveal: Surface patterns worth investigating, but not proof of moisture or a hidden defect on their own.

Not every failure ends in smoke. A failed steam trap just leaks money quietly: the DOE's steam tip sheet notes that in systems not maintained for three to five years, 15% to 30% of steam traps may have failed, against under 5% where a regular inspection program runs. A worn bearing grinds itself down without drama. Heat gives both away.

The common thread is comparison. Why is one connection warmer than its neighbors? Why has this bearing changed since its baseline? For electrical systems, the DOE guide puts it plainly: the main value is locating problems so they can be diagnosed and repaired, and "How hot is it?" is usually of far less importance. Those questions only pay off when load, surface properties and measurement conditions are understood. The patterns are clues to investigate, not diagnoses to read from color alone.

Who Thermal Inspection Is Not For

Thermal inspection is not worth running on every asset, and some sites are better served by buying it as a service. Four situations call for something else:

  • Low-consequence equipment. Small parts, non-critical or redundant equipment can reasonably run to failure, as the DOE guide's reliability-centered maintenance table recommends. Inspecting them costs more than it saves.
  • Faults without a thermal signature. Plenty of failure modes do not change a surface temperature in time to matter. ASTM E1934 is explicit that an infrared examination "does not assure" proper operation and that other tests and maintenance are still needed.
  • No commitment to training and follow-up. The DOE guide advises against buying predictive equipment for in-house use without a serious commitment to training, implementation and repair, and suggests contracting a survey service instead, which it calls the most cost-effective route for smaller, less maintenance-intensive facilities.
  • Assets that need constant watching. Where a fault can develop between rounds and the consequence is severe, permanently installed sensors do the job better than any person with a camera.

A Camera Is Not a Crystal Ball: Limits and Safety

Thermal imaging is powerful and easy to oversell. Its limits belong in the maintenance plan, not the small print.

  • Emissivity and reflections matter. Shiny metal reflects surrounding infrared and can read far cooler than it is, while sunlight, nearby hot equipment and the inspector's own body all show up as false spots. ISO 18434-1 includes procedures for compensating for emissivity and reflected temperature; in the field, changing the viewing angle from a safe position helps separate a reflection from a real hot spot.
  • Load changes the story. A lightly loaded connection may barely warm up even with a defect. Inspect under representative load and record it. A cool image of idle equipment is not a clean bill of health.
  • It needs a clear view. Thermal cameras cannot see through metal panel doors or ordinary glass. An infrared window gives access without opening the enclosure, but its transmission and coverage must be accounted for, as must distance and resolution for small targets.
  • It is a snapshot. ASTM E1934 notes that an infrared examination provides data about the equipment at the time of examination only, and that it is not remedial in itself.
  • A hot spot is a finding, not a repair order. Confirm it, find the cause and judge the urgency. Under ASTM E1934, opinions on causes and corrective action require knowledge beyond thermography itself.

One rule sits above all of these: the camera does not make live equipment any safer. It does not remove shock or arc-flash hazards, replace required protective equipment, or authorize anyone to open an energized enclosure. Every inspection stays inside the site's electrical safety procedures: in the U.S., OSHA 29 CFR 1910.333 limits work on energized parts to qualified persons; in Europe, EN 50110-1:2023 sets the requirements for any work on, with or near electrical installations.

Standards and Qualifications for Thermal Inspection

A program built on recognized standards is easier to defend to auditors, insurers and management. The documents most often referenced are:

  • NFPA 70B: Electrical equipment maintenance programs, including infrared inspection. A standard with mandatory requirements since the 2023 edition; check which edition and obligations apply to your facility.
  • ISO 17359:2018: General procedures for setting up a condition monitoring program for any machine
  • ISO 18434-1:2008: Thermography for machinery condition monitoring: procedures, severity criteria, emissivity and reflected-temperature compensation, safety
  • ISO 18436-7:2014: Qualification and assessment of thermography personnel, in a three-category classification
  • ASTM E1934-99a(2024): Responsibilities of the end user and the thermographer, and documentation for qualitative and quantitative inspections
  • OSHA 1910.333 / EN 50110-1: Electrical safety work practices for anyone working on or near energized equipment

Personnel qualification usually follows ISO 18436-7 or the ASNT Level I to III scheme, which the DOE guide describes as running from competent equipment use (Level I), through full diagnostics and recommendations (Level II), to supervising and teaching (Level III).

Handheld, Fixed or Wearable: Choosing the Inspection Method

So why did the breaker still fail? Not because the warning was invisible. Because nobody was looking when it appeared. A scheduled survey is valuable evidence, but it is a snapshot, and faults develop or only show up under different loads in the months between visits. A sharper sensor does not fix a timing problem.

Any thermal program can be tested with four plain questions:

  • Can the equipment be seen from an approved inspection position?
  • Is thermal equipment on hand during ordinary rounds, not just on survey day?
  • Does the technician know which changes to escalate?
  • Does every finding reach someone responsible for acting on it?

A hot spot saved in an image folder is not a maintenance action. The answer to the coverage problem is more chances to notice a change, with a clear response path, and three methods provide them. They are different tools for different jobs, and most programs use more than one.

  • Handheld camera. Best for: Formal surveys, detailed radiometric measurement, documentation and reporting. Limits: Occupies a hand; a separate task outside ordinary rounds.
  • Fixed sensors. Best for: Continuous watching of selected critical assets between inspections. Limits: Cost per point; only covers what it is aimed at.
  • Wearable. Best for: Thermal checks built into everyday rounds, hands-free, with the thermal layer in the inspector's own view. Limits: Battery and imaging range set the job: ThermalGlass Explorer runs up to two hours, −40 °C to 330 °C, so it suits targeted rounds, not continuous monitoring.

Software ties the methods together: baseline images and recorded operating conditions turn snapshots into trends. AI-assisted analysis can flag changes but still needs comparable data and qualified review, and linking findings to a CMMS is an integration decision of its own.

Wearable thermal inspection with ThermalGlass Explorer

Wearables are the shift Holoindustry cares most about: bringing a thermal view into the rounds technicians already walk, so checking for heat stops being a separate handheld-camera task. ThermalGlass Explorer pairs a 240 × 320 thermal camera, imaging at up to 27 Hz with a typical sensitivity of 65 mK, with a see-through display and Fused Vision™, Holoindustry's technology that overlays real-time thermal data onto the wearer's natural field of view. Edge detection sharpens outlines, so the overlay shows which breaker or bearing is warm, not just a colored blob. Temperature thresholds can trigger alerts, and a companion app handles settings, streaming and recording. A short video of ThermalGlass Explorer on a factory floor shows the workflow.

Two boundaries apply. Hands-free does not mean attention-free: stop in a safe position before inspecting, and never read an overlay while climbing or crossing traffic. And ThermalGlass Explorer's published imaging range is −40 °C to 330 °C with up to two hours of battery, so it suits targeted rounds, not whole-shift monitoring or furnace shells. Wearables fill the gaps between formal surveys and fixed sensors; they do not replace either.

From Finding to Fix: The Detect-to-Repair Loop

A thermal program only prevents failures if every finding runs through a complete loop. Detection is the beginning, not the outcome.

  1. Observe. From an approved position, notice the pattern and record the asset, location and operating conditions.
  2. Assess. A qualified person confirms the finding, investigates the cause and sets the priority. Some findings need immediate escalation or shutdown, not a planned delay.
  3. Act. Assign an owner and a work order, then repair under the required isolation procedures.
  4. Verify. Re-inspect under comparable load and record whether the anomaly is gone.

Replay the 3:12 a.m. scene with that loop in place. Three weeks earlier, on an authorized round, a technician checks the breaker through an infrared window and sees one connection running warmer than its neighbors. A qualified electrician confirms the problem, and the repair is booked for a controlled, locked-out outage that weekend. A follow-up inspection under load shows the hot spot is gone. The night shift never sees the emergency. That is how good maintenance should feel: uneventful.

Not every fault gives that much warning, and some demand action today. The point is not a guaranteed easy fix. It is the chance to make the decision yourself, before the failure makes it for you.

How to Start a Thermal Inspection Program

A thermal inspection program can start small and grow with evidence. ISO 17359 describes the general steps for setting up condition monitoring; in practice, five moves cover most of it:

  1. Choose ten high-priority assets. Identify the failure modes worth watching, safe inspection positions and baseline conditions.
  2. Add thermal checks to approved rounds. Match frequency to risk and expected deterioration, not convenience.
  3. Give every finding a destination. Record the asset, evidence, conditions, owner and next action in your maintenance process.
  4. Train for interpretation, not color. Teach comparison, measurement limits and when to call a specialist.
  5. Keep the wider program. Formal surveys, other condition monitoring methods and any inspections your site requires all stay.

The DOE guide estimates that a facility can expect thermography to pay back in 12 months or less on average, faster where downtime costs thousands of dollars an hour, and suggests that smaller sites use a contracted survey service instead. The figures date from 2010, so treat them as a direction, not a forecast.

Frequently Asked Questions

How often should thermal inspections be done?

Thermal inspection frequency should follow risk, not habit. Critical assets with fast-developing failure modes need more frequent checks than stable, low-consequence ones, and for electrical equipment NFPA 70B includes maintenance interval guidance. Steam systems are a good example of risk-based timing: the DOE recommends testing high-pressure steam traps weekly to monthly and low-pressure traps annually.

What temperature difference counts as a problem?

No single temperature difference applies to every asset. A difference matters when it appears between comparable components under similar load, or when it grows against the asset's own baseline. ISO 18434-1 gives guidance on setting severity criteria for each application, and a qualified person should judge urgency before any repair is scheduled.

Is thermal imaging better than vibration analysis?

Thermal imaging and vibration analysis answer different questions. Thermography covers a wide range of electrical, mechanical and process equipment quickly, while vibration analysis goes deeper into rotating machinery. Most condition monitoring programs use both, chosen asset by asset according to which warning can be detected reliably.

Do thermographers need certification?

Thermographers are usually qualified under ISO 18436-7 or the ASNT Level I to III scheme, and electrical safety rules require that anyone working near energized equipment is qualified for that work. Which certification a site needs depends on its standards, insurer and internal procedures.

What ROI can a plant expect from thermal imaging?

Return depends on downtime cost and how the program acts on findings. The U.S. Department of Energy's O&M guide estimates an average thermography payback of 12 months or less, and savings of 8% to 12% for predictive maintenance over a preventive-only program. Both figures are older estimates; a plant's own avoided failures are the better measure.

Keep the Smoke In

The old electricians were right, in their way. Everything does run on smoke, and maintenance is the craft of keeping it in. Heat talks first. Smoke talks last. A thermal inspection program decides whether someone is listening to the first one.

Want to see what your next inspection round looks like in thermal? Explore ThermalGlass Explorer or request a demonstration from Holoindustry on your own equipment.

Sources

  1. U.S. Department of Energy, Federal Energy Management Program. O&M Best Practices Guide, Release 3.0, Chapter 5: Types of Maintenance Programs (2010).
  2. U.S. Department of Energy, Federal Energy Management Program. O&M Best Practices Guide, Release 3.0, Chapter 6: Predictive Maintenance Technologies (2010).
  3. Nowlan, F. S.; Heap, H. F. Reliability-Centered Maintenance. United Airlines for the U.S. Department of Defense (1978).
  4. ASTM International. ASTM E1934-99a(2024), Standard Guide for Examining Electrical and Mechanical Equipment with Infrared Thermography.
  5. ISO. ISO 18434-1:2008, Condition monitoring and diagnostics of machines, Thermography, Part 1: General procedures (confirmed 2023).
  6. ISO. ISO 18436-7:2014, Requirements for qualification and assessment of personnel, Part 7: Thermography (confirmed 2025).
  7. ISO. ISO 17359:2018, Condition monitoring and diagnostics of machines, General guidelines (confirmed 2023).
  8. NFPA. NFPA 70B, Standard for Electrical Equipment Maintenance.
  9. OSHA. 29 CFR 1910.333, Selection and use of work practices.
  10. CENELEC. EN 50110-1:2023, Operation of electrical installations, Part 1: General requirements.
  11. U.S. Department of Energy, Advanced Manufacturing Office. Steam Tip Sheet #1: Inspect and Repair Steam Traps (2012).
  12. Holoindustry. ThermalGlass Explorer: product page and technical parameters.