Industries / Industrial equipment and services / Industrial services / Thermography
Electrical surveys
An electrical survey points a heat camera at switchgear, panels, motor control centers, transformers and substation connections while they carry current, to find joints and parts that run hotter than they should.
5
Milestones achieved, from 1970s to 2011.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1970s
Scans moved to the dark
Most thermal cameras sold before 1980 sensed mid-wave infrared, 3 to 5 micrometers (µm) in wavelength, where sunlight glinting off metal can look like a hot spot. Electric utilities, among the first users, scheduled their outdoor scans for early morning or after sundown, and in 2013 a FLIR sales director wrote that some still do.
LessonThe US Bureau of Reclamation's 2011 thermography manual warns that night and off-peak hours also bring lighter loads, which "may not adequately heat a problem area".
Credit
electric utilities, as told by Ed Kochanek of FLIR Systems
1988
The Navy writes survey rules
MIL-STD-2194(SH), a US Navy standard dated 12 February 1988, set out how to survey a ship's electrical equipment with a thermal imager: at the dock first, then under way for gear that runs only at sea, with each item in service under normal load for at least half an hour beforehand. It called any spot 10 °C or more above the equipment's reference temperature a hot spot and coded severity in bands: 10 to 25 °C above reference, fix at the next routine maintenance; 25 to 40 °C, failure probable; 40 to 70 °C, failure almost certain; 70 °C and above, failure imminent, so stop the survey and tell the officers in charge. The Navy cancelled the standard without replacement on 9 March 1995.
Credit
US Navy
2000
Wind cools the hot spots
Moving air carries heat away from a surface, so wind or a fan lowers the temperature a camera sees. At SPIE's Thermosense XXII conference in 2000, Robert P. Madding and Bernard R. Lyon Jr. of FLIR's Infrared Training Center reported a simple test, blowing air on a hot spot on a fuse cutout (the fuse holder on a power pole), and wrote that thermographers "do not know how to compensate" for the cooling. The Bureau of Reclamation's 2011 manual, which draws on their work, says there is "no reliable way to correct for wind quantitatively" and that "there are no simple load correction factors"; the factors it reprints from a 2002 paper by Madding and colleagues multiply a temperature rise read at 50% load by 3, and one read at 30% load by 7.
LessonA reading taken in wind or at light load can make a serious fault look minor, so the load and the wind at the time belong in every report.
Credit
Robert P. Madding and Bernard R. Lyon Jr
2005
An insurer scans its clients
From March 2005 to June 2010, thermographers at CNA, a US commercial insurer, ran 6,445 infrared surveys of clients' electrical systems with 22 FLIR P65 cameras and logged 47,077 faults, 7.3 per report; 3,067 of the surveys found at least one fault that CNA rated serious or critical. CNA's own estimate is that the program saved clients about $4 for every $1 of survey. Its bands are temperature rises above normal of 1 to 9 °F for minor, 10 to 34 °F intermediate, 35 to 74 °F serious, and 75 °F or more critical.
LessonSeverity bands are conventions, not physics: a bulletin from Global Risk Consultants, a risk consulting firm in New Jersey, uses nearly the same numbers in degrees Celsius (1 to 10, 10 to 35, 35 to 75, and 75 °C or more), so a 75 °F rise, about 42 °C, that CNA calls critical is only serious on its scale.
Credit
Thomas Gray of CNA Insurance, in a paper for FLIR's InfraMation conference
2011
Windows instead of open doors
The US Bureau of Reclamation, which calls itself the country's second-largest hydropower producer, set out its thermography practice in November 2011 in a manual for its power plants (FIST volume 4-13). It says that opening panels to scan live parts "raises safety concerns and may upset the thermal equilibrium", and recommends IR windows, viewing panes made of a material that lets infrared through, or small ports, listed by Underwriters Laboratories (UL) and meeting the Institute of Electrical and Electronics Engineers' (IEEE) requirements for viewing panes; ordinary glass and some plastics block infrared. For severity it adopts the table in the InterNational Electrical Testing Association's (NETA) 2005 Maintenance Testing Specifications, which calls a part 1 to 3 °C hotter than a similar part under similar load a possible deficiency, and more than 15 °C hotter a major deficiency to repair immediately, and it uses those values, it says, "in the absence of consensus standards".
Credit
Bureau of Reclamation, Hydroelectric Research and Technical Services Group; NETA
Watch for
Always-on cameras in a switchyard
Flir and its partner Hi-Tech say fixed thermal cameras on pan-tilt mounts, with software that maps each transformer, circuit breaker and connection and raises an alarm when a set temperature is passed, now watch the 220 kV switchyard of NALCO's aluminum smelter at Angul around the clock. They report about 20 hours a year less outage time and a full return on the cost in under two years.
Achieved whenNALCO or an outside auditor publishes the switchyard's outage records for the years before and after, with the faults the cameras caught.
Claimed by
Flir, June 2026
Crystal windows that last
An IR window's optic must let infrared through for the life of the panel, and the camera must be set for the share it passes, its transmission. In a June 2026 article Flir says its calcium fluoride (CaF2) windows "naturally resist moisture absorption, preventing degradation over time". IRISS, another window maker, wrote that a CaF2 window in an air-conditioned switchgear cabinet had "zero infrared or visual transmission" after two years, and in 2009 Joe DeMonte of TEGG Corporation, an electrical maintenance firm, found that one CaF2 window had lost about 69% of its transmission in two years in an office.
Achieved whenan independent laboratory ages windows of each material from several makers at controlled humidity and publishes their transmission over time.
Claimed by
Flir, June 2026
Open
01
Age IR windows in public
A year-long bench test of the IR windows sold for switchgear, with no electrical equipment involved. Buy several windows of each optic material on sale, such as plain and coated calcium fluoride and polymer, keep sets in sealed boxes at fixed humidity, and once a month measure how much infrared each passes by imaging a steady hot target through it and without it. Publish every window's transmission over time and the temperature error a thermographer would make by keeping the value set when the window was new.
The briefCredit · unclaimed02
Open wind and load data
A public data set from a bench rig that measures how a faulty bolted joint's temperature rise changes with current and with wind, so the corrections thermographers use can be checked. Copper bus bar joints are bolted at the correct torque and at set levels of looseness or with corroded faces, carry direct current from a supply that stays below 10 V so nothing on the bench can give a shock, and are filmed with a thermal camera while a fan blows air across them at measured speeds.
The briefCredit · unclaimed
Open 01 · the brief
Age IR windows in public
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A year-long bench test of the IR windows sold for switchgear, with no electrical equipment involved. Buy several windows of each optic material on sale, such as plain and coated calcium fluoride and polymer, keep sets in sealed boxes at fixed humidity, and once a month measure how much infrared each passes by imaging a steady hot target through it and without it. Publish every window's transmission over time and the temperature error a thermographer would make by keeping the value set when the window was new.
Why it is still open
The makers disagree, and the one measured test is small. Flir wrote in June 2026 that CaF2 resists moisture; IRISS, in an article by Martin Robinson, reports a CaF2 window going opaque within two years and advises against crystals. DeMonte's 2009 test covered one CaF2 window and one polymer window, both kept two years in an office: through the CaF2 window, with the setting it had when new, a target at about 45.6 °C read 28.0 °C, and one at 349.6 °C read 252.1 °C, while the polymer window read both within 0.5%. UL 50V, the standard for infrared viewports, leaves suitability to "the conditions that prevail in actual service", as DeMonte quotes it. Web and GitHub searches in November 2026 found no independent aging test.
Built with what exists
IR windows bought from makers such as IRISS and Flir; sealed plastic boxes held at fixed humidity over saturated salt solutions, a standard laboratory method, with a humidity logger in each; a temperature-controlled hot plate painted with high-emissivity paint, or a blackbody calibrator, as the target; a radiometric thermal camera (one that records a temperature for every pixel); Python for the analysis.
First step
Buy one CaF2 and one polymer window, measure each against a hot plate at 50 °C and at 150 °C, seal one of each in a box over water, and measure them again four weeks later.
Done looks like
At least four window types from at least two makers, three samples of each, at three humidity levels (about 30%, 60% and 90% relative humidity), measured monthly for 12 months at two target temperatures; all images and a table of transmission by window and month published under an open license, with the share of windows whose transmission fell by more than 10 percentage points.
Open 02 · the brief
Open wind and load data
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public data set from a bench rig that measures how a faulty bolted joint's temperature rise changes with current and with wind, so the corrections thermographers use can be checked. Copper bus bar joints are bolted at the correct torque and at set levels of looseness or with corroded faces, carry direct current from a supply that stays below 10 V so nothing on the bench can give a shock, and are filmed with a thermal camera while a fan blows air across them at measured speeds.
Why it is still open
Madding and Lyon's 2000 wind test used one fuse cutout, and its numbers sit in a paywalled SPIE paper. The Bureau of Reclamation's 2011 manual says there is no reliable wind correction and no simple load correction, and reprints one set of load factors from a 2002 conference paper. Susanna Spinsante and colleagues at the Polytechnic University of Marche, with IMESA, an Italian switchgear maker, loosened six busbar joint screws in a 12 kV switchgear from the recommended 45 Nm of torque down to 4 Nm and measured temperature at 27 points, at one current, the rated 630 A, with no wind; they write that continuous temperature records of switchgear are "extremely rare or even not existing" (2022). GitHub and Hugging Face searches in November 2026 found no data set of electrical hot spots measured against controlled current and wind.
Built with what exists
A commercial low-voltage, high-current DC power supply of a few hundred amps; copper bus bar, bolts and a torque wrench; thermocouples and a logger such as Pico Technology's TC-08; a hot-wire anemometer and a variable-speed fan; a radiometric thermal camera; a guard around the rig, since the bars get hot; Python to fit how temperature rise grows with current.
First step
Bolt two bus bar pieces with one loose bolt, run the joint at 50%, 75% and 100% of the supply's current with no wind, and log it beside a sound joint until both settle.
Done looks like
At least six joint conditions, each at five currents (20% to 100% of the bar's rating) and four wind speeds (0 to 5 m/s), held until temperatures settle; all images and logs released under an open license; and for each joint condition, the fitted relation between current and temperature rise, with its spread, set beside the factors in Reclamation's manual.