Industries / Industrial equipment and services / Industrial services / Motor rewinding
Insulation
Insulation is everything that keeps current inside the copper of a winding: the enamel on the wire, the sheets and tapes between coils and iron, and the varnish or resin that binds them, and a rewind replaces all of it.
6
Milestones achieved, from 1913 to 2005.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1913
Heat sets the rating
Lamme of Westinghouse and Steinmetz of General Electric wrote in a joint paper for the American Institute of Electrical Engineers (AIEE) that insulation is used up at a rate that depends on its operating temperature, and their work became the basis of the temperature classes in the AIEE's Standard No. 1, according to a 1962 paper from Westinghouse. Those classes were defined by what a material was made of, which did not fit the new synthetic materials, so in 1955 to 1957 the International Electrotechnical Commission (IEC) and the AIEE rewrote them to rest on "accepted test and experience", and the IEC issued the result as Publication 85 in 1957.
Credit
Lamme (Westinghouse) and Steinmetz (General Electric); the IEC and the AIEE
1948
Aging treated as chemistry
In 1930 V. M. Montsinger reported laboratory aging tests on class A insulation, in air and in oil, showing that the rate of aging roughly doubled for each 8 °C rise in temperature. In 1948 Thomas W. Dakin showed that thermal aging follows the theory of chemical reaction rates, and replaced what he called the "approximate 7 to 10 rule" with a formula for carrying the results of short aging tests at high temperature over to the lower temperatures of service.
Credit
V. M. Montsinger; Thomas W. Dakin
c. 1950
A hard groundwall
The groundwall is the main insulation between a coil and the grounded iron around it. From about 1915, mica and cotton tapes were soaked with hot asphalt using a vacuum and pressure cycle, and these asphalt-mica systems served for about 35 years; about 1950 Westinghouse developed Thermalastic, a polyester resin system that the Combined Cycle Journal calls the first "hard" groundwall, and other makers followed with their own. Westinghouse later replaced the polyester with epoxy, which its engineers reported gave a mechanically stronger coil that stood up better to heating and cooling.
Lessonevery hard groundwall system still relies on mica, because, in the journal's words, no man-made product comes near mica's resistance to partial discharge.
Credit
Westinghouse
1990s
Stators sealed in one piece
Siemens introduced global vacuum pressure impregnation (GVPI) for generator stator windings, in which the complete wound stator, rather than each coil, is soaked in resin under vacuum and pressure. The Combined Cycle Journal reports major cost advantages and good quality, and adds that if such a winding fails, it is difficult to rewind.
Lessona process that makes a winding cheaper to build can make it harder to repair.
Credit
Siemens
1995
Drive pulses wear out wire
A variable-speed drive, also called an inverter, feeds a motor a train of steep voltage pulses rather than a smooth wave, and reflections in the cable can raise them further: in an example from the utility PG&E, a 460 V motor saw pulses of about 1,200 V at its terminals. In 1995 Weijun Yin and colleagues reported that ordinary magnet wire insulation lasts a short time under such pulses at high voltage and temperature, and in 1997 they described an improved wire for inverter-fed motors; NEMA's motor standard MG 1 Part 31 requires low-voltage inverter-duty windings to survive an impulse of 3.1 times their rated voltage. PG&E's guidance says field experience shows most premature failures under drives were in motors rated under 25 hp.
Lessonthe drive and its cable, not only the nameplate voltage, set the stress a winding has to carry.
Credit
Weijun Yin, K. D. Bultemeier, D. Barta and D. Floryan; NEMA
2005
Two kinds of drive winding
During 2005 the IEC split its draft rules for windings fed by drives in two. Type I windings, usually low-voltage random-wound stators whose round wire lies loosely in the slots, must have no partial discharge at any point in their life, so test coils are aged by heat, vibration and humidity and then checked for discharge (IEC 60034-18-41); Type II windings, usually higher-voltage form-wound coils, are expected to see discharge and must resist it, mainly with mica (IEC 60034-18-42). Engineers from GE and Iris Power wrote in 2007 that these were the first complete documents for judging insulation for use with inverters.
Credit
IEC; Meredith Stranges, Greg Stone and Dennis Bogh
Watch for
Nanoparticles in mica tape
TMEIC says it has developed what it calls the world's first nanocomposite mica insulating tape, with nanoparticles of about 1 to 100 nm spread evenly through it, so the insulation layer is thinner and the coil holds more copper. It says an 800 kVA, 6.6 kV generator built with the tape in September 2025 was about 10% smaller, with 20% less copper loss and 0.2% higher efficiency than its conventional models, and that the tape can be used when stator coils are replaced in existing machines, at local service shops.
Achieved whena customer or test laboratory outside TMEIC publishes measured losses or insulation life for a machine built or rewound with the tape.
Claimed by
TMEIC, December 2025
Open
01
Test the ten-degree rule
A public table of published thermal endurance results for motor insulation, from tests on materials, on magnet wire and on whole systems built as motorettes (small test windings), giving the life measured at each test temperature, and for each one the temperature rise that halves its life, fitted by the chemical-rate method Dakin set out in 1948.
The briefCredit · unclaimed02
Rewinds that run on drives
A public table of rewound motors that run on variable-frequency drives, built from repair shops' job records: each motor's size and voltage, the drive and cable length where the customer reported them, the magnet wire and insulation used in the rewind (standard or inverter-grade), the number of varnish treatments, and whether and when the same motor, matched by serial number, came back with another winding failure, with survival curves that compare the materials.
The briefCredit · unclaimed
Open 01 · the brief
Test the ten-degree rule
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public table of published thermal endurance results for motor insulation, from tests on materials, on magnet wire and on whole systems built as motorettes (small test windings), giving the life measured at each test temperature, and for each one the temperature rise that halves its life, fitted by the chemical-rate method Dakin set out in 1948.
Why it is still open
Trade bodies still state the rule as one number. The UK's Association of Electrical and Mechanical Trades (AEMT) wrote in January 2022 that "for every 10°C of additional heat to the windings, motor insulation life is cut in half". Montsinger's 1930 tests found the aging rate doubling every 8 °C, and Dakin in 1948 called it an "approximate 7 to 10 rule" and replaced it. Web and Crossref searches found restatements of the rule, and single studies, but no compiled table of published endurance results.
Built with what exists
The Crossref and OpenAlex APIs to find papers, and free full-text reports from NASA's technical reports server, DTIC and OSTI; a model built for reading long scanned reports, such as Muse Ember (see the Note "The jobs agents repeat"), to pull life-versus-temperature tables; Python with SciPy for the fits; a spreadsheet.
First step
Collect the free NASA and DTIC motorette reports (for example NASA's 2024 and 2025 winding-lifetime papers and DTIC report AD0744951 from 1968) and enter their curves by hand to settle the fields.
Done looks like
At least 100 endurance curves from at least 40 sources entered, each with its material or system, test method and thermal class; a published median and range of the halving interval by class; a second person checks 20 entries against the original papers.
Open 02 · the brief
Rewinds that run on drives
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public table of rewound motors that run on variable-frequency drives, built from repair shops' job records: each motor's size and voltage, the drive and cable length where the customer reported them, the magnet wire and insulation used in the rewind (standard or inverter-grade), the number of varnish treatments, and whether and when the same motor, matched by serial number, came back with another winding failure, with survival curves that compare the materials.
Why it is still open
The advice on rewinding drive-fed motors is about materials, not measured results. The US Department of Energy's tip sheet on inverter-duty motors says premature insulation failures began when fast-switching drives were introduced and that spikes have reached 2,150 V in a 480 V system; PG&E's guidance says most premature failures seen in the field were in motors under 25 hp; EASA's 2021 webinar on inverter-duty rewinding teaches shops the materials and processes for such windings. A November 2026 search found guidance and forum advice, but no published data from shops or plants on how long rewound drive-fed motors last with each choice of materials.
Built with what exists
Anonymized job records from cooperating shops; a model that answers the same fixed questions about each job ticket, such as Gemini Decision 1 (see the Note "A thousand questions, one pass"), with a person checking a sample; Python with pandas and the lifelines survival-analysis library.
First step
Ask one shop to list five years of rewind jobs for customers who run drives, and mark by serial number which motors came back with another winding failure.
Done looks like
At least 400 rewound motors known to run on drives, from at least five shops, each followed for at least two years; the share that failed again within two years for each winding material, with 95% confidence intervals; a second person checks 40 records against the job files; data and code published.