A transformer has no single most important part: it is a set of sub-systems, and a failure in any of them takes it off line. This video summarises 42 technical webinars on temperature, gases, moisture, bushings, on-load tap changers and through faults: what each measurement tells you and what it costs to ignore it.
Key points
- The winding hot spot is calculated, not measured, and under the IEC loading guide every 6 °C above 98 °C doubles the ageing rate of the paper.
- Each fault temperature produces its own gas; hydrogen, the least soluble, is the first to appear.
- Almost all the water is in the paper, not the oil, and what shows in the oil depends on temperature.
- A bushing failure is sudden and about half the time ends in a fire; online monitoring relies on three healthy currents cancelling out.
- The on-load tap changer is the only moving part: the energy its motor draws on each tap change portrays the mechanism.
Only as good as its worst sub-system
A transformer is many sub-systems, not one machine, and a failure in any one of them can trip the unit off line. Its condition is only as good as its worst-performing sub-system, so a monitor that watches one thing well is still blind. The more failure modes covered, the higher the confidence in the verdict.
The hot spot sets the life expectancy
The measurement that governs the life of the unit is the winding hot spot, and no probe in the oil reaches it. It is worked out from three things: the top-oil temperature measured in the oil, the load current measured through the CTs, and the winding gradient from the heat-run test. The sessions are clear about how wrong the old gauges can be: in one case, a unit three months in service had the winding gauge reading 50 °C while the oil gauge read 65 °C.
The cost of getting it wrong is not linear. Under the IEC loading guide, at a steady 98 °C the transformer ages one year per year; at 110 °C it ages at four times that rate; at 128 °C, thirty-two times, so two hours become two and a half days. Even below 50 °C there is a minimum ageing rate of about a day a year. Directly measured fibre-optic probes are now standard practice.
Dissolved gas: each gas names a temperature
Gas is the second pillar, and the sessions teach it as a thermometer: the gas that appears tells you how hot the fault is.
| Fault temperature | Characteristic gas |
|---|---|
| Above 150 °C | hydrogen and methane |
| Above 250 °C | ethane |
| Above 350 °C | ethylene |
| Above 500 °C | acetylene: arcing |
| CO and CO₂ | the paper, not the oil |
| A falling CO₂/CO ratio | cellulose is degrading |
Hydrogen appears in every thermal condition, which is what makes a single-gas monitor a useful smoke detector even though it cannot name the fault.
Why hydrogen first
Hydrogen is the least soluble of the diagnostic gases, so it leaves the oil and collects in the headspace at roughly 20 to 1 against the dissolved concentration in normal operation. Its dispersion in oil is uneven and depends on circulation, which is why a spot sample can mislead and a trend cannot. Judge the alarm by rate of change, not by the absolute number: high hydrogen over a short period generally means active arcing.
Moisture: the water is in the paper, the oil only reports it
This is the subject the sessions return to most often. In a 50 MVA transformer with 1.5 % water in the paper, almost all the water is in the paper, and how much appears in the oil depends entirely on temperature:
| Oil temperature | Water in oil | Water in paper | Oil reading |
|---|---|---|---|
| 20 °C | 0.10 kg | 15.84 kg | 10 ppm |
| 40 °C | 0.21 kg | 15.73 kg | 20 ppm |
| 60 °C | 0.52 kg | 15.43 kg | 49 ppm |
| 80 °C | 1.45 kg | 14.49 kg | 137 ppm |
Same transformer, same water: ten ppm or a hundred and thirty-seven, depending on when you sampled. And the moisture in the paper decides three practical limits: the bubbling temperature (the real limit on how far you can overload), the dielectric strength of the barriers at the bottom of the winding, and the ageing rate of the winding insulation itself. Online relative saturation tracks it continuously and corrects for temperature; a laboratory ppm figure alone cannot be used in a partition model.
Bushings: sudden, and they take the transformer with them
Bushings take six of the 42 sessions because of the severity. About half of bushing failures cause a fire that destroys the transformer; around 80 % occur after ten to twelve years in service; recommended offline testing is every four to five years, and events happen between those tests.
The online measurement works because of the construction of the bushing itself. Capacitive foils divide the voltage evenly; C1 is the capacitance from the centre conductor to the test tap, the one that ages, and C2 runs from the tap to earth and is in circuit only once the tap is monitored. The test tap was always there, normally just bolted to earth. Tan delta rises with moisture and with temperature.
What the monitor looks for is balance. Three healthy phases produce three currents that cancel; a defect changes one phase’s magnitude and angle, and the sum stops being zero. From the imbalance come power factor and capacitance. Dry, sound insulation has almost no temperature dependency; contaminated insulation does, and steeply. The catch: breakdown products are conductive and settle along the inside of the lower porcelain, where an offline test at 10 kV may not be enough voltage to see contamination that shows plainly at rated voltage.
The on-load tap changer: the only part that moves
The failure statistics from one large utility make the point. Flashover in the tap changer was the single largest category, and 14 of those 28 cases went on to destroy a winding. Mechanical failure of the tap changer led to further winding failures, as did overheating of connections, contamination and hot spots on connections — all of it developing inside a compartment nobody opens.
There are four ways in, with no outage: the motor energy (joules per tap change), the temperature differential between the compartment and the main tank, tap counters per position rather than just a total, and DGA in the compartment, with ratios that name the fault. The cleverest is the motor energy: most mechanical problems in the drive change the power the motor draws, so the joules per tap change become a picture of the linkage, the gears and the lubrication.
Damage that accumulates from outside
Two more stresses come from outside the transformer. Long through faults raise thermal degradation exponentially; degraded insulation loses volume and the winding loses clamping force; less clamping means more displacement and less withstand next time. And geomagnetic disturbance part-saturates the core and generates harmonics, whose tell-tale is even harmonics rising above the odd ones.
What the 42 sessions keep coming back to
- The hot spot governs the life, and it is calculated, not measured.
- Gas names the temperature of a fault; rate of change names the urgency.
- The water is in the paper; an oil ppm without its temperature is not a measurement.
- Bushings and tap changers fail suddenly, and both can be watched in service.
- Identical units in parallel behave differently. Treat each as an individual.
Related equipment
Dynamic Ratings covers these sub-systems with the E3 transformer monitor, the B100 electronic temperature monitor and the C50 bushing monitor; the GRIDSCAN5000 measures dissolved hydrogen continuously. The rest of the range is under transformer testing and monitoring. To discuss your fleet, contact our engineers.