Hot Spot Analysis of Transformers
Theory: from losses to temperature to life
Overview
How much does running a transformer slightly over rating affect its life?
More than you'd think. Transformer life is set by how fast the paper insulating the windings degrades and becomes brittle with heat, and that rate depends exponentially on temperature: for non-upgraded paper it doubles every 6 K. More load raises copper loss with current squared, lifting both oil temperature and the winding-to-oil gradient. Computing the hottest point (hot spot) tells how fast life is consumed. Detailed work finds loss distribution by electromagnetic analysis and oil flow and temperature by CFD, but start with the IEC 60076-7 equations for a feel.
Hot-spot temperature and ageing rate
θ_a is ambient, Δθ_or rated top-oil rise (52 K), R the ratio of load to no-load loss at rating (6), K load factor, H hot-spot factor (1.3), g_r rated average winding-to-oil gradient (20 K), x = 0.8 and y = 1.3 (natural oil and air cooling example). V is the relative ageing rate of non-thermally-upgraded paper.
The 6 K doubling rule
That insulation-paper degradation accelerates with temperature was shown experimentally by Montsinger in the 1930s, spreading the rule of thumb that life halves every 6–8 K. It rests on the Arrhenius idea that reaction rates rise exponentially with temperature. IEC 60076-7 now uses a doubling every 6 K referenced to 98 °C for non-upgraded paper, with a different formula and higher reference for thermally upgraded paper.
Worked examples
Example 1: 20 °C ambient
| Load factor K | Top-oil temperature | Winding gradient | Hot spot | Ageing rate V |
|---|---|---|---|---|
| 0.8 | 58.7 ℃ | 19.5 K | 78.2 ℃ | 0.10 |
| 1.0 | 72.0 ℃ | 26.0 K | 98.0 ℃ | 1.00 |
| 1.2 | 87.2 ℃ | 33.0 K | 120.1 ℃ | 12.9 |
| 1.5 | 113.1 ℃ | 44.0 K | 157.2 ℃ | 929 |
Example 2: 30 °C ambient
| Load factor K | Hot spot | Ageing rate V |
|---|---|---|
| 0.8 | 88.2 ℃ | 0.32 |
| 1.0 | 108.0 ℃ | 3.17 |
| 1.2 | 130.1 ℃ | 40.9 |
| 1.5 | 167.2 ℃ | 2,950 |
At 1.5 times load you use two and a half years of life in a day. Surely that's not allowed?
This is the steady state if 1.5 times load continued indefinitely. Real oil temperature rises slowly with a time constant of hours, so short overloads don't drive the hot spot this high. IEC guidance therefore computes temperature over time and sums low- and high-ageing periods through the day to assess life consumption; night-time hours at K 0.8 with ageing rate 0.1 can offset modest daytime overload. But if the hot spot gets too high, moisture in the paper can form bubbles and cause dielectric failure regardless of ageing, so temperature itself has limits. Ambient matters too: on a 30 °C day even rated load triples the ageing rate.
Analysis workflow
- Set rated oil rise and winding gradient from heat-run test results.
- Find winding eddy-loss distribution by electromagnetic analysis to locate the hot spot.
- Use CFD of oil flow and winding temperature to confirm the hot-spot factor.
- Compute temperature over time and life consumption from daily load and ambient profiles.
- Calibrate against fibre-optic temperature measurements.
“Top-oil temperature looked normal”
In one transformer top-oil temperature stayed below its alarm limit, yet inspection found paper at the top of the winding more degraded than expected. CFD showed a winding section with narrow oil ducts where oil barely flowed; there the winding-to-oil gradient exceeded design assumptions, equivalent to a hot-spot factor of about 1.6 rather than 1.3. Oil-guiding parts were added and fibre-optic sensors installed to monitor the hot spot directly.
Common mistakes
Mistakes and fixes
| Mistake | Effect | Fix |
|---|---|---|
| Judging by top-oil only | Hot spot missed | Compute and monitor hot spot |
| Steady equations for short overloads | Ageing overestimated | Compute transient |
| Constant ambient | Summer ageing missed | Include seasonal variation |
| Fixed hot-spot factor | Local hot zones missed | Confirm with CFD |
| Confusing paper types | Wrong life | Choose formula by paper |
I'd like to learn related topics.
Related pages include induction heating, eddy-current heating, Joule heating, microwave heating and dielectric heating.