Why can transformers exceed rating briefly?

Their large oil volume takes time to warm. The top-oil time constant is around three hours (180 min here), so raising load doesn't reach steady temperature at once. Sustained 1.5 load would drive the hot spot to 167 °C, but a 30-minute overload stops at 115.8 °C.

How much life does an overload cost?

Insulation paper ages twice as fast for every 6 K rise in hot-spot temperature. Here a 1.5 overload for 30 minutes, including 3 hours afterwards, used 2.8 hours of life at the 98 °C reference; 1 hour used 10.5 hours and 2 hours 61.5 hours. Normal 0.7 load uses about 2.7 hours per day, so a 1-hour overload equals about 4 days.

What is needed to compute an overload?

Rated values, load exponents and time constants for top-oil rise and winding gradient. Oil time constants are hours and winding ones minutes, so both are solved in time. Judging by steady-state formulas alone either over-restricts short overloads or misses the danger of long ones.

Transformer Hot Spot Analysis

Category: Coupled Analysis – Electromagnetic-Thermal | Revised 2026-10-01

Theory: two time constants, oil and winding

Overview

🙋

On summer evenings when demand peaks, is it safe to load a transformer above its rating?

🎓

Often yes, depending on duration and temperature. Transformer life is set by ageing of the paper insulating the windings, and the ageing rate depends on the hottest point of the winding (the hot spot). Raising load increases winding loss with current squared and raises temperature. But transformers are cooled by large volumes of oil that take hours to warm. The winding-to-oil gradient catches up in minutes, but oil temperature lags, so a short overload raises the hot spot only modestly. A long overload, by contrast, approaches steady temperature and ages the paper rapidly. Overload decisions are therefore based on temperature over time and the life it consumes, not steady temperature.

Ultimate temperatures

$$ \Delta\theta_{o,\infty} = \Delta\theta_{or}\left[\frac{1+RK^2}{1+R}\right]^x,\qquad \Delta\theta_{h,\infty} = H g_r K^y $$

K is load factor, Δθ_or rated top-oil rise (52 K), R ratio of load to no-load loss at rating (6), H g_r rated hot-spot-to-oil gradient (26 K), x = 0.8, y = 1.3. Steady values use the same settings as transformer hot spot (steady state).

Time response and ageing rate

$$ \tau_o\frac{d\Delta\theta_o}{dt} = \Delta\theta_{o,\infty} - \Delta\theta_o,\quad \tau_w\frac{d\Delta\theta_h}{dt} = \Delta\theta_{h,\infty} - \Delta\theta_h,\quad V = 2^{(\theta_h - 98)/6} $$

τ_o is the oil time constant (180 min) and τ_w the winding time constant (7 min), both illustrative. V is paper ageing rate relative to 98 °C (non-thermally-upgraded paper). Ambient is 30 °C.

Coffee Break Trivia

Half per 8 degrees, double per 6

In 1930 Montsinger reported that transformer insulation life halves for roughly every 8 °C rise — an empirical rule applying the exponential temperature dependence of reaction rates to paper ageing. Later work revised the figure, and current international standards use a doubling of ageing rate per 6 K for non-thermally-upgraded paper. The idea that a few degrees change life dramatically has held for nearly a century.

Worked examples

Example 1: from 0.7 to 1.5 load by overload duration (30 °C ambient)

Overload duration	Peak hot-spot temperature	Life used (at 98 °C, incl. 3 h after)
30 min	115.8℃	2.8 h
1 h	124.0℃	10.5 h
2 h	136.2℃	61.5 h
Sustained (steady state)	167.2℃	143 min to reach 140 °C

At steady 0.7 load the hot spot is 79.2 °C with ageing rate 0.114, using about 2.7 hours of life per day. A 1-hour overload (10.5 hours) equals about 4 normal days.

🙋

Extending overload from 30 minutes to 2 hours raises life used from 2.8 to 61.5 hours — more than 20 times — though duration only quadrupled.

🎓

Because ageing rate is exponential in temperature. The longer the overload, the warmer the oil and the hotter the hot spot, and at that temperature ageing doubles and redoubles. At 30 minutes the peak is 115.8 °C (about 8 times the reference rate); at 2 hours it reaches 136.2 °C (about 80 times). So overload limits are more rationally set by what temperature the hot spot reaches and how much life is used than by a fixed number of hours. In practice, forecast temperature and load predictions drive a time-stepped calculation of next-day hot-spot temperature to decide allowable overload. Note that the hot spot here is an estimate at one point, and where the winding is hottest depends on oil flow. For new designs or cooling changes, detailed analysis coupling oil flow with winding heat, or fibre-optic measurements, confirm hot-spot location and gradient.

Modelling workflow

Get rated top-oil rise and winding gradient from heat-run tests.
Fit oil and winding time constants to temperature histories from tests.
Feed load and ambient time series to solve hot-spot temperature over time.
Integrate ageing rate to obtain life consumed.
For new designs, couple oil flow and winding heat to locate the hot spot.

Coffee Break Trivia

“A transformer hotter at night under the same load”

At one substation, fibre-optic hot-spot measurements sometimes read higher at night than in daytime at the same load. Investigation showed load peaking in the early evening and oil temperature continuing to climb for hours afterwards. Night load matched daytime load, but the oil hadn't cooled yet. Adopting time-stepped calculation in operations and starting cooling fans earlier ahead of the evening peak lowered night-time hot-spot temperatures.

Common mistakes

Mistakes and fixes

Mistake	Effect	Fix
Judging overload by steady state	Short overloads over-restricted	Solve in time
Oil time constant only	Peak underestimated on load step	Include winding time constant
Ageing from average temperature	Life overestimated	Integrate ageing rate
Ignoring the after-period	Delayed heat missed	Include hours afterwards
Constant ambient	Summer hot spot wrong	Use ambient time series

🙋

I'd like to learn related topics.

🎓

Rate this article

Thank you for your feedback!

Helpful

More details

Report error