Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
IEC 60216-3: Thermal Endurance Statistics — Temperature Index Is Not Measured, It Is Extrapolated
Three Temperatures, Dozens of Specimens, One Arrhenius Line: How a Temperature Index Is Born
IEC 60216-3:2006 specifies statistical calculation methods for insulation thermal endurance (TI and RTI). The Temperature Index is not a direct measurement — it is extrapolated from accelerated ageing data at three temperatures via Arrhenius regression to the temperature corresponding to 20,000-hour life.
| Statistical Parameter | Requirement | Meaning |
|---|---|---|
| Correlation r² | ≥0.9 | Consistent ageing mechanism across test temperatures; r²<0.9 signals mechanism change — extrapolation unreliable |
| 95% Lower Confidence Bound | TI defined by this | NOT the mean. Uses the lower bound for statistical safety — “the most pessimistic reasonable estimate” |
| F-test | Regression significant | Validates the linear ln(life) vs. 1/T relationship is statistically significant |
Why three temperatures, not two? The Arrhenius equation ln(t)=A+B/T has two unknowns (A, B); mathematically, two temperatures suffice. But the third temperature tests whether the linearity assumption holds — if the third point deviates significantly, the ageing mechanism has changed at that temperature (e.g., oxidation dominant at high temperature, hydrolysis at low temperature), and simple extrapolation is invalid.
TI calculation example:
T1=220°C(493K) mean failure t1=480h
T2=200°C(473K) mean failure t2=1,800h
T3=180°C(453K) mean failure t3=7,200h
→ Regression: ln(t) = -15.2 + 9,500/T (r²=0.98)
→ 20,000h: T = 9,500/(ln(20,000)+15.2) = 9,500/25.1 = 378K = 105°C
→ TI = 105°C (95% lower bound may be 98°C)TN Lab — Temperature Index is not “measured” — it is statistically extrapolated. r²<0.9 means the test is flawed, not the material.
