Transformer Insulation Resistance Test: Polarization Index (PI), Dielectric Absorption Ratio (DAR), Test Voltage Selection & Temperature Correction to 20°C
Abstract
The insulation resistance (IR) test, performed with a DC megohmmeter (commonly a "Megger"), is the most widely applied diagnostic test on power transformers. It is simple, non-destructive, and provides a wealth of information about the condition of the winding insulation system — moisture content, contamination (conductive particles, carbon tracking), and surface leakage. The test yields three primary parameters: the 1-minute insulation resistance value (R_60), the polarization index (PI = R_10min / R_1min), and the dielectric absorption ratio (DAR = R_60s / R_30s or R_60s / R_15s). This article explains what each parameter physically represents, how to correct the measured values to a standard temperature of 20°C for comparison with historical data and acceptance criteria, test voltage selection per the transformer voltage class, and the pass/fail criteria for both new (factory/delivery) and in-service transformers per IEEE 43 and IEC 60076-1.
1. Physical Basis of Insulation Resistance
1.1 The Three Current Components
When a DC test voltage is applied across the insulation (winding-to-winding or winding-to-ground), the total measured current has three components:
1. Capacitive Charging Current (I_c): The initial surge of current that charges the geometric capacitance of the winding-to-ground and winding-to-winding. This current decays to zero within a few seconds (time constant τ = R × C, where R is the test-lead and winding resistance, and C is the winding capacitance — typically 1-50 nF per phase).
2. Absorption Current (I_a, also called polarization current): The current required to orient polar molecules (water, insulating oil, cellulose-OH groups, and ionic contaminants) in the electric field. This current decays with time as I_a ∝ t^(-n), where n ≈ 0.5-1.0. The absorption current is the physical basis for the PI and DAR ratios — insulation with higher moisture or contamination has higher absorption current that decays more slowly.
3. Leakage (Conduction) Current (I_L): The steady-state DC current flowing through the bulk insulation and across the surface (at bushings, terminal boards). This current is constant over time. It is determined by the resistivity of the insulation materials: cellulose paper ρ ≈ 10¹²-10¹⁴ Ω·m (dry), decreasing to 10⁹-10¹¹ Ω·m when wet; oil ρ ≈ 10¹²-10¹³ Ω·m (clean, dry).
The insulation resistance at time t is:
R(t) = V_DC / (I_c(t) + I_a(t) + I_L)
After sufficient time (typically 10 minutes), I_c and I_a have decayed to near zero, and R approaches the true insulation resistance: R_∞ ≈ V_DC / I_L.
1.2 What the Test Values Mean
| Parameter | Measurement | Physical Interpretation |
|---|---|---|
| R_60 (1-minute IR) | Insulation resistance after 60 seconds of applied DC voltage | Dominated by absorption current + leakage current. Sensitive to moisture and contamination. |
| R_10min (10-minute IR) | Insulation resistance after 10 minutes (600 seconds) | Approximates the true (leakage-only) resistance. Absorption current has mostly decayed. |
| PI (Polarization Index) | R_10min / R_60s | Ratio of absorption-dominated to near-steady-state resistance. Indicates moisture, contamination, and insulation porosity. Independent of test voltage and temperature (to first order). |
| DAR (Dielectric Absorption Ratio) | R_60s / R_30s (or R_60s / R_15s) | Fast version of PI — useful when 10-minute test is impractical. Less discriminating. |
2. Polarization Index (PI) and Dielectric Absorption Ratio (DAR)
2.1 PI Interpretation (per IEEE 43 / IEC 60076-1)
| PI Value | Insulation Condition | Action |
|---|---|---|
| >2.0 | Good — insulation is dry and clean | Accept for service |
| 1.5-2.0 | Questionable — marginal moisture or contamination | Investigate — may be acceptable with other positive test results |
| 1.0-1.5 | Poor — significant moisture or contamination | Dry out or recondition before placing in service |
| <1.0 | Dangerous — insulation is wet or severely contaminated | Do not energize — requires drying and cleaning |
Important caveat: PI is temperature-dependent, though less so than the absolute IR value. At very low temperatures (<10 °C), moisture in the paper is frozen and does not contribute to the absorption current — a wet transformer at 0 °C can show a normal PI, falsely indicating good condition. The PI test should be performed when the transformer insulation temperature is ≥10 °C, ideally at the factory test temperature (20-30 °C).
PI for new, dry insulation (factory test): Typically PI > 3-5 for modern transformers with dry paper (<0.5% moisture) and clean, degasified oil. A PI of 1.8-2.0 for a new transformer at the factory test should be investigated — it may indicate incomplete drying during the vapor-phase drying process or moisture ingress during assembly.
2.2 DAR Interpretation
DAR is used when a 10-minute test is not practical (multiple windings to test, limited outage time):
DAR = R_60s / R_30s (IEEE standard) DAR = R_60s / R_15s (alternative)
| DAR Value | Interpretation |
|---|---|
| >1.4 | Acceptable (dry) |
| 1.25-1.4 | Marginal |
| <1.25 | Unacceptable (wet/contaminated) |
DAR is less sensitive than PI because the absorption current has not decayed as much at 60 seconds — if time permits, always perform the full PI test.
3. Test Voltage Selection
3.1 Voltage Selection per Winding Rated Voltage
The DC test voltage is selected based on the rated AC voltage of the winding being tested, per IEEE 43 and IEC 60076-1:
| Winding Rated Voltage (kV RMS, line-to-line) | DC Test Voltage (V) | Megger Range |
|---|---|---|
| <1 kV | 500 | 500 V setting |
| 1-5 kV | 500-1,000 | 1 kV setting |
| 5-12 kV | 1,000-2,500 | 2.5 kV setting |
| 12-36 kV | 2,500-5,000 | 5 kV setting |
| >36 kV | 5,000 | 5 kV setting |
For power transformers (≥11 kV): 5,000 V DC is the standard test voltage. Most modern insulation testers (Megger MIT series, S1 series) have a 5 kV range and are suitable for all power transformer testing.
Caution: Do not apply a test voltage higher than the winding's rated insulation level. A 5 kV DC test voltage is equivalent to approximately 3.5 kV RMS AC in terms of dielectric stress — well below the insulation level of any winding rated ≥11 kV, and safe for all power transformer windings.
3.2 Test Connections
For each winding, three insulation resistance tests are typically performed:
| Test | Connection | Measures |
|---|---|---|
| HV to (LV + Ground) | Megger (+) to HV terminal, (-) to LV terminals and ground | HV winding insulation to ground (through the LV winding and tank) |
| LV to (HV + Ground) | Megger (+) to LV terminal, (-) to HV terminals and ground | LV winding insulation to ground |
| HV to LV | Megger (+) to HV terminal, (-) to LV terminal (both windings floating from ground) | Inter-winding insulation (HV-to-LV) |
The Guard Terminal: The megger's guard terminal (G) is connected to surfaces where surface leakage current might bypass the insulation being measured. For a transformer bushing, the guard is connected to the bushing test tap or a conductive band wrapped around the porcelain between the HV terminal and the grounded flange. Without the guard, surface leakage across a contaminated bushing surface will lower the measured IR, giving a false indication of internal insulation degradation.
4. Temperature Correction to 20 °C
4.1 Why Temperature Correction is Essential
Insulation resistance decreases exponentially with increasing temperature because:
- The ionic conductivity of the cellulose and oil increases (more mobile ions at higher temperature)
- Moisture solubility increases (more water moves from the paper to the oil, increasing the oil's conductivity)
- The absorption current decays faster (shorter relaxation time for polar molecules)
IR at 80 °C is typically 5-20× lower than IR at 20 °C. Without temperature correction, a perfectly healthy transformer tested on a hot summer day (40 °C winding temperature) will show an IR value far below the pass criterion established at 20 °C, causing unnecessary alarm.
4.2 Correction Formula (IEEE 43 / IEC 60076-1)
The standard correction formula:
R_corr(20°C) = R_measured(t°C) × K_T
Where K_T is the temperature correction factor:
For oil-immersed transformers (IEEE 43): K_T = e^(0.0695 × (T - 20))
Where T is the winding temperature in °C at the time of the test.
Alternative (simplified, IEC 60076-1):
| Temperature (°C) | Correction Factor K_T |
|---|---|
| 10 | 0.50 |
| 20 | 1.0 |
| 30 | 2.0 |
| 40 | 4.0 |
| 50 | 7.5 |
| 60 | 15.0 |
| 70 | 25.0 |
| 80 | 40.0 |
Rule of thumb: IR doubles for every 10 °C decrease in temperature; halves for every 10 °C increase.
4.3 Practical Procedure
- Measure the winding temperature: The most accurate method is to measure the winding DC resistance and calculate the temperature from the resistance ratio (hot resistance / cold resistance at known reference temperature). Alternatively, use the top-oil thermometer, adding 5-10 °C for the winding-to-oil gradient.
- Perform the IR test at 5 kV DC, recording R_60s, R_10min, and the winding temperature.
- Calculate PI = R_10min / R_60s (no temperature correction needed for PI).
- Apply the temperature correction to R_60s: R_60s_corr = R_60s × K_T.
- Compare R_60s_corr to historical data (trending) or to the minimum acceptable value.
4.4 Minimum Acceptable IR (Corrected to 20°C)
Per IEEE 43, the minimum insulation resistance for AC rotating machines (applied by extension to transformers):
R_min (MΩ) = (V_rated + 1) (for kV-class rated voltage)
For a 132 kV transformer: R_min = 133 MΩ. However, this formula was developed for generators. For modern clean, dry transformer insulation, typical R_60 values at 20 °C:
- New transformer (factory): >1,000-10,000 MΩ
- In-service (dry): >500 MΩ
- Action threshold (investigate): <100 MΩ
- Do-not-energize threshold: <50 MΩ (unless the transformer is hot and the temperature correction is applied — 50 MΩ at 20 °C could be 5 MΩ at 60 °C)
FAQ
Q: Why do we use DC voltage for insulation resistance testing instead of AC?
A DC test measures the resistive (leakage) component of the insulation impedance, which is the quantity related to moisture, contamination, and aging. An AC test would measure the combined impedance of the resistance and the capacitive reactance (impedance Z), and the capacitive component (dominating at 50/60 Hz) would mask the resistive component — a wet, degraded winding could appear indistinguishable from a dry, healthy one because both have similar capacitance. The DC test, especially the time-dependent decay (PI), separates the resistive and absorption components, providing diagnostic information that an AC test cannot.
Q: Can I perform the insulation resistance test on a transformer that is still hot from operation?
Yes — and it is actually preferable to test the transformer near its operating temperature because the insulation condition at operating temperature is what matters. However, you must correctly record the winding temperature and apply the temperature correction to compare with historical values taken at other temperatures. The PI test is less temperature-sensitive and can be interpreted without correction if the temperature is above 10 °C. Note: if the transformer is cooling during the 10-minute test, the absorption current will be affected by the changing temperature — for best accuracy, perform the test at a stable temperature or record the temperature at the beginning and end and use the average.
Q: What does a PI value less than 1.0 mean?
PI < 1.0 means the insulation resistance is decreasing over time — R_10min is LESS than R_60s. This is a severe finding that indicates: (1) the absorption current is not decaying normally (which suggests the insulation is not polarizing — dry paper polarizes; wet paper is already saturated with water and does not further polarize), or (2) the leakage current is increasing during the test, indicating moisture migration within the insulation under the influence of the DC field (electroendosmosis — water molecules are drawn toward the cathode). A PI < 1.0 is a definitive indication that the transformer requires drying before it can be safely energized.
Q: Is DAR (R_60s/R_30s) a reliable substitute for PI?
DAR is a time-saving alternative but less discriminating. The absorption current decays approximately as t^(-n), so the ratio between times that are close together (30 s to 60 s) is inherently small — a DAR of 1.3 corresponds to n ≈ 0.38, while a DAR of 1.1 corresponds to n ≈ 0.14. The natural variability of the absorption exponent n between different insulation types, temperatures, and contamination levels makes DAR less universally applicable than PI. If you have time for only one test, R_60s corrected to 20 °C with a clear pass/fail threshold is more useful than DAR. Use PI whenever possible.
Q: What test voltage should I use for a 33 kV transformer winding?
5,000 V DC is the standard and recommended test voltage for all power transformer windings rated ≥11 kV. A 33 kV winding's insulation system is designed for a BIL of 170-200 kV peak — 5 kV DC is far below any stress that could cause insulation damage. Do not use 500 V or 1,000 V for windings rated ≥11 kV because the lower voltage may not adequately polarize the insulation, giving a falsely high PI (low voltage = low absorption current magnitude, but the relative decay of a small signal is harder to measure accurately). For windings rated 1-5 kV, use 1,000-2,500 V.
Q: How do I interpret a sudden drop in insulation resistance from one annual test to the next?
A sudden drop (>50% decrease in R_60 corrected to 20 °C) between annual tests — with PI unchanged — suggests: (1) moisture ingress (check the silica gel breather, conservator bladder, and gaskets for leaks), (2) oil contamination (check the oil BDV and moisture — if the oil moisture has increased while the paper moisture is stable, the IR will drop because oil contributes to the leakage path), or (3) surface contamination on bushings (clean the bushings and re-test with the guard terminal connected — if the IR recovers, the problem was surface leakage, not internal insulation degradation). If both R_60 and PI decrease significantly, the internal insulation has taken on moisture — drying is required.
References / Standards
| Reference | Title |
|---|---|
| IEEE 43-2013 | IEEE Recommended Practice for Testing Insulation Resistance of Electric Machinery |
| IEC 60076-1:2011 | Power transformers — Part 1: General |
| IEEE C57.12.90-2021 | IEEE Standard Test Code for Liquid-Immersed Distribution, Power, and Regulating Transformers |
| IEEE C57.152-2013 | IEEE Guide for Diagnostic Field Testing of Fluid-Filled Power Transformers, Regulators, and Reactors |
| IEC 60422:2013 | Mineral insulating oils in electrical equipment — Supervision and maintenance guidance |
*Authored by Du Fu, Production Engineer at ZY POWER. The insulation resistance test is deceptively simple — its diagnostic value depends entirely on correct temperature correction, proper guard terminal usage, and consistent PI trending over time. A single IR value without temperature correction is meaningless.*
Download This Guide as PDF
Save this technical guide for offline reference. Includes all tables, specifications, and contact information.
Related Articles
2000kVA Dry-Type Transformer Ventilation Design: Loss Budgeting, Air Inlet Sizing, Fan CFM, and Temperature Control Logic
A 2000 kVA dry-type transformer dissipates approximately 24 kW of heat at full load — 3.6 kW in the core (no-load loss, a constant iron loss independent of load) and 20.0 kW in the windings (load loss, proportional to the square of the load
35kV Substation Transformer Selection Guide: S22 vs. SCB14, GB 50053 Compliance, and Capacitor Bank Coordination
The 35 kV voltage level is uniquely Chinese. While IEC standards recognize 36 kV as a standard Um, the specific 35 kV nominal voltage with 35/10.5 kV or 35/6.3 kV transformation ratios is a defining feature of China's medium-voltage distrib
50Hz/60Hz Transformer Compatibility Guide: Flux, Saturation Risk, Derating Rules, and Dual-Frequency Design
A transformer designed for 50 Hz and a transformer designed for 60 Hz are physically different machines. The difference is not in the nameplate — both might say "2000 kVA, 11/0.4 kV, Dyn11" — but in the iron core: the cross-sectional area,