1. Introduction
Insulation resistance test is among the most important routine tests for power transformers and shunt reactors. They measure:
- insulation resistance between each winding and ground (tank),
- insulation resistance between different windings, and
- where accessible, between tank, core frame and core.
These values characterize the condition of the main insulation system and reveal problems such as moisture ingress, surface contamination, insulation ageing and incorrect core grounding.
The same measurements—taken as a function of time—are also used to calculate the Polarization Index (PI) and Absorption Index (AI), which provide a deeper view of insulation quality.
2. Test Principles, Preparation and Depolarization
In the insulation resistance test a DC voltage is applied and the resulting leakage current is measured with an insulation resistance meter (“megger”). Resistance is read at different times after energization (typically at 15 s, 60 s and, when required, up to 10 min). The test must be performed with great care, because polarization and surface leakage strongly influence the result.
2.1 General preparation
Before any measurements:
- Isolate the transformer from the system; all external connections must be disconnected.
- Bond tank, core and frame together and ground them to form a common reference during winding-to-ground tests.
- Clean and dry the bushings; contamination and moisture on bushing surfaces cause misleadingly low readings.
- Perform the test when ambient humidity is reasonably low and record the humidity for later interpretation.
- Never perform insulation resistance tests while the transformer is under vacuum.
For each winding under test, all its terminals at the bushing are short-circuited together. Windings that are not under test are normally connected to the tank/ground.
2.2 Depolarization between tests
Dielectric absorption and residual charges in the insulation can influence subsequent measurements. To obtain stable and reproducible results, a depolarization step is used:
- After finishing a set of measurements, connect all windings together and bond them to the tank (ground).
- Maintain this connection for at least twice the duration of the preceding voltage application or the time taken for the resistance to reach its steady value (whichever is shorter).
- This allows the insulation to discharge and depolarize adequately before the next test.
Using the guard terminal of the insulation resistance meter, according to the connection tables in Section 3, further reduces the influence of surface leakage and decreases the number of depolarization cycles needed.
3. Typical Test Connections and Sequences
Insulation resistance test is performed in a defined order. Below is practical connection schemes.
3.1 Two-winding power transformer (HV / LV)
For a three-phase, two-winding transformer the main tests are:
| Test No. | Measured insulation | “+” lead | “–” lead | Guard lead | Notes |
| 1 | HV – Ground | HV terminals (all phases shorted) | Tank / ground | LV terminals | |
| 2 | LV – Ground | LV terminals | Tank / ground | HV terminals | |
| 3 | Depolarization | – | – | – | |
| 4 | HV – LV | HV terminals | LV terminals | Tank / ground |
Note: Connect HV, LV and tank together for depolarization.
Figure 1: Test configuration for insulation resistance measurement of the HV–GND, LV–GND, and HV–LV
3.2 Three-winding power transformer (HV / LV / TV)
For a three-winding transformer, typical tests and guard connections are:
| Test No. | Measured insulation | “+” lead | “–” lead | Guard lead |
| 1 | HV – Ground | HV | Tank / ground | LV + TV |
| 2 | LV – Ground | LV | Tank / ground | HV + TV |
| 3 | TV – Ground | TV | Tank / ground | HV + LV |
| 4 | Depolarization | – | – | – |
| 5 | HV – LV | HV | LV | TV + tank |
| 6 | TV – LV | LV | TV | HV + tank |
These connections ensure that when one winding is measured to ground, the other windings are used as a guard to bypass surface leakage currents.
3.3 Three-winding autotransformer (HV / LV common winding + TV)
For an autotransformer, the common HV/LV winding is treated as a single body for some tests:
| Test No. | Measured insulation | Megger “+” lead | Megger “–” lead | Guard lead |
| 1 | (HV + LV) – Ground | (HV + LV) | Tank / ground | TV |
| 2 | TV – Ground | TV | Tank / ground | (HV + LV) |
| 3 | (Depolarization) | – | – | – |
| 4 | (HV + LV) – TV | (HV + LV) | TV | Tank / ground |
The exact sequence can follow utility practice, but the principle is to measure all relevant insulation paths with suitable guard connections and depolarization intervals between groups of tests.
3.4 Core and frame insulation resistance
Where the core and frame connections are accessible, a separate core insulation test can be performed (distinct from the main winding tests):
- Temporarily disconnect the core and frame from earth.
- Measure insulation resistance between:
- core – frame,
- core – earth,
- frame – earth,
using 500 V DC for 1 minute. After the test, restore the permanent core grounding.
4. Test Voltage, Timing and Temperature Correction
4.1 Test voltage
- The preferred DC test voltage is the value recommended by the transformer manufacturer.
- If this information is not available, a 5 kV DC test voltage is commonly used for main insulation resistance tests.
- In all cases, the test voltage must not exceed the rated RMS voltage of the winding under test.
- For core-insulation tests, 500 V DC applied for 1 minute is typical.
4.2 Measurement times
During each test:
- Apply the DC voltage and read the insulation resistance at 15 seconds and 60 seconds.
- Where more detailed assessment is required, continue the test for up to 10 minutes, recording the resistance every minute.
- The DC voltage should remain approximately constant; if leakage current rises continuously without stabilizing, the test must be stopped.
These multiple readings are used to compute PI and AI (Section 5.3).
4.3 Temperature and correction to 20 °C
Insulation resistance is strongly temperature-dependent. For meaningful comparison:
- Preferably, tests should be carried out close to 20 °C so that little correction is required, or at least repeated at similar temperatures from one maintenance interval to the next.
- The measured resistance (R) is converted to its value at 20 °C, R_20, using a temperature correction factor (a) obtained from a table or curve based on oil temperature:
where:
– insulation resistance referred to 20 °C,
- R – measured insulation resistance,
– temperature correction factor (typically derived from oil temperature; interpolation is used for intermediate temperatures).
All limits and comparisons in the next section refer to 10-minute values corrected to 20 °C.
5. Interpretation of Results
5.1 Main insulation (windings to ground / between windings)
Desired insulation resistance values at 20 °C depend on voltage class and previous test results:
- The 1-minute corrected value should not be about 20% lower than the previous test result for the same transformer, unless the value is very high (above about 20 GΩ, small variations are less significant).
- For transformers with rated voltage ≥ 145 kV, insulation resistance at 20 °C is expected to be at least 1 GΩ.
- For transformers with rated voltage ≤ 72.5 kV, insulation resistance at 20 °C should not be less than 500 MΩ.
If these conditions are not satisfied, further tests (such as dielectric dissipation factor / power factor, oil quality and moisture analysis) are recommended.
It is particularly useful to trend results over time. A gradual decrease in resistance or a change in the shape of the resistance-versus-time curve may indicate:
- moisture ingress,
- ageing or cracking of solid insulation,
- contamination on internal surfaces, or
- incomplete oil drying after maintenance.
5.2 When to extend tests to 10 minutes
Measuring resistance up to 10 minutes provides additional diagnostic information and is recommended:
- during commissioning tests,
- during detailed periodic tests or fault investigations, where the extra time (roughly one-hour additional test time for the whole transformer) is acceptable,
- whenever previous tests showed that the resistance reaches its steady value later than about 6 minutes.
5.3 Polarization Index (PI) and Absorption Index (AI)
When the 10-minute value is available, the Polarization Index is defined as:
If the test is only continued up to 1 minute, the Absorption Index is used instead:
These indices are meaningful mainly when the absolute resistance is below about 20 GΩ; for very high resistance values they tend to saturate and are less useful.
The Typical values for AI and PI are shown in tables.
| PI value | Condition |
| < 1.0 | Dangerous |
| 1.0 – 1.1 | Poor |
| 1.1 – 1.25 | Questionable |
| 1.25 – 2.0 | Fair |
| > 2.0 | Good |
| AI value | Condition |
| 1.0 – 1.25 | Questionable |
| 1.25 – 1.6 | Good |
| > 1.6 | Excellent |
PI and AI must always be interpreted together with the absolute insulation resistance and with knowledge of the insulation liquid. For example, new transformers filled with very clean mineral oil may show low PI simply because the leakage current is extremely small and dominated by capacitive effects.
5.4 Core insulation resistance
For the core and core-frame insulation test (500 V DC, 1-minute values corrected to 20 °C) a typical interpretation is:
| Transformer / reactor condition | Core insulation resistance at 20 °C | Assessment |
| New unit | > 500 MΩ | Normal |
| In-service unit | > 100 MΩ | Normal |
| In-service unit | 10–100 MΩ | Insulation ageing – monitor trend |
| In-service unit | < 10 MΩ | Requires detailed investigation |
Abnormally low core insulation resistance may indicate damaged core support insulation, incorrect core grounding arrangements, or contamination/moisture at the core and frame.
6. Conclusions
Insulation resistance testing, carried out systematically between windings and ground, between windings themselves, and—where possible—between core, frame and tank, is a powerful but relatively simple diagnostic tool for power transformers. By following defined connection schemes with appropriate guard leads, observing depolarization intervals, using suitable test voltages and correcting all readings to 20 °C, engineers can obtain reliable data that reflects the true condition of the insulation system.
