Polarization index (PI) and DAR — insulation assessment of large motors and transformers

The insulation resistance testing guide covers the standard procedure: measuring a single insulation resistance value after a fixed measurement time. On large motors, generators and transformers (typically from several MVA upward), however, that single value is considerably less discriminating than on a small circuit: the winding capacitance is then so large that the resistance reading keeps rising for a considerable time. This article covers two supplementary assessments tailored to that: the polarization index (PI) and the dielectric absorption ratio (DAR), both based on IEEE 43.

Why a single reading is not enough

When a DC voltage is applied to a winding, the measured current initially builds up from three components: a brief charging current from the winding capacitance, a gradually declining absorption current (polarization of the insulating material), and an eventually stable leakage current. On a large winding with significant capacitance, the absorption current still dominates for minutes, so the resistance reading at, say, 1 minute is still rising sharply. A single reading at a fixed point in time then mainly indicates how far that build-up has progressed, not primarily the actual condition of the insulation.

Dielectric absorption ratio (DAR)

DAR is the faster of the two assessments: the ratio between the measured insulation resistance after 60 seconds and after 30 seconds:

$$DAR = \frac{R_{60s}}{R_{30s}}$$

DAR is a quick screening tool, suitable as an initial indication, but less discriminating on windings with substantial capacitance, where the absorption process runs over a longer timescale than the 30-60-second window covered by DAR.

Polarization index (PI)

PI uses a longer window and is therefore the more complete assessment for machines with significant winding capacitance: the ratio between the insulation resistance after 10 minutes and after 1 minute:

$$PI = \frac{R_{10min}}{R_{1min}}$$

Guideline values under IEEE 43

IEEE 43 gives the following practical guideline values for a minimum acceptable PI:

Insulation class	Minimum PI
Class A (older machines)	1.5
Class B, F, H (most present-day machines)	2.0

A PI clearly below these guideline values indicates moisture absorption, contamination or ageing of the insulation; a PI at or just below 1.0 (barely rising resistance) is a strong indicator of severely moist or contaminated insulation.

Note: IEEE 43 is specifically written for rotating machines (motors, generators). For liquid-cooled power transformers, a different, transformer-specific practice applies for diagnostic field testing, with its own reference framework. In practice, PI and DAR are nonetheless also applied to transformer windings as a supplementary, comparable indication, but without directly adopting the exact IEEE 43 threshold values.

Practical relevance

For the periodic insulation assessment of large motors, generators and (with the caveat above) transformers, a single insulation resistance measurement provides insufficient information to distinguish a trend from a snapshot. PI and/or DAR, measured and documented at every periodic test, make it possible to recognise a gradual insulation degradation before the absolute insulation resistance value itself has become too low.

Common mistakes

Assessing only the absolute insulation resistance value on a large machine, without calculating PI or DAR — an apparently "adequate" resistance value can conceal a declining trend that only becomes visible via PI/DAR.
Using DAR as a substitute for PI on machines with large winding capacitance — DAR's shorter time window cannot fully capture the absorption process of a large winding.
Applying IEEE 43 thresholds for rotating machines directly to power transformers as an exact standard, rather than as a supplementary, indicative comparison.
Not keeping the test voltage or test duration consistent between successive periodic measurements — PI and DAR are only meaningfully comparable over time if the measurement conditions (test voltage, temperature, test duration) remain comparable.