Interpreting PD Patterns to IEC 60270: A Practical Pattern Recognition Guide

Interpreting PD Patterns to IEC 60270: A Practical Pattern Recognition Guide

1. What IEC 60270 covers — and what it does not

IEC 60270 is the high-voltage test technique standard for partial discharge measurement. It defines the measurement of apparent charge, expressed in picocoulombs, including the calibration procedure, the measurement circuit, and the requirements on the measuring instrument.

What it deliberately does not do is tell you what a discharge is. Two defects with the same apparent charge can have completely different causes and consequences. The magnitude tells you how much; the pattern tells you what kind.

This is why a PD report containing only a pC figure is incomplete: it answers the measurement question and leaves the diagnosis open.

2. What PRPD data is

Phase-resolved partial discharge data plots each detected discharge as a point, with:

  • Phase on the horizontal axis (the AC cycle, conventionally 0–360°)
  • Charge on the vertical axis
  • Density shown by colour or intensity, since discharges cluster

The result is a picture of where in the voltage cycle the discharges occur. That distribution — not the peak value — is what identifies the defect type.

3. The four pattern families

Family Pattern features Typical location Field implication
Internal void Two clusters, broadly symmetric around the zero crossings, charge rising with test voltage; stable shape Cable body insulation, joint insulation Insulation defect; trending matters more than the absolute value
Surface discharge Asymmetric clusters, often clustered near the zero crossings or spread along the cycle; sensitive to moisture Terminations, contaminated or poorly prepared surfaces, contaminated joints Often reparable by re-terminating and cleaning
Corona Strongly asymmetric, concentrated in one half-cycle, stable; frequently independent of the cable Sharp points, hardware, nearby energised equipment, test leads Usually external; check whether it is a genuine test-circuit artefact
Floating electrode / treeing Irregular and unstable, shifting shape between measurements, sometimes erratic magnitude Joints, terminations, degraded interfaces Requires investigation; instability itself is the finding

The value of this classification is practical: it points to where to look and what kind of action follows. A symmetric internal pattern in the cable body and an asymmetric surface pattern at a termination are different jobs with different costs.

4. How the pattern responds to voltage

The pattern's behaviour as voltage changes is as informative as its shape.

  • Inception voltage (PDIV): the voltage at which discharges first appear. A low PDIV relative to operating voltage means the defect is active under service stress.
  • Extinction voltage (PDEV): the voltage at which discharges stop on the way down. A large hysteresis between inception and extinction is characteristic of certain defect types and indicates the discharge is being sustained by trapped charge.
  • Pattern stability: a pattern whose shape is unchanged across the voltage steps suggests a defined geometric defect; a pattern that grows and shifts suggests instability or multiple sites.

Record the pattern at more than one voltage. A single pattern at a single voltage discards most of the available diagnostic information.

5. Separating PD patterns from interference

Not everything that produces a discharge-looking pattern is PD in the asset.

  • Corona in air produces a pattern that is highly asymmetric and does not change when the cable circuit is reconfigured. Verify by isolating the suspected source.
  • Switching noise from the test source tends to be phase-locked to the source but does not scale with test voltage in the same way as PD.
  • Communication and drive noise is usually broadband and non-repetitive in phase; it appears as scattered points rather than clusters.
  • Surface contamination on the test leads can produce a genuine surface-discharge pattern that is in the circuit rather than the asset.

The systematic test: change one thing at a time — voltage, source isolation, circuit layout — and observe which pattern component moves with it.

6. A practical identification workflow

Step Action Output
1 Record background noise with the circuit assembled but not energised Noise floor and its pattern
2 Apply voltage in steps and record PRPD at each step Pattern set with voltage dependence
3 Identify the phase distribution and symmetry of each cluster Pattern family classification
4 Determine PDIV and PDEV Voltage margin to operating stress
5 Test the pattern's dependence on circuit changes Confirmation that the source is in the asset
6 Localise (PD mapping) and correlate with joints Position of the discharge
7 Repeat the measurement at the same point Stability and spread

Steps 1 to 5 classify the defect; step 6 turns the classification into an actionable location.

7. What pattern recognition cannot tell you

Being explicit about the limits prevents over-claiming in reports.

  • It does not give remaining life. Patterns indicate type and activity, not a date of failure.
  • It is indicative, not definitive — a pattern is strong evidence of a defect class, not proof of a specific physical mechanism.
  • Multiple simultaneous defects can superimpose into an ambiguous pattern.
  • Pattern libraries are instrument- and method-dependent; a pattern recorded with one bandwidth is not directly comparable with another.

The correct use is to combine pattern classification with magnitude, voltage dependence, location and trend. Any one of those alone is an incomplete basis for a maintenance decision.

FAQ

Does IEC 60270 require pattern analysis? No. IEC 60270 covers the measurement of apparent charge. Pattern analysis is a diagnostic layer applied on top of a compliant measurement, and it is what converts a magnitude into an identified defect type.

Can a pattern identify the exact defect location? No. Pattern analysis classifies the defect type. Determining the position requires PD mapping — time-of-flight or reflectometry techniques — correlated with joint positions from as-built records.

Why is the same defect type reported differently by two instruments? Apparent charge depends on the measurement circuit and bandwidth, and pattern appearance depends on the instrument's detection and processing. Comparisons are valid only within the same instrument family, method and configuration.

Is a symmetric pattern always worse than an asymmetric one? No. Symmetry is a classification feature, not a severity scale. Severity depends on magnitude, voltage margin, location, and whether the activity is changing over time.

How should pattern data be stored? Store the raw phase-resolved data with the test conditions — voltage step, calibration factor, circuit configuration, instrument settings. A stored image without its conditions cannot be compared with a later measurement.

Related reading

Back to blog

Leave a comment