How to Interpret Partial Discharge PRPD Patterns: A Field Engineer’s Guide

You’re looking at a PRPD plot from a transformer test. There are clusters of dots forming distinct shapes against the 50 Hz voltage waveform. One cluster sits at the voltage peaks. Another spreads across the whole cycle. A third forms a thin band with a distinctive “rabbit ear” shape.

Each cluster tells a different story. One is corona from a sharp point on a high-voltage conductor. Another is an internal void inside the insulation. The third is a floating metal component, un-bonded, picking up charge from the field around it.

Reading these patterns is a skill. It’s also the single most useful capability a commissioning engineer or diagnostic specialist can develop for medium- and high-voltage equipment. This article is the practical guide — what the patterns look like, what they mean, and how to use that information to plan intervention.

What a PRPD Plot Actually Shows

Before interpreting patterns, you need to understand exactly what the plot represents.

A Phase-Resolved Partial Discharge (PRPD) plot is a three-dimensional scatter plot compressed into two dimensions:

• X-axis: Phase angle of the AC test voltage (0° to 360°)
• Y-axis: Apparent charge magnitude (in picocoulombs, pC)
• Color intensity (or dot density): Number of PD pulses occurring at each phase/magnitude combination over the measurement period

Overlaid on the plot is the AC voltage waveform (a sinusoid from 0° to 360°). Each PD pulse appears as a dot positioned at the phase angle where it occurred and at the charge magnitude it produced.

Over a typical measurement time of several minutes, thousands to millions of PD pulses accumulate into characteristic patterns. The shape, position, and density of these patterns reveal the underlying defect type.

Why phase matters

The phase angle tells you when during the voltage cycle each discharge occurs. Different defect types produce discharges at characteristic phase angles because the physical conditions that trigger them (field strength, polarity, space charge state) vary across the AC cycle.

For example:

• Void discharges tend to cluster around the voltage zero-crossings on the rising and falling voltage slopes
• Corona clusters at the voltage peaks (where field strength is maximum)
• Surface tracking often shows asymmetric patterns (different on positive vs negative half-cycles)

Phase position is the single most important feature for defect classification.

Why magnitude matters less than you’d think

Many engineers’ first instinct is to focus on the maximum apparent charge — the “biggest” discharges. But pattern shape and phase distribution are far more diagnostic than magnitude alone.

A 500 pC void discharge is fundamentally the same phenomenon as a 50 pC void discharge — it’s telling you there’s a void of a particular size, and the size is roughly proportional to the charge. What matters for intervention planning is that it is a void discharge, not how big it is.

Magnitude matters when:

• Comparing to historical baselines (is the PD getting worse?)
• Comparing to equipment-specific limits (is the PD above acceptance thresholds?)
• Understanding severity (a 10,000 pC pulse is serious, period)

Magnitude doesn’t tell you what the defect is. Phase position does.

The Four Fundamental PD Types

Partial discharges in power equipment fall into four categories, each with characteristic physics and characteristic PRPD patterns. The Vahidi & Teymouri textbook classifies these as:

1. Corona discharge — discharges in gas (air, oil) around sharp points on conductors
2. Surface discharge — discharges along the interface between two insulation materials (e.g., solid-gas, solid-solid)
3. Internal (cavity) discharge — discharges inside voids within solid insulation
4. Electric treeing — a late-stage failure mode where discharges form branching conductive tracks through solid insulation

Additionally, a fifth pattern type often encountered in practice:

5. Floating component discharge — discharges from an ungrounded metal part that’s picking up charge from the electric field

Each produces a distinct PRPD signature. Once you learn the four main signatures, you can identify most problems you’ll encounter in the field.

Pattern 1: Internal Void Discharge

The most common and most important PD pattern in aged medium- and high-voltage equipment.

What’s happening physically

A gas-filled void exists inside the solid insulation — typically from manufacturing, thermal cycling, or mechanical stress. Per Vahidi & Teymouri, these cavities “are generally formed in solid or liquid insulating materials and they are generally filled with gas or air. When the gas inside the cavity is overstressed, such discharges are taking place.”

The electric field concentrates in the void because the gas has lower permittivity than the surrounding solid material. When the field exceeds the breakdown strength of the gas, a discharge occurs. The discharge quenches quickly (microseconds), the field recovers, and the process repeats on each voltage cycle.

Characteristic PRPD signature

Phase position: Clusters appear on the rising slopes of the voltage waveform — roughly 0°–90° on the positive half-cycle and 180°–270° on the negative half-cycle.

Pattern shape: Two roughly symmetric clusters, one on each polarity. Often described as a “rabbit ear” or “turtle back” pattern.

Symmetry: Usually symmetric between positive and negative half-cycles. Strong asymmetry suggests the void is close to one electrode (affecting the local field geometry differently on each polarity).

Magnitude distribution: Wide — from near-threshold levels up to a maximum corresponding to the void size. Typical range: 10 pC to several hundred pC in healthy MV equipment; several thousand pC in aged or defective equipment.

What it tells you

An internal void pattern means there’s physical damage or defect inside the solid insulation. This is the most dangerous PD type because:

• The discharge occurs inside the insulation volume, invisible from outside
• Each pulse chemically attacks the void boundary, progressively enlarging it
• Left unchecked, the void eventually extends into a failure path

Severity indicators

Severity of void PD correlates with:

• Maximum apparent charge — larger = bigger void
• Repetition rate — more frequent = more active degradation
• Voltage dependence — if PDIV (inception voltage) is well above operating voltage, less urgent; if PDIV is at or below operating voltage, urgent

Where you’ll see it

• Aged motor stator insulation (mica tape delaminating from conductor)
• Power transformer windings (void formation from thermal cycling)
• Cross-linked polyethylene (XLPE) cable insulation (manufacturing voids, water trees evolving to electric trees)
• Epoxy-molded switchgear bushings (voids at casting boundaries)

Pattern 2: Surface Discharge

What’s happening physically

A discharge propagates along the boundary between two different insulation materials — most commonly along the surface of a solid insulator in air or oil. Per Vahidi & Teymouri, surface discharge “takes place on the outer surface of a solid or liquid insulation… on the joint surface of two insulation of two different materials. The electrical strength on the joint surface is less than the electrical strength inside each insulation.”

Contamination, moisture, or physical damage on the surface creates the initial conducting path. Once started, surface discharges can propagate progressively, forming carbonized tracks that extend across the surface.

Characteristic PRPD signature

Phase position: Extends across wider phase ranges than void discharges — typically 0°–120° on positive half-cycle and 180°–300° on negative half-cycle. Often appears asymmetric between the two polarities.

Pattern shape: Broader clusters than void PD, often with a more elongated shape running along the phase axis. Can form “fan” or “wing” shapes extending from voltage peaks down toward zero-crossings.

Asymmetry: Typically asymmetric — surface discharge on the boundary between a conductor and insulator behaves differently in positive vs negative half-cycle due to space charge accumulation on the surface.

Magnitude distribution: Wide and variable. Can reach very high magnitudes (thousands of pC) when surface tracking is well developed.

What it tells you

Surface discharge indicates:

• Surface contamination on insulators (industrial pollution, salt, moisture)
• Physical damage to insulator surfaces (mechanical scratches, impact damage)
• Inadequate creepage distance for the applied voltage
• Carbonized tracks from previous PD activity (tends to be progressive)

Severity indicators

Surface PD is often considered more recoverable than void PD in early stages (cleaning the surface can eliminate the problem). However, once surface tracking has progressed to carbonized tracks, the damage is permanent and the insulator typically requires replacement.

Severity indicators:

• Magnitude — low magnitude surface PD from light contamination is generally acceptable
• Progression rate — rapidly worsening surface PD suggests active tracking, requires urgent attention
• Visible tracking — any visible carbonized paths on the insulator mean permanent damage

Where you’ll see it

• Outdoor insulators with pollution accumulation
• Bushings in contaminated environments
• End-turn insulation on large motors (if end-winding contamination is present)
• Cable termination boundaries (stress cone, termination body interfaces)

Pattern 3: Corona Discharge

What’s happening physically

Corona is a form of PD that occurs in gas around sharp points on high-voltage conductors. Per IEC 60270: “Corona is a form of partial discharge that occurs in gaseous media around conductors.”

The electric field concentrates at sharp points (edges, burrs, protrusions) until it exceeds the breakdown strength of the surrounding gas (usually air or SF6). Ionization occurs in the small region immediately around the sharp point, producing a corona discharge.

Corona is common and generally not an immediate threat, but it’s important to distinguish from more dangerous PD types.

Characteristic PRPD signature

Phase position: Highly concentrated at the voltage peaks — roughly 60°–120° on the positive half-cycle and 240°–300° on the negative half-cycle.

Pattern shape: Narrow, tight clusters near the voltage peaks. Often described as “compact” or “concentrated” compared to other PD types.

Asymmetry: Often strongly asymmetric. Negative corona (corona at the negatively-stressed electrode) typically starts at lower voltages and produces more pulses than positive corona. You’ll often see a dense cluster on one half-cycle and a sparser cluster on the other.

Magnitude distribution: Relatively narrow — corona pulses tend to have similar magnitudes. Typical range: 1 pC to 100 pC depending on the sharpness of the source and the gas pressure.

What it tells you

Corona indicates a sharp point, edge, or protrusion at high voltage. Sources include:

• Burrs on conductor ends from cutting or stripping
• Sharp edges on grading rings or corona rings
• Damaged or deformed shielding on HV bus work
• Floating wires or strands sticking out from cables
• Manufacturing defects at conductor endpoints

Severity indicators

Corona is generally not immediately destructive — it occurs in a small volume of gas, not inside solid insulation. However, in certain environments:

• Corona in oil can degrade oil quality over time (generating H2, CH4) — this is the form of PD often detectable by DGA
• Corona in SF6-insulated switchgear can produce toxic byproducts (SO2, HF) from decomposition
• Corona at transformer bushings can lead to audible buzzing, RF interference, and eventual flashover in outdoor equipment

Severity indicators:

• Location — corona inside a critical region (e.g., inside a GIS compartment) is more serious than corona on an exposed bushing in air
• Rate of progression — increasing corona suggests surface degradation or contamination worsening
• Associated gas byproducts (in oil or SF6 systems) — these indicate corona-driven chemistry

Where you’ll see it

• HV air-insulated bus work (most common — not usually a problem)
• Outdoor transformer bushings (corona from dirt accumulation or damage)
• Cable terminations (corona at conductor endpoints if semiconducting layers are damaged)
• Generator connections (corona from damaged shielding)
• SF6 GIS (corona from contamination or damaged components)

Pattern 4: Floating Component Discharge

What’s happening physically

A metal component that should be grounded (or at defined potential) has lost its connection and is “floating.” It picks up charge from the electric field around it through capacitive coupling. When the accumulated charge reaches breakdown threshold, a discharge occurs between the floating component and an adjacent surface.

Common sources:

• Broken grounding bond on a metallic shield
• Cracked weld on a metallic part inside the transformer tank
• Loose screws or fasteners that have lost electrical contact
• Insulation damage that has exposed a metal part

Characteristic PRPD signature

Phase position: Narrow, symmetric clusters. The phase angle depends on the capacitive relationship between the floating component and the surrounding conductors, but typical position is centered around each voltage zero crossing (90° before and after the peaks).

Pattern shape: Very narrow phase distribution — the pattern looks like a vertical bar or line at two specific phase angles on each half-cycle.

Symmetry: Highly symmetric — nearly identical patterns on positive and negative half-cycles.

Magnitude distribution: Often narrow — the component charges up, discharges at a specific voltage threshold, and repeats. This produces highly repeatable magnitudes.

What it tells you

Floating component PD means there’s a physical problem that needs to be found and fixed — not a gradual degradation issue. A component that should be grounded has lost its connection. The defect is usually:

• Mechanical (broken bond, cracked weld, loose fastener)
• Intermittent (intermittent connection due to vibration or thermal cycling)
• Progressive (worsening until the floating component eventually causes a major fault)

Severity

Floating component PD is mid-severity. It’s not as urgent as active void discharge (the component isn’t destroying insulation around it), but it’s worse than simple corona. It often progresses to bigger problems:

• The repeated discharges can damage surrounding insulation
• The floating component can eventually cause an arc-over
• In transformers, a floating metal part can cause oil contamination

The defect needs to be found and corrected during the next planned outage at latest.

Where you’ll see it

• Power transformers (loose shields, broken bonds on tank components)
• GIS compartments (floating cast aluminum components from damaged welds)
• Motor stators (broken bonds on end-winding support rings)
• Cable accessories (damaged grounding on metallic screens)

Mixed Patterns and How to Unpick Them

Real equipment rarely shows a single clean pattern. Most PD test reports on aged equipment show mixed patterns — void + corona + surface + noise all overlapping.

Step-by-step analysis approach

Step 1: Identify the noise floor. Before interpreting patterns, identify what’s noise. Random scattered dots with no pattern correlation are usually electromagnetic noise — typically present at low magnitude (<20 pC on a modern PD system), randomly distributed across all phases.

Step 2: Look for distinct clusters. Separate the clusters by eye. Are there clusters at the voltage peaks? At the rising slopes? At the zero-crossings? Each cluster is potentially a different PD source.

Step 3: Classify each cluster. For each identified cluster, ask:

• Where is it in phase? (Peak → corona; rising slope → void; around zero-crossing → floating)
• What’s the shape? (Narrow → corona/floating; broad → void/surface)
• Is it symmetric? (Symmetric → void/floating; asymmetric → surface/corona)

Step 4: Cross-check with voltage dependency. Raise the test voltage slightly and watch each cluster. Does it:

• Grow in magnitude and rate? → Typical PD response
• Grow in rate only, similar magnitudes? → Corona (the source point is fixed, more pulses per cycle)
• Suddenly appear at a threshold voltage? → Floating component (discharge at specific charge threshold)

Step 5: Consider equipment knowledge. What’s the equipment? Where are likely defect sites? The expected PD patterns on a transformer differ from those on a cable termination.

The “three-cluster” rule

In my field experience, most faulty MV/HV equipment shows one dominant PD pattern plus secondary activity. If you identify one dominant pattern that accounts for 70%+ of the activity, that’s typically your primary defect. Secondary patterns may be:

• Normal background PD (low-level void activity in any aged insulation)
• Corona from known sources (grading rings, bus work)
• Measurement noise or pickup from nearby equipment

Focus intervention on the dominant pattern first. Secondary patterns may or may not warrant attention depending on their trend.

Common Misinterpretations

Calling every peak-concentrated pattern “corona.” Corona is one source of peak-concentrated PD, but floating component PD can also appear near peaks depending on geometry. Check symmetry and cluster width before classifying.

Assuming symmetric pattern = void. Void PD is typically symmetric, but floating component PD is also symmetric. The distinguishing feature is cluster width — void PD spreads over 20°–40° of phase, while floating PD is often in bursts of <5° phase width.

Treating magnitude as primary indicator. A 5000 pC corona event is far less concerning than a 500 pC void discharge in solid insulation. Pattern classification always comes before magnitude interpretation.

Over-interpreting limited data. PRPD patterns stabilize after several minutes of measurement. Patterns observed in the first 30 seconds may be misleading. Always let the measurement run until patterns are stable.

Ignoring voltage history. Patterns at rated voltage differ from patterns at elevated test voltage. Acceptance testing at 1.5× rated voltage may show PD patterns that don’t occur at operating voltage — which may or may not be acceptable depending on the equipment standard.

The Role of Noise in PRPD Plots

PD measurement is always performed against a background of electrical noise. Understanding noise is essential to avoid misinterpreting it as PD.

Types of noise

Per Vahidi & Teymouri, the origins of noise in PD measurement include:

• Interference from power grids — harmonics, switching transients
• High-voltage source interference — thyristor switching, corona in test equipment
• Terminal and contact noise — loose connections, bad crimps
• Electromagnetic environmental noise — radio transmissions, nearby equipment
• Grounding loops — circulating currents from multiple ground paths

How noise appears on PRPD plots

• Random noise: Uniformly distributed dots with no phase correlation — looks like random “snow” across the plot
• Periodic noise: Tight clusters at specific phase angles unrelated to PD physics (e.g., at 80° on positive half-cycle due to a local thyristor switching)
• Narrowband noise: Linear clusters at specific magnitudes — often from radio frequency pickup

Noise discrimination

Modern PD analyzers include noise discrimination features:

• Phase-window filtering — reject pulses outside expected PD phase windows
• Magnitude thresholds — reject pulses below a noise floor
• Statistical pattern recognition — reject clusters that don’t match PD pattern shapes
• Time-domain gating — reject pulses outside expected PD timescales (<1 μs)

For offline testing per IEC 60270, performing measurements in a shielded lab minimizes noise. For online monitoring, noise discrimination is essential — and interpretation requires experience.

Using PRPD Patterns to Plan Intervention

The end goal of interpreting PRPD patterns is deciding what to do. Here’s a decision framework:

Urgent action (days to weeks)

• Void PD with magnitudes >5,000 pC on MV equipment
• Rapidly increasing PD (doubling in less than 6 months)
• PD inception voltage at or below operating voltage
• Any PD with visible correlating symptoms (abnormal temperature, unusual sounds, gas accumulation)

Action: De-energize, isolate, perform detailed diagnostic testing, plan repair or replacement.

Scheduled intervention (next planned outage)

• Moderate void PD (500–5,000 pC) with stable or slowly-increasing trend
• Floating component PD at any level
• Surface PD with visible contamination (clean at outage)
• Progressive patterns showing clear degradation trend

Action: Document, trend, plan intervention during next scheduled outage. Continue monitoring.

Continued monitoring

• Low-level void PD (<500 pC) with stable trend
• Corona from known sources in non-critical locations
• Mild surface PD in outdoor equipment where cleaning is periodic

Action: Document baseline, repeat testing annually, watch for trend changes.

No action needed

• Corona from grading rings on outdoor bus work (normal)
• Light background PD within typical levels for the equipment type and age
• PRPD noise without actual PD clusters

Action: Document as baseline, no intervention.

The trending imperative

Single PD measurements are far less useful than trended measurements. Whether a 200 pC void PD is concerning depends entirely on whether it’s been stable at 200 pC for three years or has increased from 50 pC to 200 pC in six months.

Establish baselines at commissioning and trend at least annually for critical equipment. Interpret patterns and magnitudes in the context of the trend, not as isolated snapshots.

Equipment-Specific Pattern Guides

Different equipment types have characteristic common patterns.

Power transformers

Most common patterns:

• Internal void (stator winding void, bushing paper voids)
• Corona (from HV leads, grading rings)
• Oil discharges (low-intensity discharges in oil, appear as noisy void-like patterns)
• Floating component (from damaged shields, cracked welds inside the tank)

Per Vahidi & Teymouri, common PD-inducing locations inside transformers:

• Conductors inside insulation
• Metallic sharp edges
• Too-small distances between neighboring conductors
• Metal parts not connected at potential or grounded
• Oil gaps, air bubbles in oil or insulating parts
• Too much gas or water content in oil
• Too much water content in insulating components

Specific indicators:

• Fault diagnosis should be correlated with DGA results — H2 and CH4 generation correlates with corona/void PD activity
• PD magnitudes should be trended against historical baselines for that specific transformer
• Acceptance test PD per IEC 60076-3 typically limits to 300 pC at 1.3 × rated voltage for power transformers

Rotating machines (motors/generators)

Most common patterns:

• Stator slot discharges (void PD between coil and slot wall)
• Internal void discharges (in mica-epoxy insulation)
• End-turn surface discharges (from contamination on end-windings)
• Delamination discharges (between conductor and ground-wall insulation)

Specific indicators per IEEE 1434:

• Phase-resolved PRPD analysis distinguishes slot discharge from internal void discharge
• Stator bar coupler measurements differ from off-line measurements
• Machine-specific interpretation criteria apply (larger machines tolerate more PD due to insulation volume)

Cable systems

Most common patterns:

• Internal void (manufacturing voids, water trees evolved to electric trees)
• Surface discharge (at termination boundaries, if semiconducting layer is damaged)
• Floating component (on metallic screen if grounding bond is broken)
• Corona (at conductor ends if semiconducting layer has receded)

Specific indicators:

• Online PD monitoring increasingly used for MV and HV cables
• Termination PD is most common — inspect terminations first
• Cable joint PD patterns differ from termination PD (different geometry)

SF6-insulated switchgear (GIS)

Most common patterns:

• Corona (from particles or protrusions on conductor surfaces)
• Floating component (from damaged cast components, broken bonds)
• Surface tracking (on spacer insulators if contaminated)

Specific indicators:

• UHF PD detection (300–3000 MHz) is standard for GIS
• SF6 decomposition byproducts (SO2, HF) correlate with active PD
• Particle-induced corona is common and often not immediately serious (particles can migrate to low-field zones)

FAQ

Key Takeaways

• PRPD patterns encode the type of PD source — phase position and pattern shape are more diagnostic than magnitude.
• Four fundamental patterns: Internal void (rising slope clusters, symmetric), surface discharge (broad asymmetric clusters), corona (narrow clusters at peaks), floating component (narrow clusters near zero-crossings).
• Phase position is the most diagnostic feature. Peaks → corona; rising slopes → void; around zero-crossings → floating; broad and asymmetric → surface.
• Void PD is the most dangerous because it occurs inside solid insulation and progressively destroys it.
• Corona is often acceptable — it occurs in gas, not solid insulation, and doesn’t typically lead to immediate failure.
• Surface PD is often remediable by cleaning, unless tracking has progressed to carbonization.
• Floating component PD indicates a physical defect (broken bond, cracked weld) that needs repair at next outage.
• Real equipment shows mixed patterns. Identify the dominant pattern and intervene on that; treat secondary patterns according to their trend.
• Trending is essential — single measurements are less useful than trended data. Establish baselines and watch for changes.
• Intervention priorities: Urgent for void PD >5,000 pC or rapidly increasing patterns; scheduled for moderate patterns; monitor for low-level stable patterns.
• Equipment context matters — expected patterns differ between transformers, motors, cables, and GIS.

Standards and References Used in This Article

Reference	Key Content
IEC 60270:2000+AMD1:2015	PD measurement quantities (apparent charge in pC), measurement circuits, calibration requirements
IEC 60076-3	Power transformer PD acceptance test requirements
IEC 62478	UHF and acoustic PD measurement (for GIS and online monitoring)
IEEE 1434	IEEE Guide for the Measurement of Partial Discharges in AC Electric Machinery
Vahidi & Teymouri (2019)	Quality Confirmation Tests for Power Transformer Insulation Systems (Springer) — Chapter 5 on PD types and measurement
CIGRE Technical Brochure 227	Life Management Techniques for Power Transformers