Transformer Partial Discharge Testing — IEC 60270 Methods, Acceptance Criteria, and Online vs. Offline PD Monitoring

Introduction

Partial discharge (PD) is the silent killer of transformer insulation. It does not cause immediate failure — instead, it erodes insulation incrementally, creating carbonized tracks, enlarging voids, and eventually bridging the insulation gap when you least expect it. A transformer that passes every other factory test can still carry PD defects that will shorten its service life from 30 years to 5.

PD testing — both in the factory and on-site — is the most effective non-destructive technique for detecting incipient insulation defects. This article explains the physics of partial discharge, the IEC 60270 measurement standard, acceptance criteria by transformer type, the critical distinction between online and offline testing, and answers to the questions engineers actually ask.

What Is Partial Discharge — The Physics

Partial discharge is a localized electrical discharge that partially bridges the insulation between conductors. Unlike a full breakdown, PD does not create a complete conducting path — the discharge occurs across a small void, gas bubble, or surface imperfection, while the rest of the insulation remains intact.

The Three Classic PD Mechanisms

1. Internal Discharge (Voids and Cavities) Gas-filled cavities within solid or liquid insulation have lower permittivity and lower dielectric strength than the surrounding material. The electric field concentrates across the void, causing it to break down at a voltage lower than the bulk insulation withstand. Each discharge deposits charge on the cavity walls and produces chemical byproducts that degrade the surrounding material.

*Common in:* cast-resin transformers with microscopic air inclusions, oil-paper insulation with gas bubbles from thermal decomposition, poorly vacuum-impregnated windings.

2. Surface Discharge (Tracking) Occurs along the interface between insulation and a different dielectric (typically air). Contamination, moisture, and sharp potential gradients accelerate surface tracking.

*Common in:* transformer bushings with pollution deposits, creepage paths on terminal boards, dirty insulation surfaces.

3. Corona Discharge Ionization of the surrounding medium (air or oil) around a sharp metallic point at high potential. While corona itself may not immediately threaten insulation, the ozone, nitrogen oxides, and UV radiation it generates degrade organic materials over time.

*Common in:* poorly radiused HV connections, burrs on winding conductors, sharp edges on core clamping hardware.

The PD Pulse

Each PD event is extremely fast — risetimes in the nanosecond range, durations of tens to hundreds of nanoseconds. The charge transferred is tiny: picocoulombs (10⁻¹² C) to nanocoulombs (10⁻⁹ C). But the repetition rate can be thousands of pulses per second, and the cumulative damage is real.

The energy released per pulse is:

W_pulse ≈ ½ × q × V_inception

Where q is the apparent charge and V_inception is the inception voltage. Hundreds of thousands of such pulses per day adds up.

IEC 60270 — The Measurement Standard

IEC 60270:2000 (with Amendment 1:2015) is the foundational standard for PD measurement on electrical apparatus. It defines:

Measurement Circuit

The standard measurement circuit consists of:

• Coupling capacitor (Ck): Provides a low-impedance path for the high-frequency PD current pulse while blocking the power-frequency voltage
• Measuring impedance (Zm): Converts the PD current pulse into a voltage signal for the detector
• PD detector: A bandpass instrument that filters, amplifies, and quantifies the PD pulses

The measurement can be made at either the coupling capacitor terminal or the test object terminal, depending on accessibility and background noise.

Apparent Charge — pC vs. pc

The fundamental quantity measured is apparent charge (q) , expressed in picocoulombs (pC) . It is called "apparent" because it is the charge that would need to be injected at the measurement terminals to produce the same reading — not the actual charge transferred at the discharge site, which is almost always larger but inaccessible to direct measurement.

Critical note on units:

• Strictly, "pc" is an abbreviation for "parsec" — use pC (picocoulombs)
• 1 pC = 10⁻¹² C = the charge of approximately 6.24 × 10⁶ electrons
• For context: a PD level of 10 pC is an almost imperceptibly small electrical event; a level of 10,000 pC (10 nC) is a significant discharge that may be audible

Frequency Bands

IEC 60270 specifies two measurement frequency ranges:

• Wideband: 30–500 kHz (most common for factory testing)
• Narrowband: 9–30 kHz or 100–500 kHz with bandwidth of 9–30 kHz (better for noisy environments)

The choice depends on the background noise spectrum at the test site. Wideband captures more of the PD pulse energy and gives truer charge readings; narrowband provides better signal-to-noise ratio in electrically noisy environments.

Calibration

Before every PD test, the measurement system must be calibrated by injecting a known charge pulse (typically 5–50 pC) across the test object terminals using a calibrator and measuring the detector response. This establishes the scale factor (pC / mV) for the measurement.

This step is mandatory. Test reports without calibration data are worthless.

Acceptance Criteria — What PD Level Is Acceptable?

IEC 60076-3 Limits

Transformer Type / Voltage	Maximum Permitted PD Level	Notes
Liquid-immersed, Um ≤ 72.5 kV	250 pC	General criterion
Liquid-immersed, Um ≤ 72.5 kV	50 pC	If specified in contract
Liquid-immersed, Um > 72.5 kV	100 pC	Unless otherwise agreed
Dry-type (cast-resin), per IEC 60076-11	10 pC	Very strict — reflects higher PD sensitivity of cast resin

Why 10 pC for Cast-Resin Transformers?

Cast-resin insulation is solid, not self-healing like oil. Once a PD-induced carbon track forms inside epoxy, it is permanent and grows with each subsequent discharge. Oil, by contrast, can absorb some PD byproducts and redistribute them through circulation. This fundamental difference means cast-resin transformers require much tighter PD limits — a unit measuring 50 pC that would pass for an oil transformer is a definite reject for a cast-resin unit.

Background Noise Limit

IEC 60270 requires that background noise during the test be less than 50% of the specified PD limit. If the limit is 10 pC, noise must be below 5 pC — this is challenging in many factory environments and requires careful shielding, filtering, and grounding.

Common Industry Practice (Beyond IEC)

Many utilities and industrial users impose tighter in-house criteria:

• PD ≤ 50 pC for all oil-immersed distribution transformers (regardless of Um)
• PD ≤ 100 pC at 1.5 × Um/√3 for power transformers (pre-commissioning test)
• PD ≤ 250 pC as an alarm threshold for in-service monitoring
• PD ≤ 500 pC as a trip/investigate threshold for online monitors

Online vs. Offline PD Testing

Offline (Factory / Site Commissioning)

Offline PD testing is conducted with the transformer disconnected from the network, energized from an external test source. Advantages:

• Controlled environment — you can suppress background noise
• True PD inception and extinction voltage (PDIV / PDEV) can be determined
• Conforms to IEC 60270 for acceptance testing
• Can be combined with induced voltage withstand test for efficiency

Disadvantages:

• Requires specialized HV test equipment (resonant test set, variable-frequency source)
• Transformer must be out of service for 4–8 hours
• Test voltage may not replicate in-service stress conditions (harmonics, transients)

Online (In-Service Monitoring)

Online PD monitoring uses permanently installed sensors (capacitive couplers at bushings, UHF antennas inside the tank, acoustic sensors on the tank wall, or HFCTs on neutral/ground connections) to detect PD while the transformer is energized and loaded.

Advantages:

• Continuous monitoring — catches intermittent PD that offline testing might miss
• No outage required
• Trending over weeks/months reveals deterioration rate
• Can localize PD source using acoustic triangulation or UHF time-of-flight

Disadvantages:

• Extremely high background noise from corona on bushings, switching operations, and neighboring equipment
• Sensor location and type critically affect sensitivity
• Interpretation requires experienced analysts — false alarms are common
• Not a substitute for acceptance testing; used for condition-based maintenance

When to Use Which

Scenario	Recommended Approach
Factory acceptance	Offline IEC 60270 — mandatory
Site commissioning	Offline IEC 60270 — recommended benchmark
Critical power transformers (>50 MVA)	Online monitoring — permanent installation
Aging fleet / suspect units	Offline survey first, online if budget permits
GIS-connected transformers	UHF sensors in GIS — highly effective

Frequently Asked Questions

FAQ

Q: My transformer measured 180 pC during factory PD testing. The spec limit is 100 pC. Is it a reject?

Under IEC 60076-3, yes — if the contract specifies ≤100 pC. Before rejecting, verify three things: (1) that the measured PD was from the transformer and not external corona or lab noise, (2) that the calibration was correct, and (3) that the test was conducted at the specified voltage for the specified duration. External noise can masquerade as transformer PD — ask the factory to repeat the test with enhanced shielding. If the PD is confirmed internal, the manufacturer should locate the source (acoustic or electrical PD location) and rework or replace the defective component.

Q: Can I perform PD testing on a transformer that is already filled with oil?

Yes — in fact, offline PD testing is routinely performed on oil-filled transformers both in the factory and at site. The oil is an integral part of the insulation system. Ensure the oil has been properly processed (degassed, dehydrated, filtered) before testing — gas bubbles in oil are PD sources and will produce false-positive readings. Allow at least 24–48 hours of settling time after oil processing before PD testing.

Q: What is the difference between PD Inception Voltage (PDIV) and PD Extinction Voltage (PDEV)?

PDIV is the voltage at which PD pulses first appear (typically defined as exceeding the background noise level by a given margin) as the test voltage is raised. PDEV is the voltage at which PD pulses disappear as the test voltage is lowered. PDEV is always lower than PDIV due to residual charge in the void. A large gap between PDIV and PDEV (hysteresis) suggests a physically large void. For acceptance purposes, PDIV should be comfortably above the maximum operating voltage — ideally ≥ 1.2 × Um/√3.

Q: Is PD testing required for all transformers or just MV/HV units?

Under IEC 60076-1, PD measurement is a routine test only for transformers with Um ≥ 72.5 kV. For transformers with Um < 72.5 kV, PD testing is a special test — meaning it is only performed if specified in the purchase order. However, for cast-resin dry-type transformers per IEC 60076-11, PD testing is a routine test regardless of voltage class, with a ≤10 pC limit. Many informed buyers specify PD testing for all MV transformers (Um ≥ 12 kV) as a special test because the incremental cost is negligible compared to a field failure.

Q: Can online PD monitoring replace periodic offline testing?

No. Online monitoring is a complementary technology, not a replacement. Offline testing per IEC 60270 in a controlled environment gives the most accurate absolute PD magnitude, provides PDIV/PDEV, and serves as a calibrated baseline. Online monitoring adds the dimension of time — it tells you whether PD is getting worse, when it occurs (under what load/temperature conditions), and can trigger alarms. Best practice: establish an offline baseline at commissioning, install online monitoring on critical units, and repeat offline testing every 5–10 years depending on criticality.

Q: Why do some standards reference pC and others mV or dB?

Different PD measurement techniques produce different units. IEC 60270 measures apparent charge in pC using a coupling capacitor and measuring impedance — this is the reference method for transformer acceptance testing. UHF techniques (common in GIS and online monitoring) measure signal amplitude in dBm or mV because the sensor is essentially an antenna; these cannot be directly calibrated in pC. Acoustic PD detection measures sound pressure in mV or dB, useful for localization but not for quantifying discharge severity. Always check which measurement technique was used before comparing PD test results.

References & Standards

• IEC 60270:2000 + AMD1:2015 — High-voltage test techniques — Partial discharge measurements
• IEC 60076-3:2018 — Power transformers — Insulation levels, dielectric tests (PD limits)
• IEC 60076-11:2018 — Dry-type transformers (PD routine test, 10 pC limit)
• IEC 62478:2016 — Acoustic and electromagnetic methods for PD measurement
• IEEE C57.113-2010 — Recommended Practice for Partial Discharge Measurement in Liquid-Filled Power Transformers
• CIGRE TB 676 — Partial Discharges in Transformers
• CIGRE TB 366 — Guidelines for Unconventional Partial Discharge Measurements