Online vs Offline Partial Discharge Testing: Which Method, When, and Why

You’ve decided your equipment needs partial discharge testing. The next question is operational: do you take the equipment out of service for offline measurement, or install sensors and monitor while it operates? The answer drives equipment selection, budget, outage planning, and the diagnostic value you’ll get.

This article compares offline and online PD testing in detail — how each works, what each can and cannot detect, what each costs, and when to choose which. It assumes you already understand why PD testing matters; if you need that background first, see our PD Testing vs Insulation Resistance article.

The Fundamental Difference

Both methods detect the same thing — partial discharges inside insulation — but they do it under fundamentally different conditions.

Offline PD testing: the equipment is de-energized, isolated from the power system, and connected to a dedicated test voltage source. PD measurements happen in controlled laboratory or field conditions. Test voltage can be varied, allowing measurement of PD inception and extinction voltages.

Online PD testing: the equipment continues operating in normal service. PD activity is detected through sensors that don’t interrupt power flow. The system measures whatever PD occurs at actual operating voltage and load.

Each approach answers a different question:

• Offline asks: “Under controlled test voltage, what is this equipment’s PD behavior?”
• Online asks: “Under actual operating conditions, is this equipment producing PD activity?”

Both questions are useful. Neither answers the other.

Offline PD Testing: How It Works

Offline testing follows the procedures defined in IEC 60270:2000+AMD1:2015. The standard specifies the measurement circuit, calibration method, and instrument requirements.

The standard test circuit

The equipment under test (EUT) is connected through an external high-voltage source. A coupling capacitor (Ck) connects to the EUT terminals, and a measuring impedance (Zm) sits in series with the coupling capacitor. PD pulses travel through this loop and produce measurable signals at the measuring impedance.

The basic flow:

1. EUT is fully isolated from the power grid
2. Calibrated test voltage source applied (typically variable, up to 1.5–2× rated)
3. Coupling capacitor and measuring impedance complete the measurement loop
4. Wide-band amplifier and digital acquisition capture PD pulses
5. Software processes pulses into apparent charge values, PRPD patterns, and statistical analysis

Frequency ranges per IEC 60270

IEC 60270 specifies measurement bandwidths that determine pulse resolution and sensitivity:

Instrument type	f₁ (lower)	f₂ (upper)	Bandwidth (Δf)
Wide-band	30 kHz – 100 kHz	up to 1 MHz	100 kHz – 900 kHz
Narrow-band	—	—	9 kHz – 30 kHz with center 50 kHz – 1 MHz

Wide-band instruments capture more pulse detail and have better pulse resolution (<10 μs between consecutive pulses). Narrow-band instruments are more selective against external noise but lose some pulse detail.

For a transformer or rotating machine winding, the upper measurement frequency is often reduced to a few hundred kHz because the equipment’s inductance attenuates higher frequencies before they reach the measurement terminals.

Calibration

This is the operational advantage of offline testing. A calibrator injects a known charge (typically 5 pC, 10 pC, 100 pC, 1000 pC) directly across the EUT terminals before the actual test. The measurement system records this known input, scales its readings to picocoulombs, and the ensuing measurements are quantitatively traceable to the calibration.

This calibration is repeatable, transferable between test sites, and comparable across different instruments. It’s why offline PD measurements appear as “150 pC at 1.5×Un” in test reports — that’s a calibrated, defensible quantitative value.

Test voltage variation

Because the test voltage source is dedicated to the test, you can:

• Slowly raise voltage to find PD inception voltage (PDIV)
• Lower voltage to find PD extinction voltage (PDEV)
• Hold at specific test points (1.0×Un, 1.3×Un, 1.5×Un, 1.7×Un) and characterize PD behavior
• Apply voltage above operating to stress-test marginal insulation

This level of control is impossible online — you only see PD behavior at whatever voltage the system happens to be operating at.

Where offline excels

• Acceptance and commissioning testing — calibrated, traceable, comparable to type test data
• Acceptance against equipment-specific standards (IEC 60076-3 for transformers, IEEE 1434 for rotating machines)
• PDIV/PDEV characterization — only practical with controlled voltage source
• Detailed defect characterization — controlled environment lets you isolate variables
• Forensic analysis after a known fault — quantitative comparison to historical baselines
• Research and development — manufacturers use offline PD heavily during product qualification

Where offline falls short

• Requires equipment outage — significant operational impact for in-service equipment
• Doesn’t show real operating behavior — equipment may behave differently under actual load and temperature
• Snapshot in time — a 30-minute test misses PD events that occur intermittently
• Expensive test source for HV equipment — a test voltage source for a 400 kV transformer is its own major piece of equipment

Online PD Testing: How It Works

Online PD testing detects PD activity while equipment operates normally. Multiple sensor technologies exist, each with different physical principles and applications.

High-Frequency Current Transformers (HFCTs)

The most common online PD sensor for transformers, motors, and cables. An HFCT clamps around grounding conductors (transformer neutral, cable shield, motor frame) and detects high-frequency current pulses produced by PD activity.

Physical principle: PD pulses cause brief current flows in the grounding path. The HFCT picks up the magnetic field of these pulses and converts to a measurable voltage signal.

Frequency range: typically 100 kHz to 30 MHz, depending on the specific HFCT design.

Strengths:

• Non-intrusive — clamp around existing grounding conductor
• Continuous monitoring possible
• Good for cable and transformer applications
• Reasonably priced ($500–$3,000 per HFCT)

Limitations:

• Sensitive to noise from other electrical equipment
• Quantitative calibration in pC is difficult (impedance of the path is unknown and variable)
• Wide-band noise from power electronics can obscure low-magnitude PD

UHF (Ultra-High Frequency) Sensors

Detect electromagnetic radiation from PD pulses in the 300 MHz – 3 GHz range. Common for gas-insulated switchgear (GIS) where the metal enclosure becomes a waveguide for UHF signals.

Physical principle: PD pulses radiate electromagnetic energy. Internal antennas inside GIS chambers, or external antennas near other equipment, capture this radiation.

Strengths:

• Excellent noise immunity (most external interference is below 100 MHz)
• Standard for GIS — covered by IEC 62478
• Allows location estimation through time-of-arrival differences
• Continuous monitoring with reasonable sensor cost

Limitations:

• UHF measurements don’t directly convert to apparent charge in pC
• Specific to equipment with appropriate UHF sensor mounting points
• Limited sensitivity for low-energy PD (under specific conditions)

Acoustic Emission Sensors

Detect ultrasonic pressure waves produced by PD activity. Common for transformers, where acoustic sensors mounted on the tank exterior pick up oil-borne pressure waves from internal PD.

Physical principle: PD events produce acoustic shockwaves. Piezoelectric sensors convert these pressure changes to electrical signals in the 20 kHz – 500 kHz range.

Strengths:

• Excellent for fault location (acoustic time-of-arrival from multiple sensors)
• Non-intrusive — sensors mount on tank exterior
• Specific to internal events — external electrical noise doesn’t produce acoustic signals
• Can localize defects to specific tank coordinates

Limitations:

• Depends on acoustic propagation paths (oil + tank wall + air)
• Sensitivity varies with sensor placement
• Cannot directly correlate to apparent charge values
• Cross-correlated with electrical methods for confirmation

Capacitive couplers

Permanently installed coupling capacitors at equipment terminals provide a continuous PD measurement path similar to offline testing. Less common but used in specialized applications (large stator windings, critical transformers).

Strengths:

• Direct measurement at the equipment terminals
• Closer to IEC 60270 calibrated measurement
• Permanent installation enables continuous trending

Limitations:

• Higher cost (specialized equipment installation)
• Requires equipment design to accommodate sensors
• Subject to noise pickup similar to HFCTs

Other methods

• Stator slot couplers (SSCs) — embedded in motor/generator stator slots for direct PD detection
• Bushing tap couplers — connect to transformer bushing capacitive taps for terminal measurement
• Optical methods — for PD producing visible discharge (corona, surface tracking)
• Chemical methods — DGA (dissolved gas analysis) for PD activity in oil-filled transformers (covered in the DGA-PD correlation article)

Where online excels

• No outage required — equipment continues earning revenue while monitored
• Real-world operating conditions — captures PD that only occurs under specific load, temperature, or harmonic conditions
• Continuous trending — catches developing faults early
• Long observation windows — multiple weeks of data show patterns that 30-minute offline tests miss
• Fleet monitoring — single SCADA can monitor dozens of assets
• Early-stage defect detection — finds PD that’s intermittent or condition-dependent

Where online falls short

• Higher background noise — other electrical equipment, motors, switching, harmonics
• No quantitative calibration in pC — values are typically relative (mV, dB, or vendor-specific scale)
• PDIV/PDEV cannot be measured — voltage is whatever the system delivers
• Pattern interpretation is harder — noise mixed with PD signals
• Specialized expertise required for analysis and interpretation
• Initial installation cost can be substantial for fleet-wide deployment

Sensitivity Comparison

This is one of the most important differences and most often misunderstood.

Offline sensitivity

Modern offline PD systems per IEC 60270 routinely achieve 1 pC sensitivity in shielded laboratory conditions. Field offline measurements typically achieve 5–10 pC sensitivity due to environmental noise. A defect producing 50 pC of apparent charge is clearly detectable; a 5 pC defect is at the noise floor.

For acceptance testing of new equipment, IEC 60076-3 typically specifies PD limits around 100–300 pC at 1.3×Un — well above the measurement floor.

Online sensitivity

Online PD systems typically have higher noise floors due to operational interference:

• HFCT in clean substation environment: 5–20 pC equivalent (after calibration against known PD source)
• HFCT in noisy industrial environment: 50–200 pC equivalent
• UHF in GIS: very low for in-band signals, but quantitative comparison difficult
• Acoustic on transformer tank: highly position-dependent, generally 100+ pC equivalent for confident detection

The “equivalent” qualifier matters. Online sensors detect PD-related signals, but converting those to apparent charge values requires assumptions about the propagation path, calibration source, and signal processing. The numbers are useful for trending but not directly comparable to IEC 60270 calibrated measurements.

What this means in practice

If your goal is detecting PD activity above 100 pC, both methods will work. Online monitoring catches it continuously; offline testing catches it during the test window.

If your goal is detecting PD activity below 50 pC, offline testing is significantly more reliable. Online detection at this level requires very clean signal environments and sophisticated noise discrimination.

If your goal is detecting PD activity that only happens under specific conditions (high load, high temperature, harmonic stress), online monitoring is the only practical method. Offline testing only sees what happens during the test.

Calibration: Where They Diverge Most

Calibration is the largest practical difference between online and offline.

Offline calibration

Per IEC 60270, calibration is performed by injecting a known charge across the EUT terminals before the test. The measurement system scales its readings to this known input. Subsequent PD measurements are quantitatively traceable to the calibration source.

Standard calibration values: 5 pC, 10 pC, 50 pC, 100 pC, 500 pC, 1000 pC, 10000 pC.

Calibration uncertainty per IEC 60270 is typically ≤10% for properly maintained equipment. This means a measurement reported as “200 pC” actually represents 180–220 pC with high confidence.

This calibration is what makes offline measurements:

• Comparable across different test instruments
• Comparable to type test data
• Suitable for acceptance against quantitative standards
• Defensible in disputes

Online calibration

Online measurements are typically not calibrated to apparent charge in the IEC 60270 sense. Instead:

• Sensitivity is verified through pulse injection on test conductors
• Trending is established against equipment-specific baselines
• Relative changes are tracked rather than absolute values

A vendor might report online PD as “150 mV equivalent” or “−40 dBm” or use proprietary scaling. These values are useful within a single system over time but don’t translate directly across systems or to offline calibrated measurements.

The implication for procurement

If you need quantitative compliance with standards (IEC 60076-3 for transformer acceptance, IEEE 1434 for rotating machines), you need offline calibrated PD measurements.

If you need long-term health monitoring with trend analysis, online measurements are appropriate even without absolute calibration.

If you need both, plan for both: offline at commissioning and major outages for calibrated baselines, online between outages for trending.

Outage and Operational Impact

For utility and industrial applications, this is often the deciding factor.

Offline test outage requirements

For a transformer:

• 1–2 hours preparation (isolation, grounding, test setup)
• 1–2 hours actual testing (calibration, voltage stepping, data collection)
• 1–2 hours teardown and return to service
• Total outage: 4–8 hours for a single test

For a generator or large motor:

• Half-day to full-day outage typical
• May coincide with planned maintenance windows
• Limited frequency due to operational impact

For a cable system:

• 2–4 hours per cable for offline testing
• Can be done during scheduled feeder switching

The outage cost varies by equipment criticality:

• Critical transformer: $50k–$500k revenue impact for a multi-hour outage
• Industrial motor: $5k–$50k production impact
• Generator: $50k–$5M revenue impact at peak demand
• Distribution feeder: $1k–$10k operational cost

These numbers explain why critical equipment often has online monitoring even when offline testing would be more accurate.

Online operational impact

Online PD monitoring has minimal operational impact once installed:

• Sensor installation typically requires brief outage (hours) for one-time install
• Continuous monitoring runs alongside normal operation
• Data analysis happens in background
• Alarms and exceptions trigger investigation, not automatic shutdown

For critical equipment with online monitoring already installed, the operational case is strong: continuous visibility with no ongoing outage cost.

Cost Breakdown

Offline PD equipment

For transformer or motor PD testing per IEC 60270:

• Wide-band PD measurement system: $30,000–$100,000
• Calibrators and accessories: $5,000–$15,000
• Coupling capacitors: $3,000–$15,000 per voltage class
• Test voltage source for HV equipment: $50,000–$500,000+ (for 400+ kV testing)
• Acoustic emission system (if added): $20,000–$80,000

Total for a basic offline transformer PD capability: roughly $80,000–$200,000 not including HV test sources.

Online PD equipment

For continuous monitoring of a single transformer or motor:

• HFCT sensors: $500–$3,000 each (typically 3–6 per unit)
• UHF sensors (for GIS): $2,000–$10,000 per chamber
• Acoustic emission sensors: $1,000–$5,000 per sensor
• Multi-channel acquisition system: $20,000–$100,000
• Software and analysis platform: $10,000–$50,000
• Installation labor: $5,000–$20,000 per asset

Total for online monitoring of a single critical transformer: roughly $40,000–$150,000.

Per-test cost (offline)

For utilities with their own offline PD equipment:

• Field testing crew (3–4 person team) + equipment day rate: $5,000–$15,000 per day
• Single test campaign for one transformer: 1–2 days
• Cost per offline test: $5,000–$30,000 (including outage cost)

For utilities using contractor PD testing service:

• Contractor flat rate: $10,000–$30,000 per transformer test
• Plus outage cost separately

Total cost of ownership (5 years)

For a critical transformer:

Offline-only approach:

• Annual offline PD test: $15,000–$25,000 per year
• 5-year cost: $75,000–$125,000
• Plus outage costs: $250,000+ over 5 years

Online-only approach:

• Initial installation: $80,000
• Annual maintenance and analysis: $10,000–$20,000 per year
• 5-year cost: $130,000–$180,000
• Outage costs: minimal

Hybrid approach:

• Initial online monitoring: $80,000
• Annual online costs: $10,000
• Offline test every 2–3 years: $20,000 per occurrence
• 5-year cost: $160,000–$200,000

For non-critical equipment, periodic offline testing is significantly cheaper than online monitoring. For critical equipment, online monitoring’s continuous coverage often justifies the higher cost.

Equipment-Specific Considerations

Different equipment types favor different approaches.

Power transformers

Offline testing:

• Standard for acceptance and commissioning
• IEC 60076-3 specifies PD test as part of dielectric testing
• Comparable to manufacturer’s type test data
• Typically every 5–10 years for service equipment

Online testing:

• Increasingly common for critical units (generator step-up, main bus transformers)
• HFCT on neutral grounding conductor is the standard sensor
• Combined with DGA and oil analysis for comprehensive monitoring
• Acoustic emission for fault location when needed

Recommendation: Offline at commissioning and major outages. Online monitoring for transformers above $500k value or critical to operations.

Rotating machines (motors and generators)

Offline testing:

• Required for acceptance per IEEE 1434
• Typically performed during major outages
• Stator winding and rotor PD characterization

Online testing:

• HFCT or stator slot couplers (SSCs) embedded during manufacturing
• Continuous monitoring with daily/weekly trending
• Particularly common for large turbine generators

Recommendation: Offline for new machines and after major refurbishment. Online for generators above 50 MW or motors driving critical processes.

Gas-insulated switchgear (GIS)

Offline testing:

• Difficult and expensive to perform on installed GIS
• Limited to factory acceptance testing
• Site testing usually limited to commissioning

Online testing:

• UHF monitoring is the industry standard per IEC 62478
• Permanent UHF sensors built into GIS chambers
• Continuous monitoring with sophisticated noise discrimination

Recommendation: UHF online monitoring is essentially mandatory for modern GIS. Offline rarely useful after installation.

Cables

Offline testing:

• Common for new cable commissioning
• VLF (very low frequency) PD testing is standard
• Performed before energization

Online testing:

• HFCT on cable screen/shield
• Continuous monitoring increasingly used for critical feeders
• Particularly valuable for long cable runs and submarine cables

Recommendation: Offline for commissioning. Online for cables longer than 10 km, submarine cables, or feeders to critical loads.

Switchgear and bushings

Offline testing:

• Less common — typically tested as part of associated equipment
• Bushings tested with associated transformer

Online testing:

• UHF monitoring for GIS (covered above)
• Bushing tap couplers for transformer bushing monitoring
• Continuous monitoring of busbar systems in critical substations

Recommendation: Online monitoring for critical substations; routine inspection otherwise.

The Hybrid Approach: Best of Both

For critical equipment, the strongest approach combines both methods.

Typical hybrid program

At commissioning (offline):

• Calibrated PD measurement per IEC 60270
• Establishes quantitative baseline
• Documents acceptance against equipment standards

Continuous (online):

• HFCT or UHF sensors monitoring during operation
• Trending of relative PD activity
• Early warning of developing issues

Periodic (offline):

• Every 5–10 years, take equipment out of service for offline testing
• Re-establishes calibrated baseline
• Validates online monitoring sensitivity
• Allows PDIV/PDEV characterization

On-trigger (offline):

• When online monitoring detects significant change, schedule offline diagnostic testing
• Detailed characterization of the developing fault
• Decision-making input for repair vs replace

This approach captures the best of both: quantitative reference points from offline, continuous coverage from online.

Why hybrid works

The two methods compensate for each other’s weaknesses:

• Offline calibration provides absolute reference; online monitoring provides continuous coverage
• Offline catches PD that requires elevated voltage; online catches PD that requires actual operating conditions
• Offline gives comparable values across equipment; online gives equipment-specific trending

Most modern utility PD programs for critical equipment use this hybrid approach. The cost is higher than either method alone, but the diagnostic value is greater than the sum.

Decision Framework

Use this framework to choose the right approach:

Choose offline-only when:

• Equipment is non-critical (failure consequences are modest)
• Equipment is rarely energized (standby, peaking units)
• Budget is tight ($50k+ for online monitoring is hard to justify on small assets)
• Equipment is new and needs commissioning per standards
• Periodic snapshots are sufficient for the application

Choose online-only when:

• Equipment outages are extremely costly (utility transformers, large generators)
• Operational profile causes PD that varies with load/conditions
• Continuous coverage is mandatory (regulatory or insurance requirement)
• Fleet-wide visibility is needed (centralized SCADA monitoring)
• Manual offline testing is impractical (submarine cables, remote installations)

Choose hybrid when:

• Equipment is critical and the cost of failure justifies investment
• Quantitative compliance is needed AND continuous monitoring is needed
• You want both early warning AND defensible test data
• Long-term asset management is the goal

Skip PD testing entirely when:

• Equipment voltage is below ~3 kV (PD activity is rare; insulation resistance testing usually sufficient)
• Equipment is approaching end of life and slated for replacement
• Cost-benefit analysis doesn’t justify either method
• Other diagnostic methods provide adequate visibility (DGA for transformers, IR for motors)

Specification Checklist

When procuring PD testing equipment or services, verify these specifications:

For offline equipment

• ☐ Compliance with IEC 60270:2000+AMD1:2015
• ☐ Bandwidth specifications (wide-band or narrow-band, depending on application)
• ☐ Calibration capability per IEC 60270 (5 pC, 50 pC, 500 pC at minimum)
• ☐ Calibration uncertainty stated and ≤10%
• ☐ PRPD pattern display and statistical analysis
• ☐ Voltage source compatibility (frequency, max voltage, current capability)
• ☐ Coupling capacitor selection appropriate for equipment voltage class
• ☐ Software for trending and reporting
• ☐ Operator training included

For online monitoring

• ☐ Sensor type appropriate for equipment (HFCT, UHF, acoustic, capacitive)
• ☐ Bandwidth appropriate for application
• ☐ Noise discrimination capability
• ☐ Signal processing and pattern recognition
• ☐ Alarm thresholds configurable
• ☐ Trending and historical data storage
• ☐ SCADA integration (Modbus, IEC 61850)
• ☐ Self-diagnostics and sensor health monitoring
• ☐ Reference test signal injection capability
• ☐ Vendor support and analysis services

For hybrid programs

• ☐ Coordination plan between offline test schedule and online monitoring
• ☐ Cross-reference between online relative values and offline calibrated values
• ☐ Integrated data management (single platform for both data sources)
• ☐ Vendor agnosticism (offline equipment shouldn’t lock you into specific online platform)
• ☐ Long-term roadmap including periodic offline validation

FAQ

Can I use offline PD equipment for online monitoring?

Generally no. Offline equipment is designed for calibrated measurement under controlled conditions; it’s not built for continuous noise discrimination, automatic alarm processing, and SCADA integration. Some manufacturers offer dual-use equipment, but most facilities use dedicated systems for each purpose.

How often should I do offline PD testing if I have continuous online monitoring?

For critical equipment, every 5–10 years is reasonable — typically aligned with major maintenance outages. The offline test re-establishes calibrated baseline, validates the online monitoring sensitivity, and provides documentation for asset valuation. More frequent offline testing is usually unnecessary if online monitoring is providing acceptable coverage.

Are online PD readings comparable across different vendors’ equipment?

Generally no. Different vendors use different scaling, processing algorithms, and proprietary measurement methods. Online PD trending is best done within a single vendor’s system. For cross-vendor comparability, offline measurements per IEC 60270 are needed.

What if my online monitoring shows an alarm but I can’t take the equipment out for offline testing?

Several options:

• Trend the online data to determine if the alarm is increasing, stable, or decreasing
• Check correlated data (DGA for transformers, temperature, load patterns)
• Use mobile offline test equipment during a brief planned outage
• Consult vendor for specific interpretation of the online data
• Plan offline testing during the next scheduled outage

In practice, most online PD alarms are tracked and assessed without immediate outage. The alarm is a flag for attention, not necessarily an immediate-action trigger.

Can online PD monitoring replace insulation resistance testing?

No. They detect different things. Insulation resistance measures bulk insulation against ground (slow degradation, contamination, moisture). PD detects localized discharge activity (developing voids, surface tracking, corona). A complete monitoring program for critical equipment includes both. See our PD vs IR article for details.

Is online PD monitoring worth installing on existing equipment, or only on new equipment?

Both. Sensor installation on existing equipment is generally feasible — HFCTs clamp around grounding conductors that already exist, UHF antennas can be retrofitted to GIS, acoustic sensors mount on existing transformer tanks. The retrofit cost is comparable to factory installation. The decision is economic: critical equipment with high outage cost generally justifies retrofit; routine equipment usually doesn’t.

What’s the minimum equipment age before PD monitoring becomes useful?

For offline testing, useful from day one (commissioning) onward. For online monitoring, useful from commissioning onward but most valuable after the first 5 years when initial defects start emerging. Some operators install monitoring at commissioning and use the first years for baseline establishment.

Does online monitoring detect all PD types?

Different sensors have different sensitivities to different PD types. HFCTs detect electrical pulses well but may miss surface tracking that doesn’t produce strong electrical signals. Acoustic sensors detect mechanical wave from PD but are position-dependent. UHF detects electromagnetic radiation but is bandwidth-limited. A comprehensive monitoring system uses multiple sensor types.

What expertise is needed for online PD analysis?

Significant. Modern online PD systems include automated analysis tools, but interpretation still requires:

• Understanding of equipment-specific PD patterns
• Recognition of noise vs PD signals
• Pattern correlation across multiple sensors
• Trend analysis over time
• Coordination with other diagnostic data (DGA, temperature, vibration)

Most utilities have dedicated PD specialists or contract with diagnostic services for analysis. The data alone isn’t useful without interpretation.

How do PD test results affect insurance and asset valuation?

Increasingly, yes. Insurance companies for critical electrical assets review PD test history during renewals and after incidents. Documented offline PD compliance plus continuous online monitoring data generally improves insurance terms. For transformer asset valuation in trade or sale, recent PD test reports are a standard part of due diligence.

Key Takeaways

• Offline testing is calibrated, traceable, comparable to standards (IEC 60270, IEC 60076-3, IEEE 1434). Best for commissioning, acceptance, and detailed diagnostic characterization.
• Online monitoring is continuous, captures real operating behavior, detects intermittent PD that offline tests miss. Best for ongoing health monitoring of critical equipment.
• Sensitivity differs significantly: offline can achieve 1–10 pC; online typically 10–200 pC equivalent depending on sensor and noise environment.
• Calibration is the biggest divergence: offline is calibrated in pC per IEC 60270; online is typically uncalibrated relative trending.
• Cost depends on application: offline is cheaper for occasional testing of multiple assets; online is cheaper per year for continuous monitoring of a single critical asset.
• Hybrid is best for critical equipment: offline at commissioning and major outages, online continuously between. Combines defensible calibrated data with continuous coverage.
• Equipment-specific considerations matter: UHF online for GIS (mandatory), HFCT online for cables and transformers, acoustic for transformer fault location, offline for rotating machines per IEEE 1434.
• Skip PD testing for low-voltage equipment (below ~3 kV) — IR and other methods are more appropriate.
• Sensor selection drives capability — different sensors detect different PD types; comprehensive programs use multiple sensor types.
• Interpretation requires expertise — modern systems automate detection but human analysis is still required for diagnosis and decision-making.

Standards and References

Standard / Reference	Content
IEC 60270:2000+AMD1:2015	High-voltage test techniques — Partial discharge measurements (offline reference)
IEC 62478:2016	High voltage test techniques — Measurement of partial discharges by electromagnetic and acoustic methods
IEC 60076-3:2018	Power transformers — Insulation levels, dielectric tests, and external clearances in air
IEC 60840:2020	Power cables with extruded insulation — Test methods and requirements (cable PD testing)
IEEE 1434:2014	IEEE Guide for the Measurement of Partial Discharges in AC Electric Machinery
IEEE C57.127:2018	IEEE Guide for the Detection, Location and Interpretation of Sources of Acoustic Emissions from Electrical Discharges in Power Transformers
CIGRE TB 444	Guidelines for unconventional partial discharge measurements
CIGRE TB 502	High-voltage on-site testing with partial discharge measurement
Vahidi & Teymouri (2019)	Quality Confirmation Tests for Power Transformer Insulation Systems (Springer) — Chapters 4 and 5