Digital Partial Discharge Tester: Overvoltage and Transient Event Correlation for Root Cause Analysis
Partial discharge rarely occurs in isolation. In most cases, PD activity is triggered or intensified by overvoltage events—switching transients, lightning surges, resonance, or temporary overvoltages. A digital partial discharge tester that captures PD data without correlating it to the events that caused it provides only half the picture. This article examines the relationship between overvoltage events and PD activity, and demonstrates how correlation analysis enables accurate root cause identification and targeted preventive measures.
Why Overvoltage Events Matter for PD Diagnostics
Insulation systems are designed to withstand rated voltage continuously and specified overvoltages for limited durations. When overvoltages exceed design margins—or occur more frequently than anticipated—PD can initiate, accelerate, or escalate:
- Switching transients: Circuit breaker operations, capacitor bank switching, and load rejection generate oscillatory overvoltages with rise times from microseconds to nanoseconds. These stress insulation turn-to-turn, phase-to-phase, and phase-to-ground.
- Lightning surges: Direct or induced strikes produce impulse overvoltages that can initiate PD in weakened insulation. Even when the surge arrester operates, residual stress may be sufficient to trigger discharge.
- Resonance and ferroresonance: System reconfigurations can create resonant circuits that amplify voltage to 2–4× normal, causing sustained PD and rapid degradation.
- Temporary overvoltages (TOV): Ground faults, load rejection, or generator runaway produce TOV lasting milliseconds to seconds. PD under TOV may not be detectable during normal operation but appears when voltage rises.
- Harmonic distortion: High harmonic content distorts the voltage waveform, creating peak voltages higher than expected and introducing high-frequency components that stress insulation differently than pure 50/60 Hz.
The Correlation Challenge
Traditional PD testing measures PD at a single voltage level and assumes steady-state conditions. This approach misses the dynamic relationship between overvoltage events and PD:
- A cable that shows 20 pC at rated voltage may show 500 pC during switching transients—but the portable test never applies such transients.
- A transformer that passes factory PD tests may fail in service due to cumulative stress from repeated switching operations.
- A GIS that exhibits no PD under normal voltage may develop free particles that only discharge during voltage surges.
To identify the true root cause, PD measurement must be synchronized with event recording—capturing both the overvoltage waveform and the PD response.
Architecture for Event-Correlated PD Monitoring
A digital partial discharge tester capable of overvoltage correlation requires:
| Component | Function | Specification Example |
|---|---|---|
| Voltage divider or VT input | Continuous voltage monitoring and event capture | Bandwidth DC to 10 MHz; accuracy ±1% |
| PD sensor (HFCT, UHF, coupler) | PD pulse acquisition | 100 kHz–2 GHz depending on sensor type |
| High-speed digitizer | Simultaneous capture of voltage and PD channels | 4 channels, 14-bit, 250 MS/s per channel |
| Event trigger logic | Detect overvoltage threshold exceedance | Adjustable threshold (1.1–3.0 pu) with 1 μs response |
| Time-stamping | GPS or IRIG-B for event correlation | ±100 ns absolute time accuracy |
| Data storage | Pre-trigger and post-trigger waveform capture | Minimum 2 seconds pre-trigger, 10 seconds post-trigger |
Types of Events to Capture and Correlate
Configure the digital partial discharge tester to trigger on these event types:
- Voltage threshold exceedance: Trigger when voltage exceeds 110%, 120%, or 150% of rated value. Capture waveform and simultaneous PD activity.
- Rate of rise (dV/dt) exceedance: Trigger on fast transients (e.g., >1 kV/μs) even if peak voltage is moderate. Fast edges stress insulation differently than slow overvoltages.
- Breaker operation: Receive a contact signal from the circuit breaker auxiliary contact. Correlate PD with opening or closing operations.
- Recloser operations: Multiple reclose attempts create repeated overvoltage/PD cycles. Capture each attempt for cumulative stress assessment.
- Capacitor bank switching: Back-to-back switching generates high-frequency transients. Correlate PD with each switching event.
- Load rejection: Sudden load loss causes voltage rise. Monitor PD during the transient and recovery period.
- Fault events: When protection relays operate, capture the voltage and PD waveform to assess insulation stress during fault clearing.
Analysis Techniques for Event-Correlated PD Data
Time-Domain Correlation
Plot the voltage waveform and PD pulse train on the same time axis. Key observations:
- PD inception time relative to voltage rise (e.g., PD appears 2 ms after voltage exceeds 1.2 pu).
- PD magnitude vs. instantaneous voltage during the transient.
- PD decay time after voltage returns to normal (does PD persist or extinguish immediately?).
- Cumulative PD count per event (e.g., 50 PD pulses during one switching transient).
Phase-Resolved Analysis During Events
Generate PRPD plots specifically for the transient period. Compare to steady-state PRPD:
- Does the defect type change during overvoltage (e.g., void discharge becomes surface discharge)?
- Are new phase clusters appearing (indicating additional defect activation)?
- Does PD magnitude increase linearly or non-linearly with overvoltage?
Statistical Correlation
Over many events, calculate:
- Correlation coefficient between overvoltage magnitude and PD magnitude.
- Probability of PD inception as a function of overvoltage level.
- Trend of PD magnitude per event over time (is insulation weakening?).
- Number of PD pulses per event vs. cumulative number of events (aging curve).
Case Study: Identifying Switching-Induced PD in a Wind Farm Collector System
Problem: A 33 kV wind farm collector system experienced three cable joint failures in 18 months. Standard PD testing at rated voltage showed no significant activity. The utility suspected manufacturing defects but could not identify a common cause.
Investigation: A digital partial discharge tester with event correlation was installed at the substation. It continuously monitored voltage and PD on the suspect feeder. Over two weeks, 47 switching events were recorded (capacitor bank operations for reactive power control).
Findings: During each capacitor bank energization, a transient overvoltage of 1.8 pu with 2.5 kHz oscillation occurred. PD activity appeared 200 μs after the voltage peak, with magnitudes of 300–800 pC—far exceeding the 50 pC measured at rated voltage. The PD pattern during transients indicated surface discharge at the joint interface. Cumulative analysis showed that each switching event caused approximately 100 PD pulses, and the magnitude increased by 15% over the two-week monitoring period.
Root cause: The cable joints were manufactured with a small void at the interface between the stress cone and the cable insulation. Under normal voltage, the void was below PD inception. Under switching transients, the voltage stress exceeded inception, causing PD that eroded the interface. Repeated switching events (up to 50 per day) accelerated degradation until failure.
Solution: The utility replaced the joint design with a void-free molded joint and installed pre-insertion resistors on the capacitor bank breakers to reduce transient overvoltage magnitude. Follow-up monitoring showed PD below 20 pC during switching events. No further failures occurred over three years.
Root Cause Analysis Framework
When event-correlated PD data is available, follow this systematic framework:
- Identify the event type: Switching, lightning, fault, resonance, or other.
- Quantify the stress: Peak voltage, rise time, duration, and frequency content of the overvoltage.
- Characterize the PD response: Inception voltage, magnitude, repetition rate, and phase pattern during the event.
- Determine the defect type: Use PRPD pattern during transient to classify (void, surface, corona, floating).
- Assess cumulative damage: Calculate total PD pulses and energy per event; multiply by event frequency to estimate annual stress.
- Identify the weakest link: Compare PD response across phases, joints, or sections to locate the most vulnerable component.
- Develop mitigation: Reduce overvoltage (surge arresters, pre-insertion resistors, damping), strengthen insulation (design change), or reduce event frequency (operational change).
- Verify effectiveness: Continue monitoring after mitigation to confirm PD reduction.
Mitigation Strategies Based on Correlation Findings
| Root Cause | Mitigation Option | Expected PD Reduction |
|---|---|---|
| Capacitor bank switching transients | Pre-insertion resistors, point-on-wave closing, synchronous closing | 60–90% reduction in transient magnitude and PD |
| Repeated recloser operations | Reduce reclose attempts, increase dead time, use single-shot reclosing | Proportional to event reduction |
| Resonance during reconfiguration | Avoid resonant configurations, add damping, detune filters | Eliminates sustained overvoltage and associated PD |
| Lightning-induced PD | Improve surge arrester coordination, reduce grounding impedance | 50–80% reduction in residual overvoltage |
| Harmonic resonance | Add harmonic filters, detune capacitor banks | Reduces voltage distortion and peak stress |
Selecting a Digital Partial Discharge Tester for Event Correlation
Prioritize these features for root cause analysis applications:
- Simultaneous multi-channel acquisition with common time base (voltage + PD).
- Fast event triggering (<1 μs response) with pre-trigger buffer.
- Adjustable trigger thresholds for voltage level, dV/dt, and external contact closure.
- GPS or IRIG-B time-stamping for correlating events across multiple monitoring points.
- Long-duration waveform capture (minimum 10 seconds post-trigger).
- Software that automatically overlays voltage and PD waveforms for analysis.
- Statistical tools for event-PD correlation over weeks or months of data.
Integration with Power Quality Monitoring
Many utilities already have power quality monitors installed. Integrating PD data with power quality event records provides a comprehensive view:
- Power quality monitor records voltage sags, swells, harmonics, and transients.
- Digital partial discharge tester records PD activity with GPS time stamps.
- Correlating the two datasets identifies which power quality events cause PD.
- Combined reporting supports maintenance prioritization and regulatory compliance.
Case Study: Lightning Correlation Saves Transformer
A 230 kV transformer with a digital partial discharge tester and surge counter was monitored for 6 months. A lightning storm produced 12 detected strikes within 2 km. The tester recorded 3 events where voltage exceeded 1.5 pu. PD activity appeared during each event, with magnitudes of 150–400 pC. Between events, PD was below 10 pC. Correlation analysis showed that PD was triggered by steep-fronted lightning surges, indicating a weakened turn-to-turn insulation in the high-voltage winding. The transformer was scheduled for replacement during a planned outage 4 months later. During decommissioning, inspection confirmed turn insulation degradation consistent with repeated surge stress. The correlation data provided definitive evidence for replacement justification.
Best Practices for Event-Correlated PD Monitoring
- Define events of interest before deployment: Not every voltage fluctuation requires capture. Set thresholds based on asset criticality and known stress sources.
- Maintain accurate time synchronization: GPS time-stamping is essential for correlating events from multiple monitors or with SCADA records.
- Record baseline before events: Continuous monitoring with circular buffer ensures pre-event data is available.
- Automate event classification: Use software to tag events as switching, lightning, fault, or unknown based on waveform characteristics.
- Review events promptly: High-magnitude PD during a transient may indicate immediate risk. Set alerts for critical thresholds.
- Trend cumulative stress: Even small PD events, repeated thousands of times, cause long-term degradation. Track cumulative PD energy per asset.
- Share data across departments: Protection engineers, asset managers, and maintenance planners all benefit from event-PD correlation insights.
Limitations and Cautions
- Correlation does not always prove causation. PD may occur coincidentally with an event without being caused by it. Repeat observations strengthen causal inference.
- Some overvoltage events (e.g., very fast transients <100 ns rise time) may not be captured by standard voltage dividers. Specialized sensors are required.
- PD during transients may be masked by the transient itself if the voltage signal saturates the digitizer. Select voltage divider ratios carefully.
- High-frequency transients can couple into PD sensor cables, creating false PD indications. Use shielded cables and differential inputs.
Overvoltage and transient event correlation transforms a digital partial discharge tester from a static measurement tool into a dynamic diagnostic system. By capturing both the cause (overvoltage event) and the effect (PD response), engineers can identify root causes, quantify cumulative stress, and implement targeted mitigation. For asset owners facing unexplained insulation failures or seeking to extend equipment life, event-correlated PD monitoring provides the actionable intelligence needed to break the failure cycle and achieve lasting reliability improvements.
