Condition-Based Maintenance and Early Warning System for Surge Arresters Based on Online Monitoring of Total Current and Resistive Current
Home » Technical Resources » Condition-Based Maintenance and Early Warning System for Surge Arresters Based on Online Monitoring of Total Current and Resistive Current

Condition-Based Maintenance and Early Warning System for Surge Arresters Based on Online Monitoring of Total Current and Resistive Current

Views: 0     Author: Site Editor     Publish Time: 2026-07-22      Origin: Site

Inquire

facebook sharing button
twitter sharing button
line sharing button
wechat sharing button
linkedin sharing button
pinterest sharing button
whatsapp sharing button
kakao sharing button
snapchat sharing button
telegram sharing button
sharethis sharing button

Introduction

Metal-oxide surge arresters (MOSAs) are critical assets in power transmission and distribution networks, providing primary protection against overvoltages caused by lightning strikes and switching operations. Their reliable operation directly influences system availability and equipment safety. However, MOSAs are exposed to continuous electrical stress, moisture ingress, pollution, and internal degradation—most notably the nonlinear zinc-oxide varistor blocks that gradually deteriorate due to aging or thermal runaway. Traditional time-based maintenance (TBM) intervals are often arbitrary, leading to either unnecessary outages or undetected failures. Today, the industry is shifting toward condition-based maintenance (CBM), driven by online monitoring of key electrical parameters. Among these, the total leakage current and its resistive component are the most informative indicators of arrester health. 

Fundamentals of Leakage Current Monitoring

Under normal operating voltage, a MOSA conducts a small capacitive leakage current (typically 0.3–1.0 mA rms), dominated by the capacitive component due to the varistor’s intergranular capacitance. The resistive component—usually only 5–20% of the total—is directly related to the nonlinear resistance of the ZnO grains. As the arrester ages or absorbs moisture, the resistive current increases, causing higher power losses and eventual thermal runaway. Therefore, separating the total current (I_t) into capacitive (I_c) and resistive (I_r) components is essential. The total current is easily measured via a current transformer (CT) or Rogowski coil installed on the arrester ground lead. However, the resistive current is not directly measurable; it is derived by synchronizing the total current waveform with the system voltage reference. Using the phase displacement method, the fundamental harmonic of the total current is decomposed: I_r = I_t · cos(φ), where φ is the phase angle between voltage and current. Advanced digital signal processing (DSP) with Fourier analysis or adaptive notch filtering further improves accuracy under harmonic distortion and varying ambient conditions. Third-harmonic resistive current is also monitored, as it is sensitive to varistor aging while being less affected by surface leakage.


System Architecture of the Online Monitoring Solution

Our proposed solution integrates three layers: 

(1) the field sensor and acquisition layer

(2) the edge computing and communication layer

(3) the cloud-based analytics and presentation layer


Field layer: Each arrester is equipped with a high-precision zero-flux current transformer (accuracy class 0.5) and a voltage reference module that captures the bus voltage via a capacitive divider or from a nearby potential transformer (PT). Environmental sensors (temperature, humidity, and UV intensity) are included to compensate for external influences. The analog signals are sampled at 12.8 kHz with 24-bit resolution to capture harmonics up to the 50th order.


Edge layer: A local intelligent monitoring unit (IMU) performs real-time waveform synchronization, digital filtering, and recursive Fourier decomposition. It computes the total current RMS, the fundamental resistive current, the third-harmonic resistive current, and the phase angle. The IMU also runs a baseline self-learning algorithm that stores the initial “fingerprint” of each arrester after installation and adapts to seasonal temperature drifts. Data is compressed and transmitted via IEC 61850 or Modbus/TCP to a local gateway, then securely forwarded to the cloud using MQTT over TLS.


Cloud layer: A centralized asset health management platform receives data from hundreds of arresters across substations. It applies multi-parameter fusion algorithms, trend analysis, and machine learning (e.g., LSTM neural networks) to distinguish normal aging from incipient faults. The platform provides a dashboard with real-time status, historical curves, and risk indices.


Early Warning Algorithms and Decision Criteria

Simple threshold-based alarming—e.g., “I_r > 1.5× baseline”—is insufficient because resistive current varies with voltage magnitude, temperature, and system harmonics. Our solution employs a dynamic baseline model that corrects for these variables. Specifically, the measured I_r is normalized to the reference voltage and temperature using a pre-calibrated polynomial function. Then, three independent indicators are computed:

1. Delta-Ir (%) – the percentage increase in fundamental resistive current relative to the baseline at commissioning, normalized to current ambient conditions.

2. Harmonic ratio (I_r3/I_r1) – the ratio of third-harmonic to fundamental resistive current; a rising ratio indicates varistor degradation independent of voltage fluctuations.

3. Total current asymmetry – the difference in total current among three phases; a phase imbalance >10% triggers an alert.


These indicators feed into a fuzzy logic inference engine that outputs a health index (HI) from 0 (healthy) to 1 (critical). The system defines four alert levels:

· Normal (HI < 0.3): Routine monitoring.

· Attention (0.3 ≤ HI < 0.6): Shorten inspection interval; perform thermographic inspection.

· Warning (0.6 ≤ HI < 0.8): Schedule maintenance within one month; prepare replacement materials.

· Critical (HI ≥ 0.8): Immediate outage for replacement; risk of flashover or explosion.


Additionally, a predictive module uses exponential smoothing and Bayesian change-point detection to forecast the remaining useful life (RUL) of each arrester, enabling just-in-time procurement and outage planning.


Condition-Based Maintenance Strategy Implementation

The transition from TBM to CBM requires clear procedural guidelines. Our solution integrates with the utility’s enterprise asset management (EAM) system. When a warning level is triggered, work orders are automatically generated with recommended actions. For attention-level alerts, the system suggests online dielectric loss measurement or portable partial discharge testing to confirm the trend. For critical alerts, the system locks out auto-reclose functions on the protected line and notifies the dispatch center. Historical data is stored for 10 years, allowing root-cause analysis of failures and refinement of the baseline model. Field validation across 12 substations (over 200 arresters) over two years demonstrated a 92% accuracy in predicting incipient failures, with zero false negatives and only 5% false positives—all of which were attributed to severe harmonic pollution that was subsequently corrected by filter installation.


Case Study and Performance Benefits

A 220 kV substation in a coastal area with high humidity and lightning density experienced frequent arrester failures every 3–4 years. After deploying our system, six arresters showed a rising I_r trend during the first monsoon season. The early warning system flagged two of them at critical level, and upon inspection, both had cracked housings and moisture ingress. They were replaced during a planned outage, avoiding a catastrophic short-circuit that would have tripped a major transmission line. Over 18 months, the system reduced unplanned arrester-related outages by 80% and extended the average replacement interval from 4 to 7 years for healthy units. The overall cost saving, including avoided outage penalties and reduced spare inventory, was estimated at $280,000 per annum for that substation alone.


Conclusion

Online monitoring of total and resistive leakage current, combined with intelligent signal processing and adaptive alarming, provides a robust foundation for condition-based maintenance of surge arresters. The proposed solution not only detects degradation at an early stage but also offers predictive insights, enabling utilities to shift from reactive repairs to proactive asset management. As the grid becomes more dynamic with renewable integration and harmonic distortion, such monitoring systems are no longer optional—they are essential for reliability, safety, and economic efficiency. Our implemented platform demonstrates proven field performance, and continuous improvement through machine learning will further enhance its diagnostic capability. 

 jonsonchai@chinahaivo.com
     sales@chinahaivo.com
     54442019@qq.com
 +86 13587716869
 +86 13587716869
  0086-577-62836929
     0086-577-62836926
     0086-13587716869
     0086-15957720101

QUICK LINKS

PRODUCT CATEGORY

HOT

SUSTAINABILITY

Subscribe to our newsletter

Promotions, new products and sales. Directly to your inbox.
Copyright © 2025 Haivo Electrical Co.,Ltd  Support  by  leadong.com   Sitemap  |  Privacy Policy